aboutsummaryrefslogtreecommitdiff
path: root/test/CodeGen/X86
diff options
context:
space:
mode:
Diffstat (limited to 'test/CodeGen/X86')
-rw-r--r--test/CodeGen/X86/bitcast-and-setcc-128.ll1155
-rw-r--r--test/CodeGen/X86/bitcast-and-setcc-256.ll403
-rw-r--r--test/CodeGen/X86/mul-constant-i16.ll141
-rw-r--r--test/CodeGen/X86/mul-constant-i32.ll1589
-rw-r--r--test/CodeGen/X86/mul-constant-i64.ll1612
-rw-r--r--test/CodeGen/X86/setcc-lowering.ll65
-rw-r--r--test/CodeGen/X86/vector-sext.ll56
-rw-r--r--test/CodeGen/X86/xchg-nofold.ll37
8 files changed, 4695 insertions, 363 deletions
diff --git a/test/CodeGen/X86/bitcast-and-setcc-128.ll b/test/CodeGen/X86/bitcast-and-setcc-128.ll
new file mode 100644
index 000000000000..a681c3b0aa42
--- /dev/null
+++ b/test/CodeGen/X86/bitcast-and-setcc-128.ll
@@ -0,0 +1,1155 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=x86_64-apple-darwin -mcpu=x86-64 -mattr=+sse2 < %s | FileCheck %s --check-prefixes=SSE2-SSSE3,SSE2
+; RUN: llc -mtriple=x86_64-apple-darwin -mcpu=x86-64 -mattr=+ssse3 < %s | FileCheck %s --check-prefixes=SSE2-SSSE3,SSSE3
+; RUN: llc -mtriple=x86_64-apple-darwin -mcpu=x86-64 -mattr=+avx < %s | FileCheck %s --check-prefixes=AVX12,AVX1
+; RUN: llc -mtriple=x86_64-apple-darwin -mcpu=x86-64 -mattr=+avx2 < %s | FileCheck %s --check-prefixes=AVX12,AVX2
+; RUN: llc -mtriple=x86_64-apple-darwin -mcpu=x86-64 -mattr=+avx512f,+avx512vl,+avx512bw < %s | FileCheck %s --check-prefixes=AVX512
+
+define i8 @v8i16(<8 x i16> %a, <8 x i16> %b, <8 x i16> %c, <8 x i16> %d) {
+; SSE2-SSSE3-LABEL: v8i16:
+; SSE2-SSSE3: ## BB#0:
+; SSE2-SSSE3-NEXT: pcmpgtw %xmm1, %xmm0
+; SSE2-SSSE3-NEXT: pcmpgtw %xmm3, %xmm2
+; SSE2-SSSE3-NEXT: pand %xmm0, %xmm2
+; SSE2-SSSE3-NEXT: pextrw $7, %xmm2, %eax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: pextrw $6, %xmm2, %eax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: pextrw $5, %xmm2, %eax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: pextrw $4, %xmm2, %eax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: pextrw $3, %xmm2, %eax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: pextrw $2, %xmm2, %eax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: pextrw $1, %xmm2, %eax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: movd %xmm2, %eax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; SSE2-SSSE3-NEXT: retq
+;
+; AVX12-LABEL: v8i16:
+; AVX12: ## BB#0:
+; AVX12-NEXT: vpcmpgtw %xmm1, %xmm0, %xmm0
+; AVX12-NEXT: vpcmpgtw %xmm3, %xmm2, %xmm1
+; AVX12-NEXT: vpand %xmm1, %xmm0, %xmm0
+; AVX12-NEXT: vpextrw $7, %xmm0, %eax
+; AVX12-NEXT: andl $1, %eax
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vpextrw $6, %xmm0, %eax
+; AVX12-NEXT: andl $1, %eax
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vpextrw $5, %xmm0, %eax
+; AVX12-NEXT: andl $1, %eax
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vpextrw $4, %xmm0, %eax
+; AVX12-NEXT: andl $1, %eax
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vpextrw $3, %xmm0, %eax
+; AVX12-NEXT: andl $1, %eax
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vpextrw $2, %xmm0, %eax
+; AVX12-NEXT: andl $1, %eax
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vpextrw $1, %xmm0, %eax
+; AVX12-NEXT: andl $1, %eax
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vmovd %xmm0, %eax
+; AVX12-NEXT: andl $1, %eax
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; AVX12-NEXT: retq
+;
+; AVX512-LABEL: v8i16:
+; AVX512: ## BB#0:
+; AVX512-NEXT: vpcmpgtw %xmm1, %xmm0, %k1
+; AVX512-NEXT: vpcmpgtw %xmm3, %xmm2, %k0 {%k1}
+; AVX512-NEXT: kmovd %k0, %eax
+; AVX512-NEXT: ## kill: %AL<def> %AL<kill> %EAX<kill>
+; AVX512-NEXT: retq
+ %x0 = icmp sgt <8 x i16> %a, %b
+ %x1 = icmp sgt <8 x i16> %c, %d
+ %y = and <8 x i1> %x0, %x1
+ %res = bitcast <8 x i1> %y to i8
+ ret i8 %res
+}
+
+define i4 @v4i32(<4 x i32> %a, <4 x i32> %b, <4 x i32> %c, <4 x i32> %d) {
+; SSE2-SSSE3-LABEL: v4i32:
+; SSE2-SSSE3: ## BB#0:
+; SSE2-SSSE3-NEXT: pcmpgtd %xmm1, %xmm0
+; SSE2-SSSE3-NEXT: pcmpgtd %xmm3, %xmm2
+; SSE2-SSSE3-NEXT: pand %xmm0, %xmm2
+; SSE2-SSSE3-NEXT: movd %xmm2, %eax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm2[3,1,2,3]
+; SSE2-SSSE3-NEXT: movd %xmm0, %eax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm2[2,3,0,1]
+; SSE2-SSSE3-NEXT: movd %xmm0, %eax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm2[1,1,2,3]
+; SSE2-SSSE3-NEXT: movd %xmm0, %eax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; SSE2-SSSE3-NEXT: retq
+;
+; AVX12-LABEL: v4i32:
+; AVX12: ## BB#0:
+; AVX12-NEXT: vpcmpgtd %xmm1, %xmm0, %xmm0
+; AVX12-NEXT: vpcmpgtd %xmm3, %xmm2, %xmm1
+; AVX12-NEXT: vpand %xmm1, %xmm0, %xmm0
+; AVX12-NEXT: vpextrd $3, %xmm0, %eax
+; AVX12-NEXT: andl $1, %eax
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vpextrd $2, %xmm0, %eax
+; AVX12-NEXT: andl $1, %eax
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vpextrd $1, %xmm0, %eax
+; AVX12-NEXT: andl $1, %eax
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vmovd %xmm0, %eax
+; AVX12-NEXT: andl $1, %eax
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; AVX12-NEXT: retq
+;
+; AVX512-LABEL: v4i32:
+; AVX512: ## BB#0:
+; AVX512-NEXT: vpcmpgtd %xmm1, %xmm0, %k1
+; AVX512-NEXT: vpcmpgtd %xmm3, %xmm2, %k0 {%k1}
+; AVX512-NEXT: kmovd %k0, %eax
+; AVX512-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX512-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; AVX512-NEXT: retq
+ %x0 = icmp sgt <4 x i32> %a, %b
+ %x1 = icmp sgt <4 x i32> %c, %d
+ %y = and <4 x i1> %x0, %x1
+ %res = bitcast <4 x i1> %y to i4
+ ret i4 %res
+}
+
+define i4 @v4f32(<4 x float> %a, <4 x float> %b, <4 x float> %c, <4 x float> %d) {
+; SSE2-SSSE3-LABEL: v4f32:
+; SSE2-SSSE3: ## BB#0:
+; SSE2-SSSE3-NEXT: cmpltps %xmm0, %xmm1
+; SSE2-SSSE3-NEXT: cmpltps %xmm2, %xmm3
+; SSE2-SSSE3-NEXT: andps %xmm1, %xmm3
+; SSE2-SSSE3-NEXT: movd %xmm3, %eax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm3[3,1,2,3]
+; SSE2-SSSE3-NEXT: movd %xmm0, %eax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm3[2,3,0,1]
+; SSE2-SSSE3-NEXT: movd %xmm0, %eax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm3[1,1,2,3]
+; SSE2-SSSE3-NEXT: movd %xmm0, %eax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; SSE2-SSSE3-NEXT: retq
+;
+; AVX12-LABEL: v4f32:
+; AVX12: ## BB#0:
+; AVX12-NEXT: vcmpltps %xmm0, %xmm1, %xmm0
+; AVX12-NEXT: vcmpltps %xmm2, %xmm3, %xmm1
+; AVX12-NEXT: vandps %xmm1, %xmm0, %xmm0
+; AVX12-NEXT: vpextrd $3, %xmm0, %eax
+; AVX12-NEXT: andl $1, %eax
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vpextrd $2, %xmm0, %eax
+; AVX12-NEXT: andl $1, %eax
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vpextrd $1, %xmm0, %eax
+; AVX12-NEXT: andl $1, %eax
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vmovd %xmm0, %eax
+; AVX12-NEXT: andl $1, %eax
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; AVX12-NEXT: retq
+;
+; AVX512-LABEL: v4f32:
+; AVX512: ## BB#0:
+; AVX512-NEXT: vcmpltps %xmm0, %xmm1, %k1
+; AVX512-NEXT: vcmpltps %xmm2, %xmm3, %k0 {%k1}
+; AVX512-NEXT: kmovd %k0, %eax
+; AVX512-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX512-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; AVX512-NEXT: retq
+ %x0 = fcmp ogt <4 x float> %a, %b
+ %x1 = fcmp ogt <4 x float> %c, %d
+ %y = and <4 x i1> %x0, %x1
+ %res = bitcast <4 x i1> %y to i4
+ ret i4 %res
+}
+
+define i16 @v16i8(<16 x i8> %a, <16 x i8> %b, <16 x i8> %c, <16 x i8> %d) {
+; SSE2-SSSE3-LABEL: v16i8:
+; SSE2-SSSE3: ## BB#0:
+; SSE2-SSSE3-NEXT: pcmpgtb %xmm1, %xmm0
+; SSE2-SSSE3-NEXT: pcmpgtb %xmm3, %xmm2
+; SSE2-SSSE3-NEXT: pand %xmm0, %xmm2
+; SSE2-SSSE3-NEXT: movdqa %xmm2, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; SSE2-SSSE3-NEXT: andb $1, %al
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; SSE2-SSSE3-NEXT: andb $1, %al
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; SSE2-SSSE3-NEXT: andb $1, %al
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; SSE2-SSSE3-NEXT: andb $1, %al
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; SSE2-SSSE3-NEXT: andb $1, %al
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; SSE2-SSSE3-NEXT: andb $1, %al
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; SSE2-SSSE3-NEXT: andb $1, %al
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; SSE2-SSSE3-NEXT: andb $1, %al
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; SSE2-SSSE3-NEXT: andb $1, %al
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; SSE2-SSSE3-NEXT: andb $1, %al
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; SSE2-SSSE3-NEXT: andb $1, %al
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; SSE2-SSSE3-NEXT: andb $1, %al
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; SSE2-SSSE3-NEXT: andb $1, %al
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; SSE2-SSSE3-NEXT: andb $1, %al
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; SSE2-SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %cl
+; SSE2-SSSE3-NEXT: andb $1, %cl
+; SSE2-SSSE3-NEXT: movb %cl, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: andb $1, %al
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: movzwl -{{[0-9]+}}(%rsp), %eax
+; SSE2-SSSE3-NEXT: retq
+;
+; AVX12-LABEL: v16i8:
+; AVX12: ## BB#0:
+; AVX12-NEXT: vpcmpgtb %xmm1, %xmm0, %xmm0
+; AVX12-NEXT: vpcmpgtb %xmm3, %xmm2, %xmm1
+; AVX12-NEXT: vpand %xmm1, %xmm0, %xmm0
+; AVX12-NEXT: vpextrb $15, %xmm0, %eax
+; AVX12-NEXT: andb $1, %al
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vpextrb $14, %xmm0, %eax
+; AVX12-NEXT: andb $1, %al
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vpextrb $13, %xmm0, %eax
+; AVX12-NEXT: andb $1, %al
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vpextrb $12, %xmm0, %eax
+; AVX12-NEXT: andb $1, %al
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vpextrb $11, %xmm0, %eax
+; AVX12-NEXT: andb $1, %al
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vpextrb $10, %xmm0, %eax
+; AVX12-NEXT: andb $1, %al
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vpextrb $9, %xmm0, %eax
+; AVX12-NEXT: andb $1, %al
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vpextrb $8, %xmm0, %eax
+; AVX12-NEXT: andb $1, %al
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vpextrb $7, %xmm0, %eax
+; AVX12-NEXT: andb $1, %al
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vpextrb $6, %xmm0, %eax
+; AVX12-NEXT: andb $1, %al
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vpextrb $5, %xmm0, %eax
+; AVX12-NEXT: andb $1, %al
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vpextrb $4, %xmm0, %eax
+; AVX12-NEXT: andb $1, %al
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vpextrb $3, %xmm0, %eax
+; AVX12-NEXT: andb $1, %al
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vpextrb $2, %xmm0, %eax
+; AVX12-NEXT: andb $1, %al
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vpextrb $1, %xmm0, %eax
+; AVX12-NEXT: andb $1, %al
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vpextrb $0, %xmm0, %eax
+; AVX12-NEXT: andb $1, %al
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: movzwl -{{[0-9]+}}(%rsp), %eax
+; AVX12-NEXT: retq
+;
+; AVX512-LABEL: v16i8:
+; AVX512: ## BB#0:
+; AVX512-NEXT: vpcmpgtb %xmm1, %xmm0, %k1
+; AVX512-NEXT: vpcmpgtb %xmm3, %xmm2, %k0 {%k1}
+; AVX512-NEXT: kmovd %k0, %eax
+; AVX512-NEXT: ## kill: %AX<def> %AX<kill> %EAX<kill>
+; AVX512-NEXT: retq
+ %x0 = icmp sgt <16 x i8> %a, %b
+ %x1 = icmp sgt <16 x i8> %c, %d
+ %y = and <16 x i1> %x0, %x1
+ %res = bitcast <16 x i1> %y to i16
+ ret i16 %res
+}
+
+define i2 @v2i8(<2 x i8> %a, <2 x i8> %b, <2 x i8> %c, <2 x i8> %d) {
+; SSE2-SSSE3-LABEL: v2i8:
+; SSE2-SSSE3: ## BB#0:
+; SSE2-SSSE3-NEXT: psllq $56, %xmm2
+; SSE2-SSSE3-NEXT: movdqa %xmm2, %xmm4
+; SSE2-SSSE3-NEXT: psrad $31, %xmm4
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm4 = xmm4[1,3,2,3]
+; SSE2-SSSE3-NEXT: psrad $24, %xmm2
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm2 = xmm2[1,3,2,3]
+; SSE2-SSSE3-NEXT: punpckldq {{.*#+}} xmm2 = xmm2[0],xmm4[0],xmm2[1],xmm4[1]
+; SSE2-SSSE3-NEXT: psllq $56, %xmm3
+; SSE2-SSSE3-NEXT: movdqa %xmm3, %xmm4
+; SSE2-SSSE3-NEXT: psrad $31, %xmm4
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm4 = xmm4[1,3,2,3]
+; SSE2-SSSE3-NEXT: psrad $24, %xmm3
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm3 = xmm3[1,3,2,3]
+; SSE2-SSSE3-NEXT: punpckldq {{.*#+}} xmm3 = xmm3[0],xmm4[0],xmm3[1],xmm4[1]
+; SSE2-SSSE3-NEXT: psllq $56, %xmm0
+; SSE2-SSSE3-NEXT: movdqa %xmm0, %xmm4
+; SSE2-SSSE3-NEXT: psrad $31, %xmm4
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm4 = xmm4[1,3,2,3]
+; SSE2-SSSE3-NEXT: psrad $24, %xmm0
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,3,2,3]
+; SSE2-SSSE3-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1]
+; SSE2-SSSE3-NEXT: psllq $56, %xmm1
+; SSE2-SSSE3-NEXT: movdqa %xmm1, %xmm4
+; SSE2-SSSE3-NEXT: psrad $31, %xmm4
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm4 = xmm4[1,3,2,3]
+; SSE2-SSSE3-NEXT: psrad $24, %xmm1
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm1 = xmm1[1,3,2,3]
+; SSE2-SSSE3-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm4[0],xmm1[1],xmm4[1]
+; SSE2-SSSE3-NEXT: movdqa {{.*#+}} xmm4 = [2147483648,0,2147483648,0]
+; SSE2-SSSE3-NEXT: pxor %xmm4, %xmm1
+; SSE2-SSSE3-NEXT: pxor %xmm4, %xmm0
+; SSE2-SSSE3-NEXT: movdqa %xmm0, %xmm5
+; SSE2-SSSE3-NEXT: pcmpgtd %xmm1, %xmm5
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm6 = xmm5[0,0,2,2]
+; SSE2-SSSE3-NEXT: pcmpeqd %xmm1, %xmm0
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
+; SSE2-SSSE3-NEXT: pand %xmm6, %xmm0
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm1 = xmm5[1,1,3,3]
+; SSE2-SSSE3-NEXT: por %xmm0, %xmm1
+; SSE2-SSSE3-NEXT: pxor %xmm4, %xmm3
+; SSE2-SSSE3-NEXT: pxor %xmm4, %xmm2
+; SSE2-SSSE3-NEXT: movdqa %xmm2, %xmm0
+; SSE2-SSSE3-NEXT: pcmpgtd %xmm3, %xmm0
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm4 = xmm0[0,0,2,2]
+; SSE2-SSSE3-NEXT: pcmpeqd %xmm3, %xmm2
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; SSE2-SSSE3-NEXT: pand %xmm4, %xmm2
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
+; SSE2-SSSE3-NEXT: por %xmm2, %xmm0
+; SSE2-SSSE3-NEXT: pand %xmm1, %xmm0
+; SSE2-SSSE3-NEXT: movq %xmm0, %rax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
+; SSE2-SSSE3-NEXT: movq %xmm0, %rax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; SSE2-SSSE3-NEXT: retq
+;
+; AVX1-LABEL: v2i8:
+; AVX1: ## BB#0:
+; AVX1-NEXT: vpsllq $56, %xmm3, %xmm3
+; AVX1-NEXT: vpsrad $31, %xmm3, %xmm4
+; AVX1-NEXT: vpsrad $24, %xmm3, %xmm3
+; AVX1-NEXT: vpshufd {{.*#+}} xmm3 = xmm3[1,1,3,3]
+; AVX1-NEXT: vpblendw {{.*#+}} xmm3 = xmm3[0,1],xmm4[2,3],xmm3[4,5],xmm4[6,7]
+; AVX1-NEXT: vpsllq $56, %xmm2, %xmm2
+; AVX1-NEXT: vpsrad $31, %xmm2, %xmm4
+; AVX1-NEXT: vpsrad $24, %xmm2, %xmm2
+; AVX1-NEXT: vpshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; AVX1-NEXT: vpblendw {{.*#+}} xmm2 = xmm2[0,1],xmm4[2,3],xmm2[4,5],xmm4[6,7]
+; AVX1-NEXT: vpsllq $56, %xmm1, %xmm1
+; AVX1-NEXT: vpsrad $31, %xmm1, %xmm4
+; AVX1-NEXT: vpsrad $24, %xmm1, %xmm1
+; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
+; AVX1-NEXT: vpblendw {{.*#+}} xmm1 = xmm1[0,1],xmm4[2,3],xmm1[4,5],xmm4[6,7]
+; AVX1-NEXT: vpsllq $56, %xmm0, %xmm0
+; AVX1-NEXT: vpsrad $31, %xmm0, %xmm4
+; AVX1-NEXT: vpsrad $24, %xmm0, %xmm0
+; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
+; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm4[2,3],xmm0[4,5],xmm4[6,7]
+; AVX1-NEXT: vpcmpgtq %xmm1, %xmm0, %xmm0
+; AVX1-NEXT: vpcmpgtq %xmm3, %xmm2, %xmm1
+; AVX1-NEXT: vpand %xmm1, %xmm0, %xmm0
+; AVX1-NEXT: vpextrq $1, %xmm0, %rax
+; AVX1-NEXT: andl $1, %eax
+; AVX1-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX1-NEXT: vmovq %xmm0, %rax
+; AVX1-NEXT: andl $1, %eax
+; AVX1-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX1-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; AVX1-NEXT: retq
+;
+; AVX2-LABEL: v2i8:
+; AVX2: ## BB#0:
+; AVX2-NEXT: vpsllq $56, %xmm3, %xmm3
+; AVX2-NEXT: vpsrad $31, %xmm3, %xmm4
+; AVX2-NEXT: vpsrad $24, %xmm3, %xmm3
+; AVX2-NEXT: vpshufd {{.*#+}} xmm3 = xmm3[1,1,3,3]
+; AVX2-NEXT: vpblendd {{.*#+}} xmm3 = xmm3[0],xmm4[1],xmm3[2],xmm4[3]
+; AVX2-NEXT: vpsllq $56, %xmm2, %xmm2
+; AVX2-NEXT: vpsrad $31, %xmm2, %xmm4
+; AVX2-NEXT: vpsrad $24, %xmm2, %xmm2
+; AVX2-NEXT: vpshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; AVX2-NEXT: vpblendd {{.*#+}} xmm2 = xmm2[0],xmm4[1],xmm2[2],xmm4[3]
+; AVX2-NEXT: vpsllq $56, %xmm1, %xmm1
+; AVX2-NEXT: vpsrad $31, %xmm1, %xmm4
+; AVX2-NEXT: vpsrad $24, %xmm1, %xmm1
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
+; AVX2-NEXT: vpblendd {{.*#+}} xmm1 = xmm1[0],xmm4[1],xmm1[2],xmm4[3]
+; AVX2-NEXT: vpsllq $56, %xmm0, %xmm0
+; AVX2-NEXT: vpsrad $31, %xmm0, %xmm4
+; AVX2-NEXT: vpsrad $24, %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
+; AVX2-NEXT: vpblendd {{.*#+}} xmm0 = xmm0[0],xmm4[1],xmm0[2],xmm4[3]
+; AVX2-NEXT: vpcmpgtq %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpcmpgtq %xmm3, %xmm2, %xmm1
+; AVX2-NEXT: vpand %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpextrq $1, %xmm0, %rax
+; AVX2-NEXT: andl $1, %eax
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vmovq %xmm0, %rax
+; AVX2-NEXT: andl $1, %eax
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: v2i8:
+; AVX512: ## BB#0:
+; AVX512-NEXT: vpsllq $56, %xmm3, %xmm3
+; AVX512-NEXT: vpsraq $56, %xmm3, %xmm3
+; AVX512-NEXT: vpsllq $56, %xmm2, %xmm2
+; AVX512-NEXT: vpsraq $56, %xmm2, %xmm2
+; AVX512-NEXT: vpsllq $56, %xmm1, %xmm1
+; AVX512-NEXT: vpsraq $56, %xmm1, %xmm1
+; AVX512-NEXT: vpsllq $56, %xmm0, %xmm0
+; AVX512-NEXT: vpsraq $56, %xmm0, %xmm0
+; AVX512-NEXT: vpcmpgtq %xmm1, %xmm0, %k1
+; AVX512-NEXT: vpcmpgtq %xmm3, %xmm2, %k0 {%k1}
+; AVX512-NEXT: kmovd %k0, %eax
+; AVX512-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX512-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; AVX512-NEXT: retq
+ %x0 = icmp sgt <2 x i8> %a, %b
+ %x1 = icmp sgt <2 x i8> %c, %d
+ %y = and <2 x i1> %x0, %x1
+ %res = bitcast <2 x i1> %y to i2
+ ret i2 %res
+}
+
+define i2 @v2i16(<2 x i16> %a, <2 x i16> %b, <2 x i16> %c, <2 x i16> %d) {
+; SSE2-SSSE3-LABEL: v2i16:
+; SSE2-SSSE3: ## BB#0:
+; SSE2-SSSE3-NEXT: psllq $48, %xmm2
+; SSE2-SSSE3-NEXT: movdqa %xmm2, %xmm4
+; SSE2-SSSE3-NEXT: psrad $31, %xmm4
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm4 = xmm4[1,3,2,3]
+; SSE2-SSSE3-NEXT: psrad $16, %xmm2
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm2 = xmm2[1,3,2,3]
+; SSE2-SSSE3-NEXT: punpckldq {{.*#+}} xmm2 = xmm2[0],xmm4[0],xmm2[1],xmm4[1]
+; SSE2-SSSE3-NEXT: psllq $48, %xmm3
+; SSE2-SSSE3-NEXT: movdqa %xmm3, %xmm4
+; SSE2-SSSE3-NEXT: psrad $31, %xmm4
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm4 = xmm4[1,3,2,3]
+; SSE2-SSSE3-NEXT: psrad $16, %xmm3
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm3 = xmm3[1,3,2,3]
+; SSE2-SSSE3-NEXT: punpckldq {{.*#+}} xmm3 = xmm3[0],xmm4[0],xmm3[1],xmm4[1]
+; SSE2-SSSE3-NEXT: psllq $48, %xmm0
+; SSE2-SSSE3-NEXT: movdqa %xmm0, %xmm4
+; SSE2-SSSE3-NEXT: psrad $31, %xmm4
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm4 = xmm4[1,3,2,3]
+; SSE2-SSSE3-NEXT: psrad $16, %xmm0
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,3,2,3]
+; SSE2-SSSE3-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1]
+; SSE2-SSSE3-NEXT: psllq $48, %xmm1
+; SSE2-SSSE3-NEXT: movdqa %xmm1, %xmm4
+; SSE2-SSSE3-NEXT: psrad $31, %xmm4
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm4 = xmm4[1,3,2,3]
+; SSE2-SSSE3-NEXT: psrad $16, %xmm1
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm1 = xmm1[1,3,2,3]
+; SSE2-SSSE3-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm4[0],xmm1[1],xmm4[1]
+; SSE2-SSSE3-NEXT: movdqa {{.*#+}} xmm4 = [2147483648,0,2147483648,0]
+; SSE2-SSSE3-NEXT: pxor %xmm4, %xmm1
+; SSE2-SSSE3-NEXT: pxor %xmm4, %xmm0
+; SSE2-SSSE3-NEXT: movdqa %xmm0, %xmm5
+; SSE2-SSSE3-NEXT: pcmpgtd %xmm1, %xmm5
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm6 = xmm5[0,0,2,2]
+; SSE2-SSSE3-NEXT: pcmpeqd %xmm1, %xmm0
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
+; SSE2-SSSE3-NEXT: pand %xmm6, %xmm0
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm1 = xmm5[1,1,3,3]
+; SSE2-SSSE3-NEXT: por %xmm0, %xmm1
+; SSE2-SSSE3-NEXT: pxor %xmm4, %xmm3
+; SSE2-SSSE3-NEXT: pxor %xmm4, %xmm2
+; SSE2-SSSE3-NEXT: movdqa %xmm2, %xmm0
+; SSE2-SSSE3-NEXT: pcmpgtd %xmm3, %xmm0
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm4 = xmm0[0,0,2,2]
+; SSE2-SSSE3-NEXT: pcmpeqd %xmm3, %xmm2
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; SSE2-SSSE3-NEXT: pand %xmm4, %xmm2
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
+; SSE2-SSSE3-NEXT: por %xmm2, %xmm0
+; SSE2-SSSE3-NEXT: pand %xmm1, %xmm0
+; SSE2-SSSE3-NEXT: movq %xmm0, %rax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
+; SSE2-SSSE3-NEXT: movq %xmm0, %rax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; SSE2-SSSE3-NEXT: retq
+;
+; AVX1-LABEL: v2i16:
+; AVX1: ## BB#0:
+; AVX1-NEXT: vpsllq $48, %xmm3, %xmm3
+; AVX1-NEXT: vpsrad $31, %xmm3, %xmm4
+; AVX1-NEXT: vpsrad $16, %xmm3, %xmm3
+; AVX1-NEXT: vpshufd {{.*#+}} xmm3 = xmm3[1,1,3,3]
+; AVX1-NEXT: vpblendw {{.*#+}} xmm3 = xmm3[0,1],xmm4[2,3],xmm3[4,5],xmm4[6,7]
+; AVX1-NEXT: vpsllq $48, %xmm2, %xmm2
+; AVX1-NEXT: vpsrad $31, %xmm2, %xmm4
+; AVX1-NEXT: vpsrad $16, %xmm2, %xmm2
+; AVX1-NEXT: vpshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; AVX1-NEXT: vpblendw {{.*#+}} xmm2 = xmm2[0,1],xmm4[2,3],xmm2[4,5],xmm4[6,7]
+; AVX1-NEXT: vpsllq $48, %xmm1, %xmm1
+; AVX1-NEXT: vpsrad $31, %xmm1, %xmm4
+; AVX1-NEXT: vpsrad $16, %xmm1, %xmm1
+; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
+; AVX1-NEXT: vpblendw {{.*#+}} xmm1 = xmm1[0,1],xmm4[2,3],xmm1[4,5],xmm4[6,7]
+; AVX1-NEXT: vpsllq $48, %xmm0, %xmm0
+; AVX1-NEXT: vpsrad $31, %xmm0, %xmm4
+; AVX1-NEXT: vpsrad $16, %xmm0, %xmm0
+; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
+; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm4[2,3],xmm0[4,5],xmm4[6,7]
+; AVX1-NEXT: vpcmpgtq %xmm1, %xmm0, %xmm0
+; AVX1-NEXT: vpcmpgtq %xmm3, %xmm2, %xmm1
+; AVX1-NEXT: vpand %xmm1, %xmm0, %xmm0
+; AVX1-NEXT: vpextrq $1, %xmm0, %rax
+; AVX1-NEXT: andl $1, %eax
+; AVX1-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX1-NEXT: vmovq %xmm0, %rax
+; AVX1-NEXT: andl $1, %eax
+; AVX1-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX1-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; AVX1-NEXT: retq
+;
+; AVX2-LABEL: v2i16:
+; AVX2: ## BB#0:
+; AVX2-NEXT: vpsllq $48, %xmm3, %xmm3
+; AVX2-NEXT: vpsrad $31, %xmm3, %xmm4
+; AVX2-NEXT: vpsrad $16, %xmm3, %xmm3
+; AVX2-NEXT: vpshufd {{.*#+}} xmm3 = xmm3[1,1,3,3]
+; AVX2-NEXT: vpblendd {{.*#+}} xmm3 = xmm3[0],xmm4[1],xmm3[2],xmm4[3]
+; AVX2-NEXT: vpsllq $48, %xmm2, %xmm2
+; AVX2-NEXT: vpsrad $31, %xmm2, %xmm4
+; AVX2-NEXT: vpsrad $16, %xmm2, %xmm2
+; AVX2-NEXT: vpshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; AVX2-NEXT: vpblendd {{.*#+}} xmm2 = xmm2[0],xmm4[1],xmm2[2],xmm4[3]
+; AVX2-NEXT: vpsllq $48, %xmm1, %xmm1
+; AVX2-NEXT: vpsrad $31, %xmm1, %xmm4
+; AVX2-NEXT: vpsrad $16, %xmm1, %xmm1
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
+; AVX2-NEXT: vpblendd {{.*#+}} xmm1 = xmm1[0],xmm4[1],xmm1[2],xmm4[3]
+; AVX2-NEXT: vpsllq $48, %xmm0, %xmm0
+; AVX2-NEXT: vpsrad $31, %xmm0, %xmm4
+; AVX2-NEXT: vpsrad $16, %xmm0, %xmm0
+; AVX2-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
+; AVX2-NEXT: vpblendd {{.*#+}} xmm0 = xmm0[0],xmm4[1],xmm0[2],xmm4[3]
+; AVX2-NEXT: vpcmpgtq %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpcmpgtq %xmm3, %xmm2, %xmm1
+; AVX2-NEXT: vpand %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpextrq $1, %xmm0, %rax
+; AVX2-NEXT: andl $1, %eax
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vmovq %xmm0, %rax
+; AVX2-NEXT: andl $1, %eax
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: v2i16:
+; AVX512: ## BB#0:
+; AVX512-NEXT: vpsllq $48, %xmm3, %xmm3
+; AVX512-NEXT: vpsraq $48, %xmm3, %xmm3
+; AVX512-NEXT: vpsllq $48, %xmm2, %xmm2
+; AVX512-NEXT: vpsraq $48, %xmm2, %xmm2
+; AVX512-NEXT: vpsllq $48, %xmm1, %xmm1
+; AVX512-NEXT: vpsraq $48, %xmm1, %xmm1
+; AVX512-NEXT: vpsllq $48, %xmm0, %xmm0
+; AVX512-NEXT: vpsraq $48, %xmm0, %xmm0
+; AVX512-NEXT: vpcmpgtq %xmm1, %xmm0, %k1
+; AVX512-NEXT: vpcmpgtq %xmm3, %xmm2, %k0 {%k1}
+; AVX512-NEXT: kmovd %k0, %eax
+; AVX512-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX512-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; AVX512-NEXT: retq
+ %x0 = icmp sgt <2 x i16> %a, %b
+ %x1 = icmp sgt <2 x i16> %c, %d
+ %y = and <2 x i1> %x0, %x1
+ %res = bitcast <2 x i1> %y to i2
+ ret i2 %res
+}
+
+define i2 @v2i32(<2 x i32> %a, <2 x i32> %b, <2 x i32> %c, <2 x i32> %d) {
+; SSE2-SSSE3-LABEL: v2i32:
+; SSE2-SSSE3: ## BB#0:
+; SSE2-SSSE3-NEXT: psllq $32, %xmm2
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm4 = xmm2[1,3,2,3]
+; SSE2-SSSE3-NEXT: psrad $31, %xmm2
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm2 = xmm2[1,3,2,3]
+; SSE2-SSSE3-NEXT: punpckldq {{.*#+}} xmm4 = xmm4[0],xmm2[0],xmm4[1],xmm2[1]
+; SSE2-SSSE3-NEXT: psllq $32, %xmm3
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm2 = xmm3[1,3,2,3]
+; SSE2-SSSE3-NEXT: psrad $31, %xmm3
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm3 = xmm3[1,3,2,3]
+; SSE2-SSSE3-NEXT: punpckldq {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1]
+; SSE2-SSSE3-NEXT: psllq $32, %xmm0
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm3 = xmm0[1,3,2,3]
+; SSE2-SSSE3-NEXT: psrad $31, %xmm0
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,3,2,3]
+; SSE2-SSSE3-NEXT: punpckldq {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1]
+; SSE2-SSSE3-NEXT: psllq $32, %xmm1
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,3,2,3]
+; SSE2-SSSE3-NEXT: psrad $31, %xmm1
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm1 = xmm1[1,3,2,3]
+; SSE2-SSSE3-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; SSE2-SSSE3-NEXT: movdqa {{.*#+}} xmm1 = [2147483648,0,2147483648,0]
+; SSE2-SSSE3-NEXT: pxor %xmm1, %xmm0
+; SSE2-SSSE3-NEXT: pxor %xmm1, %xmm3
+; SSE2-SSSE3-NEXT: movdqa %xmm3, %xmm5
+; SSE2-SSSE3-NEXT: pcmpgtd %xmm0, %xmm5
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm6 = xmm5[0,0,2,2]
+; SSE2-SSSE3-NEXT: pcmpeqd %xmm0, %xmm3
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm3[1,1,3,3]
+; SSE2-SSSE3-NEXT: pand %xmm6, %xmm0
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm3 = xmm5[1,1,3,3]
+; SSE2-SSSE3-NEXT: por %xmm0, %xmm3
+; SSE2-SSSE3-NEXT: pxor %xmm1, %xmm2
+; SSE2-SSSE3-NEXT: pxor %xmm1, %xmm4
+; SSE2-SSSE3-NEXT: movdqa %xmm4, %xmm0
+; SSE2-SSSE3-NEXT: pcmpgtd %xmm2, %xmm0
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm1 = xmm0[0,0,2,2]
+; SSE2-SSSE3-NEXT: pcmpeqd %xmm2, %xmm4
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm2 = xmm4[1,1,3,3]
+; SSE2-SSSE3-NEXT: pand %xmm1, %xmm2
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
+; SSE2-SSSE3-NEXT: por %xmm2, %xmm0
+; SSE2-SSSE3-NEXT: pand %xmm3, %xmm0
+; SSE2-SSSE3-NEXT: movq %xmm0, %rax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
+; SSE2-SSSE3-NEXT: movq %xmm0, %rax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; SSE2-SSSE3-NEXT: retq
+;
+; AVX1-LABEL: v2i32:
+; AVX1: ## BB#0:
+; AVX1-NEXT: vpsllq $32, %xmm3, %xmm3
+; AVX1-NEXT: vpsrad $31, %xmm3, %xmm4
+; AVX1-NEXT: vpshufd {{.*#+}} xmm3 = xmm3[1,1,3,3]
+; AVX1-NEXT: vpblendw {{.*#+}} xmm3 = xmm3[0,1],xmm4[2,3],xmm3[4,5],xmm4[6,7]
+; AVX1-NEXT: vpsllq $32, %xmm2, %xmm2
+; AVX1-NEXT: vpsrad $31, %xmm2, %xmm4
+; AVX1-NEXT: vpshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; AVX1-NEXT: vpblendw {{.*#+}} xmm2 = xmm2[0,1],xmm4[2,3],xmm2[4,5],xmm4[6,7]
+; AVX1-NEXT: vpsllq $32, %xmm1, %xmm1
+; AVX1-NEXT: vpsrad $31, %xmm1, %xmm4
+; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
+; AVX1-NEXT: vpblendw {{.*#+}} xmm1 = xmm1[0,1],xmm4[2,3],xmm1[4,5],xmm4[6,7]
+; AVX1-NEXT: vpsllq $32, %xmm0, %xmm0
+; AVX1-NEXT: vpsrad $31, %xmm0, %xmm4
+; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
+; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm4[2,3],xmm0[4,5],xmm4[6,7]
+; AVX1-NEXT: vpcmpgtq %xmm1, %xmm0, %xmm0
+; AVX1-NEXT: vpcmpgtq %xmm3, %xmm2, %xmm1
+; AVX1-NEXT: vpand %xmm1, %xmm0, %xmm0
+; AVX1-NEXT: vpextrq $1, %xmm0, %rax
+; AVX1-NEXT: andl $1, %eax
+; AVX1-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX1-NEXT: vmovq %xmm0, %rax
+; AVX1-NEXT: andl $1, %eax
+; AVX1-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX1-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; AVX1-NEXT: retq
+;
+; AVX2-LABEL: v2i32:
+; AVX2: ## BB#0:
+; AVX2-NEXT: vpsllq $32, %xmm3, %xmm3
+; AVX2-NEXT: vpsrad $31, %xmm3, %xmm4
+; AVX2-NEXT: vpshufd {{.*#+}} xmm3 = xmm3[1,1,3,3]
+; AVX2-NEXT: vpblendd {{.*#+}} xmm3 = xmm3[0],xmm4[1],xmm3[2],xmm4[3]
+; AVX2-NEXT: vpsllq $32, %xmm2, %xmm2
+; AVX2-NEXT: vpsrad $31, %xmm2, %xmm4
+; AVX2-NEXT: vpshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; AVX2-NEXT: vpblendd {{.*#+}} xmm2 = xmm2[0],xmm4[1],xmm2[2],xmm4[3]
+; AVX2-NEXT: vpsllq $32, %xmm1, %xmm1
+; AVX2-NEXT: vpsrad $31, %xmm1, %xmm4
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
+; AVX2-NEXT: vpblendd {{.*#+}} xmm1 = xmm1[0],xmm4[1],xmm1[2],xmm4[3]
+; AVX2-NEXT: vpsllq $32, %xmm0, %xmm0
+; AVX2-NEXT: vpsrad $31, %xmm0, %xmm4
+; AVX2-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
+; AVX2-NEXT: vpblendd {{.*#+}} xmm0 = xmm0[0],xmm4[1],xmm0[2],xmm4[3]
+; AVX2-NEXT: vpcmpgtq %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpcmpgtq %xmm3, %xmm2, %xmm1
+; AVX2-NEXT: vpand %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpextrq $1, %xmm0, %rax
+; AVX2-NEXT: andl $1, %eax
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vmovq %xmm0, %rax
+; AVX2-NEXT: andl $1, %eax
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: v2i32:
+; AVX512: ## BB#0:
+; AVX512-NEXT: vpsllq $32, %xmm3, %xmm3
+; AVX512-NEXT: vpsraq $32, %xmm3, %xmm3
+; AVX512-NEXT: vpsllq $32, %xmm2, %xmm2
+; AVX512-NEXT: vpsraq $32, %xmm2, %xmm2
+; AVX512-NEXT: vpsllq $32, %xmm1, %xmm1
+; AVX512-NEXT: vpsraq $32, %xmm1, %xmm1
+; AVX512-NEXT: vpsllq $32, %xmm0, %xmm0
+; AVX512-NEXT: vpsraq $32, %xmm0, %xmm0
+; AVX512-NEXT: vpcmpgtq %xmm1, %xmm0, %k1
+; AVX512-NEXT: vpcmpgtq %xmm3, %xmm2, %k0 {%k1}
+; AVX512-NEXT: kmovd %k0, %eax
+; AVX512-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX512-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; AVX512-NEXT: retq
+ %x0 = icmp sgt <2 x i32> %a, %b
+ %x1 = icmp sgt <2 x i32> %c, %d
+ %y = and <2 x i1> %x0, %x1
+ %res = bitcast <2 x i1> %y to i2
+ ret i2 %res
+}
+
+define i2 @v2i64(<2 x i64> %a, <2 x i64> %b, <2 x i64> %c, <2 x i64> %d) {
+; SSE2-SSSE3-LABEL: v2i64:
+; SSE2-SSSE3: ## BB#0:
+; SSE2-SSSE3-NEXT: movdqa {{.*#+}} xmm4 = [2147483648,0,2147483648,0]
+; SSE2-SSSE3-NEXT: pxor %xmm4, %xmm1
+; SSE2-SSSE3-NEXT: pxor %xmm4, %xmm0
+; SSE2-SSSE3-NEXT: movdqa %xmm0, %xmm5
+; SSE2-SSSE3-NEXT: pcmpgtd %xmm1, %xmm5
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm6 = xmm5[0,0,2,2]
+; SSE2-SSSE3-NEXT: pcmpeqd %xmm1, %xmm0
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
+; SSE2-SSSE3-NEXT: pand %xmm6, %xmm0
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm1 = xmm5[1,1,3,3]
+; SSE2-SSSE3-NEXT: por %xmm0, %xmm1
+; SSE2-SSSE3-NEXT: pxor %xmm4, %xmm3
+; SSE2-SSSE3-NEXT: pxor %xmm4, %xmm2
+; SSE2-SSSE3-NEXT: movdqa %xmm2, %xmm0
+; SSE2-SSSE3-NEXT: pcmpgtd %xmm3, %xmm0
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm4 = xmm0[0,0,2,2]
+; SSE2-SSSE3-NEXT: pcmpeqd %xmm3, %xmm2
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; SSE2-SSSE3-NEXT: pand %xmm4, %xmm2
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
+; SSE2-SSSE3-NEXT: por %xmm2, %xmm0
+; SSE2-SSSE3-NEXT: pand %xmm1, %xmm0
+; SSE2-SSSE3-NEXT: movq %xmm0, %rax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
+; SSE2-SSSE3-NEXT: movq %xmm0, %rax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; SSE2-SSSE3-NEXT: retq
+;
+; AVX12-LABEL: v2i64:
+; AVX12: ## BB#0:
+; AVX12-NEXT: vpcmpgtq %xmm1, %xmm0, %xmm0
+; AVX12-NEXT: vpcmpgtq %xmm3, %xmm2, %xmm1
+; AVX12-NEXT: vpand %xmm1, %xmm0, %xmm0
+; AVX12-NEXT: vpextrq $1, %xmm0, %rax
+; AVX12-NEXT: andl $1, %eax
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vmovq %xmm0, %rax
+; AVX12-NEXT: andl $1, %eax
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; AVX12-NEXT: retq
+;
+; AVX512-LABEL: v2i64:
+; AVX512: ## BB#0:
+; AVX512-NEXT: vpcmpgtq %xmm1, %xmm0, %k1
+; AVX512-NEXT: vpcmpgtq %xmm3, %xmm2, %k0 {%k1}
+; AVX512-NEXT: kmovd %k0, %eax
+; AVX512-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX512-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; AVX512-NEXT: retq
+ %x0 = icmp sgt <2 x i64> %a, %b
+ %x1 = icmp sgt <2 x i64> %c, %d
+ %y = and <2 x i1> %x0, %x1
+ %res = bitcast <2 x i1> %y to i2
+ ret i2 %res
+}
+
+define i2 @v2f64(<2 x double> %a, <2 x double> %b, <2 x double> %c, <2 x double> %d) {
+; SSE2-SSSE3-LABEL: v2f64:
+; SSE2-SSSE3: ## BB#0:
+; SSE2-SSSE3-NEXT: cmpltpd %xmm0, %xmm1
+; SSE2-SSSE3-NEXT: cmpltpd %xmm2, %xmm3
+; SSE2-SSSE3-NEXT: andpd %xmm1, %xmm3
+; SSE2-SSSE3-NEXT: movq %xmm3, %rax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm3[2,3,0,1]
+; SSE2-SSSE3-NEXT: movq %xmm0, %rax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; SSE2-SSSE3-NEXT: retq
+;
+; AVX12-LABEL: v2f64:
+; AVX12: ## BB#0:
+; AVX12-NEXT: vcmpltpd %xmm0, %xmm1, %xmm0
+; AVX12-NEXT: vcmpltpd %xmm2, %xmm3, %xmm1
+; AVX12-NEXT: vandpd %xmm1, %xmm0, %xmm0
+; AVX12-NEXT: vpextrq $1, %xmm0, %rax
+; AVX12-NEXT: andl $1, %eax
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vmovq %xmm0, %rax
+; AVX12-NEXT: andl $1, %eax
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; AVX12-NEXT: retq
+;
+; AVX512-LABEL: v2f64:
+; AVX512: ## BB#0:
+; AVX512-NEXT: vcmpltpd %xmm0, %xmm1, %k1
+; AVX512-NEXT: vcmpltpd %xmm2, %xmm3, %k0 {%k1}
+; AVX512-NEXT: kmovd %k0, %eax
+; AVX512-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX512-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; AVX512-NEXT: retq
+ %x0 = fcmp ogt <2 x double> %a, %b
+ %x1 = fcmp ogt <2 x double> %c, %d
+ %y = and <2 x i1> %x0, %x1
+ %res = bitcast <2 x i1> %y to i2
+ ret i2 %res
+}
+
+define i4 @v4i8(<4 x i8> %a, <4 x i8> %b, <4 x i8> %c, <4 x i8> %d) {
+; SSE2-SSSE3-LABEL: v4i8:
+; SSE2-SSSE3: ## BB#0:
+; SSE2-SSSE3-NEXT: pslld $24, %xmm3
+; SSE2-SSSE3-NEXT: psrad $24, %xmm3
+; SSE2-SSSE3-NEXT: pslld $24, %xmm2
+; SSE2-SSSE3-NEXT: psrad $24, %xmm2
+; SSE2-SSSE3-NEXT: pslld $24, %xmm1
+; SSE2-SSSE3-NEXT: psrad $24, %xmm1
+; SSE2-SSSE3-NEXT: pslld $24, %xmm0
+; SSE2-SSSE3-NEXT: psrad $24, %xmm0
+; SSE2-SSSE3-NEXT: pcmpgtd %xmm1, %xmm0
+; SSE2-SSSE3-NEXT: pcmpgtd %xmm3, %xmm2
+; SSE2-SSSE3-NEXT: pand %xmm0, %xmm2
+; SSE2-SSSE3-NEXT: movd %xmm2, %eax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm2[3,1,2,3]
+; SSE2-SSSE3-NEXT: movd %xmm0, %eax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm2[2,3,0,1]
+; SSE2-SSSE3-NEXT: movd %xmm0, %eax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm2[1,1,2,3]
+; SSE2-SSSE3-NEXT: movd %xmm0, %eax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; SSE2-SSSE3-NEXT: retq
+;
+; AVX12-LABEL: v4i8:
+; AVX12: ## BB#0:
+; AVX12-NEXT: vpslld $24, %xmm3, %xmm3
+; AVX12-NEXT: vpsrad $24, %xmm3, %xmm3
+; AVX12-NEXT: vpslld $24, %xmm2, %xmm2
+; AVX12-NEXT: vpsrad $24, %xmm2, %xmm2
+; AVX12-NEXT: vpslld $24, %xmm1, %xmm1
+; AVX12-NEXT: vpsrad $24, %xmm1, %xmm1
+; AVX12-NEXT: vpslld $24, %xmm0, %xmm0
+; AVX12-NEXT: vpsrad $24, %xmm0, %xmm0
+; AVX12-NEXT: vpcmpgtd %xmm1, %xmm0, %xmm0
+; AVX12-NEXT: vpcmpgtd %xmm3, %xmm2, %xmm1
+; AVX12-NEXT: vpand %xmm1, %xmm0, %xmm0
+; AVX12-NEXT: vpextrd $3, %xmm0, %eax
+; AVX12-NEXT: andl $1, %eax
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vpextrd $2, %xmm0, %eax
+; AVX12-NEXT: andl $1, %eax
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vpextrd $1, %xmm0, %eax
+; AVX12-NEXT: andl $1, %eax
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vmovd %xmm0, %eax
+; AVX12-NEXT: andl $1, %eax
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; AVX12-NEXT: retq
+;
+; AVX512-LABEL: v4i8:
+; AVX512: ## BB#0:
+; AVX512-NEXT: vpslld $24, %xmm3, %xmm3
+; AVX512-NEXT: vpsrad $24, %xmm3, %xmm3
+; AVX512-NEXT: vpslld $24, %xmm2, %xmm2
+; AVX512-NEXT: vpsrad $24, %xmm2, %xmm2
+; AVX512-NEXT: vpslld $24, %xmm1, %xmm1
+; AVX512-NEXT: vpsrad $24, %xmm1, %xmm1
+; AVX512-NEXT: vpslld $24, %xmm0, %xmm0
+; AVX512-NEXT: vpsrad $24, %xmm0, %xmm0
+; AVX512-NEXT: vpcmpgtd %xmm1, %xmm0, %k1
+; AVX512-NEXT: vpcmpgtd %xmm3, %xmm2, %k0 {%k1}
+; AVX512-NEXT: kmovd %k0, %eax
+; AVX512-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX512-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; AVX512-NEXT: retq
+ %x0 = icmp sgt <4 x i8> %a, %b
+ %x1 = icmp sgt <4 x i8> %c, %d
+ %y = and <4 x i1> %x0, %x1
+ %res = bitcast <4 x i1> %y to i4
+ ret i4 %res
+}
+
+define i4 @v4i16(<4 x i16> %a, <4 x i16> %b, <4 x i16> %c, <4 x i16> %d) {
+; SSE2-SSSE3-LABEL: v4i16:
+; SSE2-SSSE3: ## BB#0:
+; SSE2-SSSE3-NEXT: pslld $16, %xmm3
+; SSE2-SSSE3-NEXT: psrad $16, %xmm3
+; SSE2-SSSE3-NEXT: pslld $16, %xmm2
+; SSE2-SSSE3-NEXT: psrad $16, %xmm2
+; SSE2-SSSE3-NEXT: pslld $16, %xmm1
+; SSE2-SSSE3-NEXT: psrad $16, %xmm1
+; SSE2-SSSE3-NEXT: pslld $16, %xmm0
+; SSE2-SSSE3-NEXT: psrad $16, %xmm0
+; SSE2-SSSE3-NEXT: pcmpgtd %xmm1, %xmm0
+; SSE2-SSSE3-NEXT: pcmpgtd %xmm3, %xmm2
+; SSE2-SSSE3-NEXT: pand %xmm0, %xmm2
+; SSE2-SSSE3-NEXT: movd %xmm2, %eax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm2[3,1,2,3]
+; SSE2-SSSE3-NEXT: movd %xmm0, %eax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm2[2,3,0,1]
+; SSE2-SSSE3-NEXT: movd %xmm0, %eax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm2[1,1,2,3]
+; SSE2-SSSE3-NEXT: movd %xmm0, %eax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; SSE2-SSSE3-NEXT: retq
+;
+; AVX12-LABEL: v4i16:
+; AVX12: ## BB#0:
+; AVX12-NEXT: vpslld $16, %xmm3, %xmm3
+; AVX12-NEXT: vpsrad $16, %xmm3, %xmm3
+; AVX12-NEXT: vpslld $16, %xmm2, %xmm2
+; AVX12-NEXT: vpsrad $16, %xmm2, %xmm2
+; AVX12-NEXT: vpslld $16, %xmm1, %xmm1
+; AVX12-NEXT: vpsrad $16, %xmm1, %xmm1
+; AVX12-NEXT: vpslld $16, %xmm0, %xmm0
+; AVX12-NEXT: vpsrad $16, %xmm0, %xmm0
+; AVX12-NEXT: vpcmpgtd %xmm1, %xmm0, %xmm0
+; AVX12-NEXT: vpcmpgtd %xmm3, %xmm2, %xmm1
+; AVX12-NEXT: vpand %xmm1, %xmm0, %xmm0
+; AVX12-NEXT: vpextrd $3, %xmm0, %eax
+; AVX12-NEXT: andl $1, %eax
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vpextrd $2, %xmm0, %eax
+; AVX12-NEXT: andl $1, %eax
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vpextrd $1, %xmm0, %eax
+; AVX12-NEXT: andl $1, %eax
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vmovd %xmm0, %eax
+; AVX12-NEXT: andl $1, %eax
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; AVX12-NEXT: retq
+;
+; AVX512-LABEL: v4i16:
+; AVX512: ## BB#0:
+; AVX512-NEXT: vpslld $16, %xmm3, %xmm3
+; AVX512-NEXT: vpsrad $16, %xmm3, %xmm3
+; AVX512-NEXT: vpslld $16, %xmm2, %xmm2
+; AVX512-NEXT: vpsrad $16, %xmm2, %xmm2
+; AVX512-NEXT: vpslld $16, %xmm1, %xmm1
+; AVX512-NEXT: vpsrad $16, %xmm1, %xmm1
+; AVX512-NEXT: vpslld $16, %xmm0, %xmm0
+; AVX512-NEXT: vpsrad $16, %xmm0, %xmm0
+; AVX512-NEXT: vpcmpgtd %xmm1, %xmm0, %k1
+; AVX512-NEXT: vpcmpgtd %xmm3, %xmm2, %k0 {%k1}
+; AVX512-NEXT: kmovd %k0, %eax
+; AVX512-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX512-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; AVX512-NEXT: retq
+ %x0 = icmp sgt <4 x i16> %a, %b
+ %x1 = icmp sgt <4 x i16> %c, %d
+ %y = and <4 x i1> %x0, %x1
+ %res = bitcast <4 x i1> %y to i4
+ ret i4 %res
+}
+
+define i8 @v8i8(<8 x i8> %a, <8 x i8> %b, <8 x i8> %c, <8 x i8> %d) {
+; SSE2-SSSE3-LABEL: v8i8:
+; SSE2-SSSE3: ## BB#0:
+; SSE2-SSSE3-NEXT: psllw $8, %xmm3
+; SSE2-SSSE3-NEXT: psraw $8, %xmm3
+; SSE2-SSSE3-NEXT: psllw $8, %xmm2
+; SSE2-SSSE3-NEXT: psraw $8, %xmm2
+; SSE2-SSSE3-NEXT: psllw $8, %xmm1
+; SSE2-SSSE3-NEXT: psraw $8, %xmm1
+; SSE2-SSSE3-NEXT: psllw $8, %xmm0
+; SSE2-SSSE3-NEXT: psraw $8, %xmm0
+; SSE2-SSSE3-NEXT: pcmpgtw %xmm1, %xmm0
+; SSE2-SSSE3-NEXT: pcmpgtw %xmm3, %xmm2
+; SSE2-SSSE3-NEXT: pand %xmm0, %xmm2
+; SSE2-SSSE3-NEXT: pextrw $7, %xmm2, %eax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: pextrw $6, %xmm2, %eax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: pextrw $5, %xmm2, %eax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: pextrw $4, %xmm2, %eax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: pextrw $3, %xmm2, %eax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: pextrw $2, %xmm2, %eax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: pextrw $1, %xmm2, %eax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: movd %xmm2, %eax
+; SSE2-SSSE3-NEXT: andl $1, %eax
+; SSE2-SSSE3-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; SSE2-SSSE3-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; SSE2-SSSE3-NEXT: retq
+;
+; AVX12-LABEL: v8i8:
+; AVX12: ## BB#0:
+; AVX12-NEXT: vpsllw $8, %xmm3, %xmm3
+; AVX12-NEXT: vpsraw $8, %xmm3, %xmm3
+; AVX12-NEXT: vpsllw $8, %xmm2, %xmm2
+; AVX12-NEXT: vpsraw $8, %xmm2, %xmm2
+; AVX12-NEXT: vpsllw $8, %xmm1, %xmm1
+; AVX12-NEXT: vpsraw $8, %xmm1, %xmm1
+; AVX12-NEXT: vpsllw $8, %xmm0, %xmm0
+; AVX12-NEXT: vpsraw $8, %xmm0, %xmm0
+; AVX12-NEXT: vpcmpgtw %xmm1, %xmm0, %xmm0
+; AVX12-NEXT: vpcmpgtw %xmm3, %xmm2, %xmm1
+; AVX12-NEXT: vpand %xmm1, %xmm0, %xmm0
+; AVX12-NEXT: vpextrw $7, %xmm0, %eax
+; AVX12-NEXT: andl $1, %eax
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vpextrw $6, %xmm0, %eax
+; AVX12-NEXT: andl $1, %eax
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vpextrw $5, %xmm0, %eax
+; AVX12-NEXT: andl $1, %eax
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vpextrw $4, %xmm0, %eax
+; AVX12-NEXT: andl $1, %eax
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vpextrw $3, %xmm0, %eax
+; AVX12-NEXT: andl $1, %eax
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vpextrw $2, %xmm0, %eax
+; AVX12-NEXT: andl $1, %eax
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vpextrw $1, %xmm0, %eax
+; AVX12-NEXT: andl $1, %eax
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: vmovd %xmm0, %eax
+; AVX12-NEXT: andl $1, %eax
+; AVX12-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX12-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; AVX12-NEXT: retq
+;
+; AVX512-LABEL: v8i8:
+; AVX512: ## BB#0:
+; AVX512-NEXT: vpsllw $8, %xmm3, %xmm3
+; AVX512-NEXT: vpsraw $8, %xmm3, %xmm3
+; AVX512-NEXT: vpsllw $8, %xmm2, %xmm2
+; AVX512-NEXT: vpsraw $8, %xmm2, %xmm2
+; AVX512-NEXT: vpsllw $8, %xmm1, %xmm1
+; AVX512-NEXT: vpsraw $8, %xmm1, %xmm1
+; AVX512-NEXT: vpsllw $8, %xmm0, %xmm0
+; AVX512-NEXT: vpsraw $8, %xmm0, %xmm0
+; AVX512-NEXT: vpcmpgtw %xmm1, %xmm0, %k1
+; AVX512-NEXT: vpcmpgtw %xmm3, %xmm2, %k0 {%k1}
+; AVX512-NEXT: kmovd %k0, %eax
+; AVX512-NEXT: ## kill: %AL<def> %AL<kill> %EAX<kill>
+; AVX512-NEXT: retq
+ %x0 = icmp sgt <8 x i8> %a, %b
+ %x1 = icmp sgt <8 x i8> %c, %d
+ %y = and <8 x i1> %x0, %x1
+ %res = bitcast <8 x i1> %y to i8
+ ret i8 %res
+}
diff --git a/test/CodeGen/X86/bitcast-and-setcc-256.ll b/test/CodeGen/X86/bitcast-and-setcc-256.ll
new file mode 100644
index 000000000000..06b1a76f6bae
--- /dev/null
+++ b/test/CodeGen/X86/bitcast-and-setcc-256.ll
@@ -0,0 +1,403 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=x86_64-apple-darwin -mcpu=x86-64 -mattr=+avx2 < %s | FileCheck %s --check-prefix=AVX2
+; RUN: llc -mtriple=x86_64-apple-darwin -mcpu=x86-64 -mattr=+avx512f,+avx512vl,+avx512bw < %s | FileCheck %s --check-prefix=AVX512
+
+define i4 @v4i64(<4 x i64> %a, <4 x i64> %b, <4 x i64> %c, <4 x i64> %d) {
+; AVX2-LABEL: v4i64:
+; AVX2: ## BB#0:
+; AVX2-NEXT: vpcmpgtq %ymm1, %ymm0, %ymm0
+; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX2-NEXT: vpacksswb %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpcmpgtq %ymm3, %ymm2, %ymm1
+; AVX2-NEXT: vextracti128 $1, %ymm1, %xmm2
+; AVX2-NEXT: vpacksswb %xmm2, %xmm1, %xmm1
+; AVX2-NEXT: vpand %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpextrd $3, %xmm0, %eax
+; AVX2-NEXT: andl $1, %eax
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vpextrd $2, %xmm0, %eax
+; AVX2-NEXT: andl $1, %eax
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vpextrd $1, %xmm0, %eax
+; AVX2-NEXT: andl $1, %eax
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vmovd %xmm0, %eax
+; AVX2-NEXT: andl $1, %eax
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; AVX2-NEXT: vzeroupper
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: v4i64:
+; AVX512: ## BB#0:
+; AVX512-NEXT: vpcmpgtq %ymm1, %ymm0, %k1
+; AVX512-NEXT: vpcmpgtq %ymm3, %ymm2, %k0 {%k1}
+; AVX512-NEXT: kmovd %k0, %eax
+; AVX512-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX512-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; AVX512-NEXT: vzeroupper
+; AVX512-NEXT: retq
+ %x0 = icmp sgt <4 x i64> %a, %b
+ %x1 = icmp sgt <4 x i64> %c, %d
+ %y = and <4 x i1> %x0, %x1
+ %res = bitcast <4 x i1> %y to i4
+ ret i4 %res
+}
+
+define i4 @v4f64(<4 x double> %a, <4 x double> %b, <4 x double> %c, <4 x double> %d) {
+; AVX2-LABEL: v4f64:
+; AVX2: ## BB#0:
+; AVX2-NEXT: vcmpltpd %ymm0, %ymm1, %ymm0
+; AVX2-NEXT: vextractf128 $1, %ymm0, %xmm1
+; AVX2-NEXT: vpacksswb %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vcmpltpd %ymm2, %ymm3, %ymm1
+; AVX2-NEXT: vextractf128 $1, %ymm1, %xmm2
+; AVX2-NEXT: vpacksswb %xmm2, %xmm1, %xmm1
+; AVX2-NEXT: vpand %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpextrd $3, %xmm0, %eax
+; AVX2-NEXT: andl $1, %eax
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vpextrd $2, %xmm0, %eax
+; AVX2-NEXT: andl $1, %eax
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vpextrd $1, %xmm0, %eax
+; AVX2-NEXT: andl $1, %eax
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vmovd %xmm0, %eax
+; AVX2-NEXT: andl $1, %eax
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; AVX2-NEXT: vzeroupper
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: v4f64:
+; AVX512: ## BB#0:
+; AVX512-NEXT: vcmpltpd %ymm0, %ymm1, %k1
+; AVX512-NEXT: vcmpltpd %ymm2, %ymm3, %k0 {%k1}
+; AVX512-NEXT: kmovd %k0, %eax
+; AVX512-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX512-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; AVX512-NEXT: vzeroupper
+; AVX512-NEXT: retq
+ %x0 = fcmp ogt <4 x double> %a, %b
+ %x1 = fcmp ogt <4 x double> %c, %d
+ %y = and <4 x i1> %x0, %x1
+ %res = bitcast <4 x i1> %y to i4
+ ret i4 %res
+}
+
+define i16 @v16i16(<16 x i16> %a, <16 x i16> %b, <16 x i16> %c, <16 x i16> %d) {
+; AVX2-LABEL: v16i16:
+; AVX2: ## BB#0:
+; AVX2-NEXT: vpcmpgtw %ymm1, %ymm0, %ymm0
+; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX2-NEXT: vpacksswb %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpcmpgtw %ymm3, %ymm2, %ymm1
+; AVX2-NEXT: vextracti128 $1, %ymm1, %xmm2
+; AVX2-NEXT: vpacksswb %xmm2, %xmm1, %xmm1
+; AVX2-NEXT: vpand %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpextrb $15, %xmm0, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vpextrb $14, %xmm0, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vpextrb $13, %xmm0, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vpextrb $12, %xmm0, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vpextrb $11, %xmm0, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vpextrb $10, %xmm0, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vpextrb $9, %xmm0, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vpextrb $8, %xmm0, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vpextrb $7, %xmm0, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vpextrb $6, %xmm0, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vpextrb $5, %xmm0, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vpextrb $4, %xmm0, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vpextrb $3, %xmm0, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vpextrb $2, %xmm0, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vpextrb $1, %xmm0, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vpextrb $0, %xmm0, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: movzwl -{{[0-9]+}}(%rsp), %eax
+; AVX2-NEXT: vzeroupper
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: v16i16:
+; AVX512: ## BB#0:
+; AVX512-NEXT: vpcmpgtw %ymm1, %ymm0, %k1
+; AVX512-NEXT: vpcmpgtw %ymm3, %ymm2, %k0 {%k1}
+; AVX512-NEXT: kmovd %k0, %eax
+; AVX512-NEXT: ## kill: %AX<def> %AX<kill> %EAX<kill>
+; AVX512-NEXT: vzeroupper
+; AVX512-NEXT: retq
+ %x0 = icmp sgt <16 x i16> %a, %b
+ %x1 = icmp sgt <16 x i16> %c, %d
+ %y = and <16 x i1> %x0, %x1
+ %res = bitcast <16 x i1> %y to i16
+ ret i16 %res
+}
+
+define i8 @v8i32(<8 x i32> %a, <8 x i32> %b, <8 x i32> %c, <8 x i32> %d) {
+; AVX2-LABEL: v8i32:
+; AVX2: ## BB#0:
+; AVX2-NEXT: vpcmpgtd %ymm1, %ymm0, %ymm0
+; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX2-NEXT: vpacksswb %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpcmpgtd %ymm3, %ymm2, %ymm1
+; AVX2-NEXT: vextracti128 $1, %ymm1, %xmm2
+; AVX2-NEXT: vpacksswb %xmm2, %xmm1, %xmm1
+; AVX2-NEXT: vpand %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpextrw $7, %xmm0, %eax
+; AVX2-NEXT: andl $1, %eax
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vpextrw $6, %xmm0, %eax
+; AVX2-NEXT: andl $1, %eax
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vpextrw $5, %xmm0, %eax
+; AVX2-NEXT: andl $1, %eax
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vpextrw $4, %xmm0, %eax
+; AVX2-NEXT: andl $1, %eax
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vpextrw $3, %xmm0, %eax
+; AVX2-NEXT: andl $1, %eax
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vpextrw $2, %xmm0, %eax
+; AVX2-NEXT: andl $1, %eax
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vpextrw $1, %xmm0, %eax
+; AVX2-NEXT: andl $1, %eax
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vmovd %xmm0, %eax
+; AVX2-NEXT: andl $1, %eax
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; AVX2-NEXT: vzeroupper
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: v8i32:
+; AVX512: ## BB#0:
+; AVX512-NEXT: vpcmpgtd %ymm1, %ymm0, %k1
+; AVX512-NEXT: vpcmpgtd %ymm3, %ymm2, %k0 {%k1}
+; AVX512-NEXT: kmovd %k0, %eax
+; AVX512-NEXT: ## kill: %AL<def> %AL<kill> %EAX<kill>
+; AVX512-NEXT: vzeroupper
+; AVX512-NEXT: retq
+ %x0 = icmp sgt <8 x i32> %a, %b
+ %x1 = icmp sgt <8 x i32> %c, %d
+ %y = and <8 x i1> %x0, %x1
+ %res = bitcast <8 x i1> %y to i8
+ ret i8 %res
+}
+
+define i8 @v8f32(<8 x float> %a, <8 x float> %b, <8 x float> %c, <8 x float> %d) {
+; AVX2-LABEL: v8f32:
+; AVX2: ## BB#0:
+; AVX2-NEXT: vcmpltps %ymm0, %ymm1, %ymm0
+; AVX2-NEXT: vextractf128 $1, %ymm0, %xmm1
+; AVX2-NEXT: vpacksswb %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vcmpltps %ymm2, %ymm3, %ymm1
+; AVX2-NEXT: vextractf128 $1, %ymm1, %xmm2
+; AVX2-NEXT: vpacksswb %xmm2, %xmm1, %xmm1
+; AVX2-NEXT: vpand %xmm1, %xmm0, %xmm0
+; AVX2-NEXT: vpextrw $7, %xmm0, %eax
+; AVX2-NEXT: andl $1, %eax
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vpextrw $6, %xmm0, %eax
+; AVX2-NEXT: andl $1, %eax
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vpextrw $5, %xmm0, %eax
+; AVX2-NEXT: andl $1, %eax
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vpextrw $4, %xmm0, %eax
+; AVX2-NEXT: andl $1, %eax
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vpextrw $3, %xmm0, %eax
+; AVX2-NEXT: andl $1, %eax
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vpextrw $2, %xmm0, %eax
+; AVX2-NEXT: andl $1, %eax
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vpextrw $1, %xmm0, %eax
+; AVX2-NEXT: andl $1, %eax
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: vmovd %xmm0, %eax
+; AVX2-NEXT: andl $1, %eax
+; AVX2-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX2-NEXT: movb -{{[0-9]+}}(%rsp), %al
+; AVX2-NEXT: vzeroupper
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: v8f32:
+; AVX512: ## BB#0:
+; AVX512-NEXT: vcmpltps %ymm0, %ymm1, %k1
+; AVX512-NEXT: vcmpltps %ymm2, %ymm3, %k0 {%k1}
+; AVX512-NEXT: kmovd %k0, %eax
+; AVX512-NEXT: ## kill: %AL<def> %AL<kill> %EAX<kill>
+; AVX512-NEXT: vzeroupper
+; AVX512-NEXT: retq
+ %x0 = fcmp ogt <8 x float> %a, %b
+ %x1 = fcmp ogt <8 x float> %c, %d
+ %y = and <8 x i1> %x0, %x1
+ %res = bitcast <8 x i1> %y to i8
+ ret i8 %res
+}
+
+define i32 @v32i8(<32 x i8> %a, <32 x i8> %b, <32 x i8> %c, <32 x i8> %d) {
+; AVX2-LABEL: v32i8:
+; AVX2: ## BB#0:
+; AVX2-NEXT: pushq %rbp
+; AVX2-NEXT: Lcfi0:
+; AVX2-NEXT: .cfi_def_cfa_offset 16
+; AVX2-NEXT: Lcfi1:
+; AVX2-NEXT: .cfi_offset %rbp, -16
+; AVX2-NEXT: movq %rsp, %rbp
+; AVX2-NEXT: Lcfi2:
+; AVX2-NEXT: .cfi_def_cfa_register %rbp
+; AVX2-NEXT: andq $-32, %rsp
+; AVX2-NEXT: subq $32, %rsp
+; AVX2-NEXT: vpcmpgtb %ymm1, %ymm0, %ymm0
+; AVX2-NEXT: vpcmpgtb %ymm3, %ymm2, %ymm1
+; AVX2-NEXT: vpand %ymm1, %ymm0, %ymm0
+; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX2-NEXT: vpextrb $15, %xmm1, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, (%rsp)
+; AVX2-NEXT: vpextrb $14, %xmm1, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, (%rsp)
+; AVX2-NEXT: vpextrb $13, %xmm1, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, (%rsp)
+; AVX2-NEXT: vpextrb $12, %xmm1, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, (%rsp)
+; AVX2-NEXT: vpextrb $11, %xmm1, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, (%rsp)
+; AVX2-NEXT: vpextrb $10, %xmm1, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, (%rsp)
+; AVX2-NEXT: vpextrb $9, %xmm1, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, (%rsp)
+; AVX2-NEXT: vpextrb $8, %xmm1, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, (%rsp)
+; AVX2-NEXT: vpextrb $7, %xmm1, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, (%rsp)
+; AVX2-NEXT: vpextrb $6, %xmm1, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, (%rsp)
+; AVX2-NEXT: vpextrb $5, %xmm1, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, (%rsp)
+; AVX2-NEXT: vpextrb $4, %xmm1, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, (%rsp)
+; AVX2-NEXT: vpextrb $3, %xmm1, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, (%rsp)
+; AVX2-NEXT: vpextrb $2, %xmm1, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, (%rsp)
+; AVX2-NEXT: vpextrb $1, %xmm1, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, (%rsp)
+; AVX2-NEXT: vpextrb $0, %xmm1, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, (%rsp)
+; AVX2-NEXT: vpextrb $15, %xmm0, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, (%rsp)
+; AVX2-NEXT: vpextrb $14, %xmm0, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, (%rsp)
+; AVX2-NEXT: vpextrb $13, %xmm0, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, (%rsp)
+; AVX2-NEXT: vpextrb $12, %xmm0, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, (%rsp)
+; AVX2-NEXT: vpextrb $11, %xmm0, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, (%rsp)
+; AVX2-NEXT: vpextrb $10, %xmm0, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, (%rsp)
+; AVX2-NEXT: vpextrb $9, %xmm0, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, (%rsp)
+; AVX2-NEXT: vpextrb $8, %xmm0, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, (%rsp)
+; AVX2-NEXT: vpextrb $7, %xmm0, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, (%rsp)
+; AVX2-NEXT: vpextrb $6, %xmm0, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, (%rsp)
+; AVX2-NEXT: vpextrb $5, %xmm0, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, (%rsp)
+; AVX2-NEXT: vpextrb $4, %xmm0, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, (%rsp)
+; AVX2-NEXT: vpextrb $3, %xmm0, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, (%rsp)
+; AVX2-NEXT: vpextrb $2, %xmm0, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, (%rsp)
+; AVX2-NEXT: vpextrb $1, %xmm0, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, (%rsp)
+; AVX2-NEXT: vpextrb $0, %xmm0, %eax
+; AVX2-NEXT: andb $1, %al
+; AVX2-NEXT: movb %al, (%rsp)
+; AVX2-NEXT: movl (%rsp), %eax
+; AVX2-NEXT: movq %rbp, %rsp
+; AVX2-NEXT: popq %rbp
+; AVX2-NEXT: vzeroupper
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: v32i8:
+; AVX512: ## BB#0:
+; AVX512-NEXT: vpcmpgtb %ymm1, %ymm0, %k1
+; AVX512-NEXT: vpcmpgtb %ymm3, %ymm2, %k0 {%k1}
+; AVX512-NEXT: kmovd %k0, %eax
+; AVX512-NEXT: vzeroupper
+; AVX512-NEXT: retq
+ %x0 = icmp sgt <32 x i8> %a, %b
+ %x1 = icmp sgt <32 x i8> %c, %d
+ %y = and <32 x i1> %x0, %x1
+ %res = bitcast <32 x i1> %y to i32
+ ret i32 %res
+}
diff --git a/test/CodeGen/X86/mul-constant-i16.ll b/test/CodeGen/X86/mul-constant-i16.ll
index e3e2737cf3e6..6d2465ddd3a8 100644
--- a/test/CodeGen/X86/mul-constant-i16.ll
+++ b/test/CodeGen/X86/mul-constant-i16.ll
@@ -188,13 +188,16 @@ define i16 @test_mul_by_11(i16 %x) {
; X86-LABEL: test_mul_by_11:
; X86: # BB#0:
; X86-NEXT: movzwl {{[0-9]+}}(%esp), %eax
-; X86-NEXT: imull $11, %eax, %eax
+; X86-NEXT: leal (%eax,%eax,4), %ecx
+; X86-NEXT: leal (%eax,%ecx,2), %eax
; X86-NEXT: # kill: %AX<def> %AX<kill> %EAX<kill>
; X86-NEXT: retl
;
; X64-LABEL: test_mul_by_11:
; X64: # BB#0:
-; X64-NEXT: imull $11, %edi, %eax
+; X64-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-NEXT: leal (%rdi,%rdi,4), %eax
+; X64-NEXT: leal (%rdi,%rax,2), %eax
; X64-NEXT: # kill: %AX<def> %AX<kill> %EAX<kill>
; X64-NEXT: retq
%mul = mul nsw i16 %x, 11
@@ -225,13 +228,16 @@ define i16 @test_mul_by_13(i16 %x) {
; X86-LABEL: test_mul_by_13:
; X86: # BB#0:
; X86-NEXT: movzwl {{[0-9]+}}(%esp), %eax
-; X86-NEXT: imull $13, %eax, %eax
+; X86-NEXT: leal (%eax,%eax,2), %ecx
+; X86-NEXT: leal (%eax,%ecx,4), %eax
; X86-NEXT: # kill: %AX<def> %AX<kill> %EAX<kill>
; X86-NEXT: retl
;
; X64-LABEL: test_mul_by_13:
; X64: # BB#0:
-; X64-NEXT: imull $13, %edi, %eax
+; X64-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-NEXT: leal (%rdi,%rdi,2), %eax
+; X64-NEXT: leal (%rdi,%rax,4), %eax
; X64-NEXT: # kill: %AX<def> %AX<kill> %EAX<kill>
; X64-NEXT: retq
%mul = mul nsw i16 %x, 13
@@ -241,14 +247,19 @@ define i16 @test_mul_by_13(i16 %x) {
define i16 @test_mul_by_14(i16 %x) {
; X86-LABEL: test_mul_by_14:
; X86: # BB#0:
-; X86-NEXT: movzwl {{[0-9]+}}(%esp), %eax
-; X86-NEXT: imull $14, %eax, %eax
+; X86-NEXT: movzwl {{[0-9]+}}(%esp), %ecx
+; X86-NEXT: leal (%ecx,%ecx,2), %eax
+; X86-NEXT: leal (%ecx,%eax,4), %eax
+; X86-NEXT: addl %ecx, %eax
; X86-NEXT: # kill: %AX<def> %AX<kill> %EAX<kill>
; X86-NEXT: retl
;
; X64-LABEL: test_mul_by_14:
; X64: # BB#0:
-; X64-NEXT: imull $14, %edi, %eax
+; X64-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-NEXT: leal (%rdi,%rdi,2), %eax
+; X64-NEXT: leal (%rdi,%rax,4), %eax
+; X64-NEXT: addl %edi, %eax
; X64-NEXT: # kill: %AX<def> %AX<kill> %EAX<kill>
; X64-NEXT: retq
%mul = mul nsw i16 %x, 14
@@ -338,14 +349,19 @@ define i16 @test_mul_by_19(i16 %x) {
; X86-LABEL: test_mul_by_19:
; X86: # BB#0:
; X86-NEXT: movzwl {{[0-9]+}}(%esp), %eax
-; X86-NEXT: imull $19, %eax, %eax
+; X86-NEXT: leal (%eax,%eax,4), %ecx
+; X86-NEXT: shll $2, %ecx
+; X86-NEXT: subl %ecx, %eax
; X86-NEXT: # kill: %AX<def> %AX<kill> %EAX<kill>
; X86-NEXT: retl
;
; X64-LABEL: test_mul_by_19:
; X64: # BB#0:
-; X64-NEXT: imull $19, %edi, %eax
-; X64-NEXT: # kill: %AX<def> %AX<kill> %EAX<kill>
+; X64-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-NEXT: leal (%rdi,%rdi,4), %eax
+; X64-NEXT: shll $2, %eax
+; X64-NEXT: subl %eax, %edi
+; X64-NEXT: movl %edi, %eax
; X64-NEXT: retq
%mul = mul nsw i16 %x, 19
ret i16 %mul
@@ -375,13 +391,16 @@ define i16 @test_mul_by_21(i16 %x) {
; X86-LABEL: test_mul_by_21:
; X86: # BB#0:
; X86-NEXT: movzwl {{[0-9]+}}(%esp), %eax
-; X86-NEXT: imull $21, %eax, %eax
+; X86-NEXT: leal (%eax,%eax,4), %ecx
+; X86-NEXT: leal (%eax,%ecx,4), %eax
; X86-NEXT: # kill: %AX<def> %AX<kill> %EAX<kill>
; X86-NEXT: retl
;
; X64-LABEL: test_mul_by_21:
; X64: # BB#0:
-; X64-NEXT: imull $21, %edi, %eax
+; X64-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-NEXT: leal (%rdi,%rdi,4), %eax
+; X64-NEXT: leal (%rdi,%rax,4), %eax
; X64-NEXT: # kill: %AX<def> %AX<kill> %EAX<kill>
; X64-NEXT: retq
%mul = mul nsw i16 %x, 21
@@ -391,14 +410,19 @@ define i16 @test_mul_by_21(i16 %x) {
define i16 @test_mul_by_22(i16 %x) {
; X86-LABEL: test_mul_by_22:
; X86: # BB#0:
-; X86-NEXT: movzwl {{[0-9]+}}(%esp), %eax
-; X86-NEXT: imull $22, %eax, %eax
+; X86-NEXT: movzwl {{[0-9]+}}(%esp), %ecx
+; X86-NEXT: leal (%ecx,%ecx,4), %eax
+; X86-NEXT: leal (%ecx,%eax,4), %eax
+; X86-NEXT: addl %ecx, %eax
; X86-NEXT: # kill: %AX<def> %AX<kill> %EAX<kill>
; X86-NEXT: retl
;
; X64-LABEL: test_mul_by_22:
; X64: # BB#0:
-; X64-NEXT: imull $22, %edi, %eax
+; X64-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-NEXT: leal (%rdi,%rdi,4), %eax
+; X64-NEXT: leal (%rdi,%rax,4), %eax
+; X64-NEXT: addl %edi, %eax
; X64-NEXT: # kill: %AX<def> %AX<kill> %EAX<kill>
; X64-NEXT: retq
%mul = mul nsw i16 %x, 22
@@ -409,14 +433,19 @@ define i16 @test_mul_by_23(i16 %x) {
; X86-LABEL: test_mul_by_23:
; X86: # BB#0:
; X86-NEXT: movzwl {{[0-9]+}}(%esp), %eax
-; X86-NEXT: imull $23, %eax, %eax
+; X86-NEXT: leal (%eax,%eax,2), %ecx
+; X86-NEXT: shll $3, %ecx
+; X86-NEXT: subl %ecx, %eax
; X86-NEXT: # kill: %AX<def> %AX<kill> %EAX<kill>
; X86-NEXT: retl
;
; X64-LABEL: test_mul_by_23:
; X64: # BB#0:
-; X64-NEXT: imull $23, %edi, %eax
-; X64-NEXT: # kill: %AX<def> %AX<kill> %EAX<kill>
+; X64-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-NEXT: leal (%rdi,%rdi,2), %eax
+; X64-NEXT: shll $3, %eax
+; X64-NEXT: subl %eax, %edi
+; X64-NEXT: movl %edi, %eax
; X64-NEXT: retq
%mul = mul nsw i16 %x, 23
ret i16 %mul
@@ -466,14 +495,19 @@ define i16 @test_mul_by_26(i16 %x) {
; X86-LABEL: test_mul_by_26:
; X86: # BB#0:
; X86-NEXT: movzwl {{[0-9]+}}(%esp), %eax
-; X86-NEXT: imull $26, %eax, %eax
+; X86-NEXT: leal (%eax,%eax,8), %ecx
+; X86-NEXT: leal (%ecx,%ecx,2), %ecx
+; X86-NEXT: subl %ecx, %eax
; X86-NEXT: # kill: %AX<def> %AX<kill> %EAX<kill>
; X86-NEXT: retl
;
; X64-LABEL: test_mul_by_26:
; X64: # BB#0:
-; X64-NEXT: imull $26, %edi, %eax
-; X64-NEXT: # kill: %AX<def> %AX<kill> %EAX<kill>
+; X64-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-NEXT: leal (%rdi,%rdi,8), %eax
+; X64-NEXT: leal (%rax,%rax,2), %eax
+; X64-NEXT: subl %eax, %edi
+; X64-NEXT: movl %edi, %eax
; X64-NEXT: retq
%mul = mul nsw i16 %x, 26
ret i16 %mul
@@ -502,14 +536,19 @@ define i16 @test_mul_by_27(i16 %x) {
define i16 @test_mul_by_28(i16 %x) {
; X86-LABEL: test_mul_by_28:
; X86: # BB#0:
-; X86-NEXT: movzwl {{[0-9]+}}(%esp), %eax
-; X86-NEXT: imull $28, %eax, %eax
+; X86-NEXT: movzwl {{[0-9]+}}(%esp), %ecx
+; X86-NEXT: leal (%ecx,%ecx,8), %eax
+; X86-NEXT: leal (%eax,%eax,2), %eax
+; X86-NEXT: addl %ecx, %eax
; X86-NEXT: # kill: %AX<def> %AX<kill> %EAX<kill>
; X86-NEXT: retl
;
; X64-LABEL: test_mul_by_28:
; X64: # BB#0:
-; X64-NEXT: imull $28, %edi, %eax
+; X64-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-NEXT: leal (%rdi,%rdi,8), %eax
+; X64-NEXT: leal (%rax,%rax,2), %eax
+; X64-NEXT: addl %edi, %eax
; X64-NEXT: # kill: %AX<def> %AX<kill> %EAX<kill>
; X64-NEXT: retq
%mul = mul nsw i16 %x, 28
@@ -519,14 +558,21 @@ define i16 @test_mul_by_28(i16 %x) {
define i16 @test_mul_by_29(i16 %x) {
; X86-LABEL: test_mul_by_29:
; X86: # BB#0:
-; X86-NEXT: movzwl {{[0-9]+}}(%esp), %eax
-; X86-NEXT: imull $29, %eax, %eax
+; X86-NEXT: movzwl {{[0-9]+}}(%esp), %ecx
+; X86-NEXT: leal (%ecx,%ecx,8), %eax
+; X86-NEXT: leal (%eax,%eax,2), %eax
+; X86-NEXT: addl %ecx, %eax
+; X86-NEXT: addl %ecx, %eax
; X86-NEXT: # kill: %AX<def> %AX<kill> %EAX<kill>
; X86-NEXT: retl
;
; X64-LABEL: test_mul_by_29:
; X64: # BB#0:
-; X64-NEXT: imull $29, %edi, %eax
+; X64-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-NEXT: leal (%rdi,%rdi,8), %eax
+; X64-NEXT: leal (%rax,%rax,2), %eax
+; X64-NEXT: addl %edi, %eax
+; X64-NEXT: addl %edi, %eax
; X64-NEXT: # kill: %AX<def> %AX<kill> %EAX<kill>
; X64-NEXT: retq
%mul = mul nsw i16 %x, 29
@@ -537,14 +583,22 @@ define i16 @test_mul_by_30(i16 %x) {
; X86-LABEL: test_mul_by_30:
; X86: # BB#0:
; X86-NEXT: movzwl {{[0-9]+}}(%esp), %eax
-; X86-NEXT: imull $30, %eax, %eax
+; X86-NEXT: movl %eax, %ecx
+; X86-NEXT: shll $5, %ecx
+; X86-NEXT: movl %eax, %edx
+; X86-NEXT: subl %ecx, %edx
+; X86-NEXT: subl %edx, %eax
; X86-NEXT: # kill: %AX<def> %AX<kill> %EAX<kill>
; X86-NEXT: retl
;
; X64-LABEL: test_mul_by_30:
; X64: # BB#0:
-; X64-NEXT: imull $30, %edi, %eax
-; X64-NEXT: # kill: %AX<def> %AX<kill> %EAX<kill>
+; X64-NEXT: movl %edi, %eax
+; X64-NEXT: shll $5, %eax
+; X64-NEXT: movl %edi, %ecx
+; X64-NEXT: subl %eax, %ecx
+; X64-NEXT: subl %ecx, %edi
+; X64-NEXT: movl %edi, %eax
; X64-NEXT: retq
%mul = mul nsw i16 %x, 30
ret i16 %mul
@@ -587,3 +641,30 @@ define i16 @test_mul_by_32(i16 %x) {
%mul = mul nsw i16 %x, 32
ret i16 %mul
}
+
+; (x*9+42)*(x*5+2)
+define i16 @test_mul_spec(i16 %x) nounwind {
+; X86-LABEL: test_mul_spec:
+; X86: # BB#0:
+; X86-NEXT: movzwl {{[0-9]+}}(%esp), %eax
+; X86-NEXT: leal 42(%eax,%eax,8), %ecx
+; X86-NEXT: leal 2(%eax,%eax,4), %eax
+; X86-NEXT: imull %ecx, %eax
+; X86-NEXT: # kill: %AX<def> %AX<kill> %EAX<kill>
+; X86-NEXT: retl
+;
+; X64-LABEL: test_mul_spec:
+; X64: # BB#0:
+; X64-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-NEXT: leal 42(%rdi,%rdi,8), %ecx
+; X64-NEXT: leal 2(%rdi,%rdi,4), %eax
+; X64-NEXT: imull %ecx, %eax
+; X64-NEXT: # kill: %AX<def> %AX<kill> %EAX<kill>
+; X64-NEXT: retq
+ %mul = mul nsw i16 %x, 9
+ %add = add nsw i16 %mul, 42
+ %mul2 = mul nsw i16 %x, 5
+ %add2 = add nsw i16 %mul2, 2
+ %mul3 = mul nsw i16 %add, %add2
+ ret i16 %mul3
+}
diff --git a/test/CodeGen/X86/mul-constant-i32.ll b/test/CodeGen/X86/mul-constant-i32.ll
index 76e46e1f1b09..b1e9a929b7f2 100644
--- a/test/CodeGen/X86/mul-constant-i32.ll
+++ b/test/CodeGen/X86/mul-constant-i32.ll
@@ -1,6 +1,12 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=i686-unknown | FileCheck %s --check-prefix=X86
-; RUN: llc < %s -mtriple=x86_64-unknown | FileCheck %s --check-prefix=X64
+; RUN: llc < %s -mtriple=x86_64-unknown -print-schedule=true -mcpu=haswell| FileCheck %s --check-prefix=X64-HSW
+; RUN: llc < %s -mtriple=x86_64-unknown -print-schedule=true -mcpu=btver2| FileCheck %s --check-prefix=X64-JAG
+; RUN: llc < %s -mtriple=i686-unknown -mul-constant-optimization=false | FileCheck %s --check-prefix=X86-NOOPT
+; RUN: llc < %s -mtriple=x86_64-unknown -mul-constant-optimization=false -print-schedule=true -mcpu=haswell| FileCheck %s --check-prefix=HSW-NOOPT
+; RUN: llc < %s -mtriple=x86_64-unknown -mul-constant-optimization=false -print-schedule=true -mcpu=btver2| FileCheck %s --check-prefix=JAG-NOOPT
+; RUN: llc < %s -mtriple=x86_64-unknown -print-schedule=true -mcpu=slm| FileCheck %s --check-prefix=X64-SLM
+; RUN: llc < %s -mtriple=x86_64-unknown -mul-constant-optimization=false -print-schedule=true -mcpu=slm| FileCheck %s --check-prefix=SLM-NOOPT
define i32 @test_mul_by_1(i32 %x) {
; X86-LABEL: test_mul_by_1:
@@ -8,10 +14,40 @@ define i32 @test_mul_by_1(i32 %x) {
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_1:
-; X64: # BB#0:
-; X64-NEXT: movl %edi, %eax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_1:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: movl %edi, %eax # sched: [1:0.25]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_1:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: movl %edi, %eax # sched: [1:0.17]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_1:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_1:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: movl %edi, %eax # sched: [1:0.25]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_1:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: movl %edi, %eax # sched: [1:0.17]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_1:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: movl %edi, %eax # sched: [1:0.50]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_1:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: movl %edi, %eax # sched: [1:0.50]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i32 %x, 1
ret i32 %mul
}
@@ -23,11 +59,47 @@ define i32 @test_mul_by_2(i32 %x) {
; X86-NEXT: addl %eax, %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_2:
-; X64: # BB#0:
-; X64-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
-; X64-NEXT: leal (%rdi,%rdi), %eax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_2:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-HSW-NEXT: leal (%rdi,%rdi), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_2:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-JAG-NEXT: leal (%rdi,%rdi), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_2:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: addl %eax, %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_2:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; HSW-NOOPT-NEXT: leal (%rdi,%rdi), %eax # sched: [1:0.50]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_2:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; JAG-NOOPT-NEXT: leal (%rdi,%rdi), %eax # sched: [1:0.50]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_2:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-SLM-NEXT: leal (%rdi,%rdi), %eax # sched: [1:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_2:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; SLM-NOOPT-NEXT: leal (%rdi,%rdi), %eax # sched: [1:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i32 %x, 2
ret i32 %mul
}
@@ -38,11 +110,46 @@ define i32 @test_mul_by_3(i32 %x) {
; X86-NEXT: imull $3, {{[0-9]+}}(%esp), %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_3:
-; X64: # BB#0:
-; X64-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
-; X64-NEXT: leal (%rdi,%rdi,2), %eax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_3:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-HSW-NEXT: leal (%rdi,%rdi,2), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_3:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-JAG-NEXT: leal (%rdi,%rdi,2), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_3:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: imull $3, {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_3:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; HSW-NOOPT-NEXT: leal (%rdi,%rdi,2), %eax # sched: [1:0.50]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_3:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; JAG-NOOPT-NEXT: leal (%rdi,%rdi,2), %eax # sched: [1:0.50]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_3:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-SLM-NEXT: leal (%rdi,%rdi,2), %eax # sched: [1:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_3:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; SLM-NOOPT-NEXT: leal (%rdi,%rdi,2), %eax # sched: [1:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i32 %x, 3
ret i32 %mul
}
@@ -54,11 +161,47 @@ define i32 @test_mul_by_4(i32 %x) {
; X86-NEXT: shll $2, %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_4:
-; X64: # BB#0:
-; X64-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
-; X64-NEXT: leal (,%rdi,4), %eax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_4:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-HSW-NEXT: leal (,%rdi,4), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_4:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-JAG-NEXT: leal (,%rdi,4), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_4:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: shll $2, %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_4:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; HSW-NOOPT-NEXT: leal (,%rdi,4), %eax # sched: [1:0.50]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_4:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; JAG-NOOPT-NEXT: leal (,%rdi,4), %eax # sched: [1:0.50]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_4:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-SLM-NEXT: leal (,%rdi,4), %eax # sched: [1:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_4:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; SLM-NOOPT-NEXT: leal (,%rdi,4), %eax # sched: [1:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i32 %x, 4
ret i32 %mul
}
@@ -69,11 +212,46 @@ define i32 @test_mul_by_5(i32 %x) {
; X86-NEXT: imull $5, {{[0-9]+}}(%esp), %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_5:
-; X64: # BB#0:
-; X64-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
-; X64-NEXT: leal (%rdi,%rdi,4), %eax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_5:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-HSW-NEXT: leal (%rdi,%rdi,4), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_5:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-JAG-NEXT: leal (%rdi,%rdi,4), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_5:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: imull $5, {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_5:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; HSW-NOOPT-NEXT: leal (%rdi,%rdi,4), %eax # sched: [1:0.50]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_5:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; JAG-NOOPT-NEXT: leal (%rdi,%rdi,4), %eax # sched: [1:0.50]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_5:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-SLM-NEXT: leal (%rdi,%rdi,4), %eax # sched: [1:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_5:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; SLM-NOOPT-NEXT: leal (%rdi,%rdi,4), %eax # sched: [1:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i32 %x, 5
ret i32 %mul
}
@@ -86,12 +264,46 @@ define i32 @test_mul_by_6(i32 %x) {
; X86-NEXT: leal (%eax,%eax,2), %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_6:
-; X64: # BB#0:
-; X64-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
-; X64-NEXT: addl %edi, %edi
-; X64-NEXT: leal (%rdi,%rdi,2), %eax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_6:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-HSW-NEXT: addl %edi, %edi # sched: [1:0.25]
+; X64-HSW-NEXT: leal (%rdi,%rdi,2), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_6:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-JAG-NEXT: addl %edi, %edi # sched: [1:0.50]
+; X64-JAG-NEXT: leal (%rdi,%rdi,2), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_6:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: imull $6, {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_6:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imull $6, %edi, %eax # sched: [4:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_6:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imull $6, %edi, %eax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_6:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-SLM-NEXT: addl %edi, %edi # sched: [1:0.50]
+; X64-SLM-NEXT: leal (%rdi,%rdi,2), %eax # sched: [1:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_6:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imull $6, %edi, %eax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i32 %x, 6
ret i32 %mul
}
@@ -104,12 +316,46 @@ define i32 @test_mul_by_7(i32 %x) {
; X86-NEXT: subl %ecx, %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_7:
-; X64: # BB#0:
-; X64-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
-; X64-NEXT: leal (,%rdi,8), %eax
-; X64-NEXT: subl %edi, %eax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_7:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-HSW-NEXT: leal (,%rdi,8), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: subl %edi, %eax # sched: [1:0.25]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_7:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-JAG-NEXT: leal (,%rdi,8), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: subl %edi, %eax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_7:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: imull $7, {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_7:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imull $7, %edi, %eax # sched: [4:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_7:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imull $7, %edi, %eax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_7:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-SLM-NEXT: leal (,%rdi,8), %eax # sched: [1:1.00]
+; X64-SLM-NEXT: subl %edi, %eax # sched: [1:0.50]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_7:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imull $7, %edi, %eax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i32 %x, 7
ret i32 %mul
}
@@ -121,11 +367,47 @@ define i32 @test_mul_by_8(i32 %x) {
; X86-NEXT: shll $3, %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_8:
-; X64: # BB#0:
-; X64-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
-; X64-NEXT: leal (,%rdi,8), %eax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_8:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-HSW-NEXT: leal (,%rdi,8), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_8:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-JAG-NEXT: leal (,%rdi,8), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_8:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: shll $3, %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_8:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; HSW-NOOPT-NEXT: leal (,%rdi,8), %eax # sched: [1:0.50]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_8:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; JAG-NOOPT-NEXT: leal (,%rdi,8), %eax # sched: [1:0.50]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_8:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-SLM-NEXT: leal (,%rdi,8), %eax # sched: [1:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_8:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; SLM-NOOPT-NEXT: leal (,%rdi,8), %eax # sched: [1:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i32 %x, 8
ret i32 %mul
}
@@ -136,11 +418,46 @@ define i32 @test_mul_by_9(i32 %x) {
; X86-NEXT: imull $9, {{[0-9]+}}(%esp), %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_9:
-; X64: # BB#0:
-; X64-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
-; X64-NEXT: leal (%rdi,%rdi,8), %eax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_9:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-HSW-NEXT: leal (%rdi,%rdi,8), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_9:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-JAG-NEXT: leal (%rdi,%rdi,8), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_9:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: imull $9, {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_9:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; HSW-NOOPT-NEXT: leal (%rdi,%rdi,8), %eax # sched: [1:0.50]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_9:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; JAG-NOOPT-NEXT: leal (%rdi,%rdi,8), %eax # sched: [1:0.50]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_9:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-SLM-NEXT: leal (%rdi,%rdi,8), %eax # sched: [1:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_9:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; SLM-NOOPT-NEXT: leal (%rdi,%rdi,8), %eax # sched: [1:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i32 %x, 9
ret i32 %mul
}
@@ -153,12 +470,46 @@ define i32 @test_mul_by_10(i32 %x) {
; X86-NEXT: leal (%eax,%eax,4), %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_10:
-; X64: # BB#0:
-; X64-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
-; X64-NEXT: addl %edi, %edi
-; X64-NEXT: leal (%rdi,%rdi,4), %eax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_10:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-HSW-NEXT: addl %edi, %edi # sched: [1:0.25]
+; X64-HSW-NEXT: leal (%rdi,%rdi,4), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_10:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-JAG-NEXT: addl %edi, %edi # sched: [1:0.50]
+; X64-JAG-NEXT: leal (%rdi,%rdi,4), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_10:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: imull $10, {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_10:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imull $10, %edi, %eax # sched: [4:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_10:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imull $10, %edi, %eax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_10:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-SLM-NEXT: addl %edi, %edi # sched: [1:0.50]
+; X64-SLM-NEXT: leal (%rdi,%rdi,4), %eax # sched: [1:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_10:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imull $10, %edi, %eax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i32 %x, 10
ret i32 %mul
}
@@ -166,13 +517,49 @@ define i32 @test_mul_by_10(i32 %x) {
define i32 @test_mul_by_11(i32 %x) {
; X86-LABEL: test_mul_by_11:
; X86: # BB#0:
-; X86-NEXT: imull $11, {{[0-9]+}}(%esp), %eax
+; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X86-NEXT: leal (%eax,%eax,4), %ecx
+; X86-NEXT: leal (%eax,%ecx,2), %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_11:
-; X64: # BB#0:
-; X64-NEXT: imull $11, %edi, %eax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_11:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-HSW-NEXT: leal (%rdi,%rdi,4), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: leal (%rdi,%rax,2), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_11:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-JAG-NEXT: leal (%rdi,%rdi,4), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: leal (%rdi,%rax,2), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_11:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: imull $11, {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_11:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imull $11, %edi, %eax # sched: [4:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_11:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imull $11, %edi, %eax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_11:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: imull $11, %edi, %eax # sched: [3:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_11:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imull $11, %edi, %eax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i32 %x, 11
ret i32 %mul
}
@@ -185,12 +572,46 @@ define i32 @test_mul_by_12(i32 %x) {
; X86-NEXT: leal (%eax,%eax,2), %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_12:
-; X64: # BB#0:
-; X64-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
-; X64-NEXT: shll $2, %edi
-; X64-NEXT: leal (%rdi,%rdi,2), %eax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_12:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-HSW-NEXT: shll $2, %edi # sched: [1:0.50]
+; X64-HSW-NEXT: leal (%rdi,%rdi,2), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_12:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-JAG-NEXT: shll $2, %edi # sched: [1:0.50]
+; X64-JAG-NEXT: leal (%rdi,%rdi,2), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_12:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: imull $12, {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_12:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imull $12, %edi, %eax # sched: [4:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_12:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imull $12, %edi, %eax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_12:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-SLM-NEXT: shll $2, %edi # sched: [1:1.00]
+; X64-SLM-NEXT: leal (%rdi,%rdi,2), %eax # sched: [1:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_12:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imull $12, %edi, %eax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i32 %x, 12
ret i32 %mul
}
@@ -198,13 +619,49 @@ define i32 @test_mul_by_12(i32 %x) {
define i32 @test_mul_by_13(i32 %x) {
; X86-LABEL: test_mul_by_13:
; X86: # BB#0:
-; X86-NEXT: imull $13, {{[0-9]+}}(%esp), %eax
+; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X86-NEXT: leal (%eax,%eax,2), %ecx
+; X86-NEXT: leal (%eax,%ecx,4), %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_13:
-; X64: # BB#0:
-; X64-NEXT: imull $13, %edi, %eax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_13:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-HSW-NEXT: leal (%rdi,%rdi,2), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: leal (%rdi,%rax,4), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_13:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-JAG-NEXT: leal (%rdi,%rdi,2), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: leal (%rdi,%rax,4), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_13:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: imull $13, {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_13:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imull $13, %edi, %eax # sched: [4:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_13:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imull $13, %edi, %eax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_13:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: imull $13, %edi, %eax # sched: [3:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_13:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imull $13, %edi, %eax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i32 %x, 13
ret i32 %mul
}
@@ -212,13 +669,52 @@ define i32 @test_mul_by_13(i32 %x) {
define i32 @test_mul_by_14(i32 %x) {
; X86-LABEL: test_mul_by_14:
; X86: # BB#0:
-; X86-NEXT: imull $14, {{[0-9]+}}(%esp), %eax
+; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx
+; X86-NEXT: leal (%ecx,%ecx,2), %eax
+; X86-NEXT: leal (%ecx,%eax,4), %eax
+; X86-NEXT: addl %ecx, %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_14:
-; X64: # BB#0:
-; X64-NEXT: imull $14, %edi, %eax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_14:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-HSW-NEXT: leal (%rdi,%rdi,2), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: leal (%rdi,%rax,4), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: addl %edi, %eax # sched: [1:0.25]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_14:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-JAG-NEXT: leal (%rdi,%rdi,2), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: leal (%rdi,%rax,4), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: addl %edi, %eax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_14:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: imull $14, {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_14:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imull $14, %edi, %eax # sched: [4:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_14:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imull $14, %edi, %eax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_14:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: imull $14, %edi, %eax # sched: [3:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_14:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imull $14, %edi, %eax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i32 %x, 14
ret i32 %mul
}
@@ -231,12 +727,46 @@ define i32 @test_mul_by_15(i32 %x) {
; X86-NEXT: leal (%eax,%eax,2), %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_15:
-; X64: # BB#0:
-; X64-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
-; X64-NEXT: leal (%rdi,%rdi,4), %eax
-; X64-NEXT: leal (%rax,%rax,2), %eax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_15:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-HSW-NEXT: leal (%rdi,%rdi,4), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: leal (%rax,%rax,2), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_15:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-JAG-NEXT: leal (%rdi,%rdi,4), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: leal (%rax,%rax,2), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_15:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: imull $15, {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_15:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imull $15, %edi, %eax # sched: [4:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_15:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imull $15, %edi, %eax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_15:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-SLM-NEXT: leal (%rdi,%rdi,4), %eax # sched: [1:1.00]
+; X64-SLM-NEXT: leal (%rax,%rax,2), %eax # sched: [1:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_15:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imull $15, %edi, %eax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i32 %x, 15
ret i32 %mul
}
@@ -248,11 +778,47 @@ define i32 @test_mul_by_16(i32 %x) {
; X86-NEXT: shll $4, %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_16:
-; X64: # BB#0:
-; X64-NEXT: shll $4, %edi
-; X64-NEXT: movl %edi, %eax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_16:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: shll $4, %edi # sched: [1:0.50]
+; X64-HSW-NEXT: movl %edi, %eax # sched: [1:0.25]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_16:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: shll $4, %edi # sched: [1:0.50]
+; X64-JAG-NEXT: movl %edi, %eax # sched: [1:0.17]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_16:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: shll $4, %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_16:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: shll $4, %edi # sched: [1:0.50]
+; HSW-NOOPT-NEXT: movl %edi, %eax # sched: [1:0.25]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_16:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: shll $4, %edi # sched: [1:0.50]
+; JAG-NOOPT-NEXT: movl %edi, %eax # sched: [1:0.17]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_16:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: shll $4, %edi # sched: [1:1.00]
+; X64-SLM-NEXT: movl %edi, %eax # sched: [1:0.50]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_16:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: shll $4, %edi # sched: [1:1.00]
+; SLM-NOOPT-NEXT: movl %edi, %eax # sched: [1:0.50]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i32 %x, 16
ret i32 %mul
}
@@ -266,13 +832,49 @@ define i32 @test_mul_by_17(i32 %x) {
; X86-NEXT: addl %ecx, %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_17:
-; X64: # BB#0:
-; X64-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
-; X64-NEXT: movl %edi, %eax
-; X64-NEXT: shll $4, %eax
-; X64-NEXT: leal (%rax,%rdi), %eax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_17:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-HSW-NEXT: movl %edi, %eax # sched: [1:0.25]
+; X64-HSW-NEXT: shll $4, %eax # sched: [1:0.50]
+; X64-HSW-NEXT: leal (%rax,%rdi), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_17:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-JAG-NEXT: movl %edi, %eax # sched: [1:0.17]
+; X64-JAG-NEXT: shll $4, %eax # sched: [1:0.50]
+; X64-JAG-NEXT: leal (%rax,%rdi), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_17:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: imull $17, {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_17:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imull $17, %edi, %eax # sched: [4:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_17:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imull $17, %edi, %eax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_17:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-SLM-NEXT: movl %edi, %eax # sched: [1:0.50]
+; X64-SLM-NEXT: shll $4, %eax # sched: [1:1.00]
+; X64-SLM-NEXT: leal (%rax,%rdi), %eax # sched: [1:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_17:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imull $17, %edi, %eax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i32 %x, 17
ret i32 %mul
}
@@ -285,12 +887,46 @@ define i32 @test_mul_by_18(i32 %x) {
; X86-NEXT: leal (%eax,%eax,8), %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_18:
-; X64: # BB#0:
-; X64-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
-; X64-NEXT: addl %edi, %edi
-; X64-NEXT: leal (%rdi,%rdi,8), %eax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_18:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-HSW-NEXT: addl %edi, %edi # sched: [1:0.25]
+; X64-HSW-NEXT: leal (%rdi,%rdi,8), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_18:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-JAG-NEXT: addl %edi, %edi # sched: [1:0.50]
+; X64-JAG-NEXT: leal (%rdi,%rdi,8), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_18:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: imull $18, {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_18:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imull $18, %edi, %eax # sched: [4:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_18:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imull $18, %edi, %eax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_18:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-SLM-NEXT: addl %edi, %edi # sched: [1:0.50]
+; X64-SLM-NEXT: leal (%rdi,%rdi,8), %eax # sched: [1:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_18:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imull $18, %edi, %eax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i32 %x, 18
ret i32 %mul
}
@@ -298,13 +934,54 @@ define i32 @test_mul_by_18(i32 %x) {
define i32 @test_mul_by_19(i32 %x) {
; X86-LABEL: test_mul_by_19:
; X86: # BB#0:
-; X86-NEXT: imull $19, {{[0-9]+}}(%esp), %eax
+; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X86-NEXT: leal (%eax,%eax,4), %ecx
+; X86-NEXT: shll $2, %ecx
+; X86-NEXT: subl %ecx, %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_19:
-; X64: # BB#0:
-; X64-NEXT: imull $19, %edi, %eax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_19:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-HSW-NEXT: leal (%rdi,%rdi,4), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: shll $2, %eax # sched: [1:0.50]
+; X64-HSW-NEXT: subl %eax, %edi # sched: [1:0.25]
+; X64-HSW-NEXT: movl %edi, %eax # sched: [1:0.25]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_19:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-JAG-NEXT: leal (%rdi,%rdi,4), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: shll $2, %eax # sched: [1:0.50]
+; X64-JAG-NEXT: subl %eax, %edi # sched: [1:0.50]
+; X64-JAG-NEXT: movl %edi, %eax # sched: [1:0.17]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_19:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: imull $19, {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_19:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imull $19, %edi, %eax # sched: [4:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_19:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imull $19, %edi, %eax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_19:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: imull $19, %edi, %eax # sched: [3:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_19:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imull $19, %edi, %eax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i32 %x, 19
ret i32 %mul
}
@@ -317,12 +994,46 @@ define i32 @test_mul_by_20(i32 %x) {
; X86-NEXT: leal (%eax,%eax,4), %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_20:
-; X64: # BB#0:
-; X64-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
-; X64-NEXT: shll $2, %edi
-; X64-NEXT: leal (%rdi,%rdi,4), %eax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_20:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-HSW-NEXT: shll $2, %edi # sched: [1:0.50]
+; X64-HSW-NEXT: leal (%rdi,%rdi,4), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_20:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-JAG-NEXT: shll $2, %edi # sched: [1:0.50]
+; X64-JAG-NEXT: leal (%rdi,%rdi,4), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_20:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: imull $20, {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_20:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imull $20, %edi, %eax # sched: [4:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_20:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imull $20, %edi, %eax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_20:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-SLM-NEXT: shll $2, %edi # sched: [1:1.00]
+; X64-SLM-NEXT: leal (%rdi,%rdi,4), %eax # sched: [1:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_20:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imull $20, %edi, %eax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i32 %x, 20
ret i32 %mul
}
@@ -330,13 +1041,49 @@ define i32 @test_mul_by_20(i32 %x) {
define i32 @test_mul_by_21(i32 %x) {
; X86-LABEL: test_mul_by_21:
; X86: # BB#0:
-; X86-NEXT: imull $21, {{[0-9]+}}(%esp), %eax
+; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X86-NEXT: leal (%eax,%eax,4), %ecx
+; X86-NEXT: leal (%eax,%ecx,4), %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_21:
-; X64: # BB#0:
-; X64-NEXT: imull $21, %edi, %eax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_21:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-HSW-NEXT: leal (%rdi,%rdi,4), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: leal (%rdi,%rax,4), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_21:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-JAG-NEXT: leal (%rdi,%rdi,4), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: leal (%rdi,%rax,4), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_21:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: imull $21, {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_21:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imull $21, %edi, %eax # sched: [4:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_21:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imull $21, %edi, %eax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_21:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: imull $21, %edi, %eax # sched: [3:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_21:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imull $21, %edi, %eax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i32 %x, 21
ret i32 %mul
}
@@ -344,13 +1091,52 @@ define i32 @test_mul_by_21(i32 %x) {
define i32 @test_mul_by_22(i32 %x) {
; X86-LABEL: test_mul_by_22:
; X86: # BB#0:
-; X86-NEXT: imull $22, {{[0-9]+}}(%esp), %eax
+; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx
+; X86-NEXT: leal (%ecx,%ecx,4), %eax
+; X86-NEXT: leal (%ecx,%eax,4), %eax
+; X86-NEXT: addl %ecx, %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_22:
-; X64: # BB#0:
-; X64-NEXT: imull $22, %edi, %eax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_22:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-HSW-NEXT: leal (%rdi,%rdi,4), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: leal (%rdi,%rax,4), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: addl %edi, %eax # sched: [1:0.25]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_22:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-JAG-NEXT: leal (%rdi,%rdi,4), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: leal (%rdi,%rax,4), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: addl %edi, %eax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_22:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: imull $22, {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_22:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imull $22, %edi, %eax # sched: [4:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_22:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imull $22, %edi, %eax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_22:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: imull $22, %edi, %eax # sched: [3:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_22:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imull $22, %edi, %eax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i32 %x, 22
ret i32 %mul
}
@@ -358,13 +1144,54 @@ define i32 @test_mul_by_22(i32 %x) {
define i32 @test_mul_by_23(i32 %x) {
; X86-LABEL: test_mul_by_23:
; X86: # BB#0:
-; X86-NEXT: imull $23, {{[0-9]+}}(%esp), %eax
+; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X86-NEXT: leal (%eax,%eax,2), %ecx
+; X86-NEXT: shll $3, %ecx
+; X86-NEXT: subl %ecx, %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_23:
-; X64: # BB#0:
-; X64-NEXT: imull $23, %edi, %eax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_23:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-HSW-NEXT: leal (%rdi,%rdi,2), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: shll $3, %eax # sched: [1:0.50]
+; X64-HSW-NEXT: subl %eax, %edi # sched: [1:0.25]
+; X64-HSW-NEXT: movl %edi, %eax # sched: [1:0.25]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_23:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-JAG-NEXT: leal (%rdi,%rdi,2), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: shll $3, %eax # sched: [1:0.50]
+; X64-JAG-NEXT: subl %eax, %edi # sched: [1:0.50]
+; X64-JAG-NEXT: movl %edi, %eax # sched: [1:0.17]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_23:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: imull $23, {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_23:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imull $23, %edi, %eax # sched: [4:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_23:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imull $23, %edi, %eax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_23:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: imull $23, %edi, %eax # sched: [3:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_23:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imull $23, %edi, %eax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i32 %x, 23
ret i32 %mul
}
@@ -377,12 +1204,46 @@ define i32 @test_mul_by_24(i32 %x) {
; X86-NEXT: leal (%eax,%eax,2), %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_24:
-; X64: # BB#0:
-; X64-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
-; X64-NEXT: shll $3, %edi
-; X64-NEXT: leal (%rdi,%rdi,2), %eax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_24:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-HSW-NEXT: shll $3, %edi # sched: [1:0.50]
+; X64-HSW-NEXT: leal (%rdi,%rdi,2), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_24:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-JAG-NEXT: shll $3, %edi # sched: [1:0.50]
+; X64-JAG-NEXT: leal (%rdi,%rdi,2), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_24:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: imull $24, {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_24:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imull $24, %edi, %eax # sched: [4:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_24:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imull $24, %edi, %eax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_24:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-SLM-NEXT: shll $3, %edi # sched: [1:1.00]
+; X64-SLM-NEXT: leal (%rdi,%rdi,2), %eax # sched: [1:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_24:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imull $24, %edi, %eax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i32 %x, 24
ret i32 %mul
}
@@ -395,12 +1256,46 @@ define i32 @test_mul_by_25(i32 %x) {
; X86-NEXT: leal (%eax,%eax,4), %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_25:
-; X64: # BB#0:
-; X64-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
-; X64-NEXT: leal (%rdi,%rdi,4), %eax
-; X64-NEXT: leal (%rax,%rax,4), %eax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_25:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-HSW-NEXT: leal (%rdi,%rdi,4), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: leal (%rax,%rax,4), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_25:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-JAG-NEXT: leal (%rdi,%rdi,4), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: leal (%rax,%rax,4), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_25:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: imull $25, {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_25:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imull $25, %edi, %eax # sched: [4:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_25:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imull $25, %edi, %eax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_25:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-SLM-NEXT: leal (%rdi,%rdi,4), %eax # sched: [1:1.00]
+; X64-SLM-NEXT: leal (%rax,%rax,4), %eax # sched: [1:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_25:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imull $25, %edi, %eax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i32 %x, 25
ret i32 %mul
}
@@ -408,13 +1303,54 @@ define i32 @test_mul_by_25(i32 %x) {
define i32 @test_mul_by_26(i32 %x) {
; X86-LABEL: test_mul_by_26:
; X86: # BB#0:
-; X86-NEXT: imull $26, {{[0-9]+}}(%esp), %eax
+; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X86-NEXT: leal (%eax,%eax,8), %ecx
+; X86-NEXT: leal (%ecx,%ecx,2), %ecx
+; X86-NEXT: subl %ecx, %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_26:
-; X64: # BB#0:
-; X64-NEXT: imull $26, %edi, %eax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_26:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-HSW-NEXT: leal (%rdi,%rdi,8), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: leal (%rax,%rax,2), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: subl %eax, %edi # sched: [1:0.25]
+; X64-HSW-NEXT: movl %edi, %eax # sched: [1:0.25]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_26:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-JAG-NEXT: leal (%rdi,%rdi,8), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: leal (%rax,%rax,2), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: subl %eax, %edi # sched: [1:0.50]
+; X64-JAG-NEXT: movl %edi, %eax # sched: [1:0.17]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_26:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: imull $26, {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_26:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imull $26, %edi, %eax # sched: [4:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_26:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imull $26, %edi, %eax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_26:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: imull $26, %edi, %eax # sched: [3:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_26:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imull $26, %edi, %eax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i32 %x, 26
ret i32 %mul
}
@@ -427,12 +1363,46 @@ define i32 @test_mul_by_27(i32 %x) {
; X86-NEXT: leal (%eax,%eax,2), %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_27:
-; X64: # BB#0:
-; X64-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
-; X64-NEXT: leal (%rdi,%rdi,8), %eax
-; X64-NEXT: leal (%rax,%rax,2), %eax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_27:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-HSW-NEXT: leal (%rdi,%rdi,8), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: leal (%rax,%rax,2), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_27:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-JAG-NEXT: leal (%rdi,%rdi,8), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: leal (%rax,%rax,2), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_27:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: imull $27, {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_27:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imull $27, %edi, %eax # sched: [4:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_27:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imull $27, %edi, %eax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_27:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-SLM-NEXT: leal (%rdi,%rdi,8), %eax # sched: [1:1.00]
+; X64-SLM-NEXT: leal (%rax,%rax,2), %eax # sched: [1:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_27:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imull $27, %edi, %eax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i32 %x, 27
ret i32 %mul
}
@@ -440,13 +1410,52 @@ define i32 @test_mul_by_27(i32 %x) {
define i32 @test_mul_by_28(i32 %x) {
; X86-LABEL: test_mul_by_28:
; X86: # BB#0:
-; X86-NEXT: imull $28, {{[0-9]+}}(%esp), %eax
+; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx
+; X86-NEXT: leal (%ecx,%ecx,8), %eax
+; X86-NEXT: leal (%eax,%eax,2), %eax
+; X86-NEXT: addl %ecx, %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_28:
-; X64: # BB#0:
-; X64-NEXT: imull $28, %edi, %eax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_28:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-HSW-NEXT: leal (%rdi,%rdi,8), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: leal (%rax,%rax,2), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: addl %edi, %eax # sched: [1:0.25]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_28:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-JAG-NEXT: leal (%rdi,%rdi,8), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: leal (%rax,%rax,2), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: addl %edi, %eax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_28:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: imull $28, {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_28:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imull $28, %edi, %eax # sched: [4:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_28:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imull $28, %edi, %eax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_28:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: imull $28, %edi, %eax # sched: [3:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_28:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imull $28, %edi, %eax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i32 %x, 28
ret i32 %mul
}
@@ -454,13 +1463,55 @@ define i32 @test_mul_by_28(i32 %x) {
define i32 @test_mul_by_29(i32 %x) {
; X86-LABEL: test_mul_by_29:
; X86: # BB#0:
-; X86-NEXT: imull $29, {{[0-9]+}}(%esp), %eax
+; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx
+; X86-NEXT: leal (%ecx,%ecx,8), %eax
+; X86-NEXT: leal (%eax,%eax,2), %eax
+; X86-NEXT: addl %ecx, %eax
+; X86-NEXT: addl %ecx, %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_29:
-; X64: # BB#0:
-; X64-NEXT: imull $29, %edi, %eax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_29:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-HSW-NEXT: leal (%rdi,%rdi,8), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: leal (%rax,%rax,2), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: addl %edi, %eax # sched: [1:0.25]
+; X64-HSW-NEXT: addl %edi, %eax # sched: [1:0.25]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_29:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-JAG-NEXT: leal (%rdi,%rdi,8), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: leal (%rax,%rax,2), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: addl %edi, %eax # sched: [1:0.50]
+; X64-JAG-NEXT: addl %edi, %eax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_29:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: imull $29, {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_29:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imull $29, %edi, %eax # sched: [4:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_29:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imull $29, %edi, %eax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_29:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: imull $29, %edi, %eax # sched: [3:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_29:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imull $29, %edi, %eax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i32 %x, 29
ret i32 %mul
}
@@ -468,13 +1519,58 @@ define i32 @test_mul_by_29(i32 %x) {
define i32 @test_mul_by_30(i32 %x) {
; X86-LABEL: test_mul_by_30:
; X86: # BB#0:
-; X86-NEXT: imull $30, {{[0-9]+}}(%esp), %eax
+; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X86-NEXT: movl %eax, %ecx
+; X86-NEXT: shll $5, %ecx
+; X86-NEXT: movl %eax, %edx
+; X86-NEXT: subl %ecx, %edx
+; X86-NEXT: subl %edx, %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_30:
-; X64: # BB#0:
-; X64-NEXT: imull $30, %edi, %eax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_30:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: movl %edi, %eax # sched: [1:0.25]
+; X64-HSW-NEXT: shll $5, %eax # sched: [1:0.50]
+; X64-HSW-NEXT: movl %edi, %ecx # sched: [1:0.25]
+; X64-HSW-NEXT: subl %eax, %ecx # sched: [1:0.25]
+; X64-HSW-NEXT: subl %ecx, %edi # sched: [1:0.25]
+; X64-HSW-NEXT: movl %edi, %eax # sched: [1:0.25]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_30:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: movl %edi, %eax # sched: [1:0.17]
+; X64-JAG-NEXT: movl %edi, %ecx # sched: [1:0.17]
+; X64-JAG-NEXT: shll $5, %eax # sched: [1:0.50]
+; X64-JAG-NEXT: subl %eax, %ecx # sched: [1:0.50]
+; X64-JAG-NEXT: subl %ecx, %edi # sched: [1:0.50]
+; X64-JAG-NEXT: movl %edi, %eax # sched: [1:0.17]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_30:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: imull $30, {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_30:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imull $30, %edi, %eax # sched: [4:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_30:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imull $30, %edi, %eax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_30:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: imull $30, %edi, %eax # sched: [3:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_30:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imull $30, %edi, %eax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i32 %x, 30
ret i32 %mul
}
@@ -488,12 +1584,46 @@ define i32 @test_mul_by_31(i32 %x) {
; X86-NEXT: subl %ecx, %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_31:
-; X64: # BB#0:
-; X64-NEXT: movl %edi, %eax
-; X64-NEXT: shll $5, %eax
-; X64-NEXT: subl %edi, %eax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_31:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: movl %edi, %eax # sched: [1:0.25]
+; X64-HSW-NEXT: shll $5, %eax # sched: [1:0.50]
+; X64-HSW-NEXT: subl %edi, %eax # sched: [1:0.25]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_31:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: movl %edi, %eax # sched: [1:0.17]
+; X64-JAG-NEXT: shll $5, %eax # sched: [1:0.50]
+; X64-JAG-NEXT: subl %edi, %eax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_31:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: imull $31, {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_31:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imull $31, %edi, %eax # sched: [4:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_31:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imull $31, %edi, %eax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_31:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: movl %edi, %eax # sched: [1:0.50]
+; X64-SLM-NEXT: shll $5, %eax # sched: [1:1.00]
+; X64-SLM-NEXT: subl %edi, %eax # sched: [1:0.50]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_31:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imull $31, %edi, %eax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i32 %x, 31
ret i32 %mul
}
@@ -505,11 +1635,124 @@ define i32 @test_mul_by_32(i32 %x) {
; X86-NEXT: shll $5, %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_32:
-; X64: # BB#0:
-; X64-NEXT: shll $5, %edi
-; X64-NEXT: movl %edi, %eax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_32:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: shll $5, %edi # sched: [1:0.50]
+; X64-HSW-NEXT: movl %edi, %eax # sched: [1:0.25]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_32:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: shll $5, %edi # sched: [1:0.50]
+; X64-JAG-NEXT: movl %edi, %eax # sched: [1:0.17]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_32:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: shll $5, %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_32:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: shll $5, %edi # sched: [1:0.50]
+; HSW-NOOPT-NEXT: movl %edi, %eax # sched: [1:0.25]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_32:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: shll $5, %edi # sched: [1:0.50]
+; JAG-NOOPT-NEXT: movl %edi, %eax # sched: [1:0.17]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_32:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: shll $5, %edi # sched: [1:1.00]
+; X64-SLM-NEXT: movl %edi, %eax # sched: [1:0.50]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_32:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: shll $5, %edi # sched: [1:1.00]
+; SLM-NOOPT-NEXT: movl %edi, %eax # sched: [1:0.50]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i32 %x, 32
ret i32 %mul
}
+
+; (x*9+42)*(x*5+2)
+define i32 @test_mul_spec(i32 %x) nounwind {
+; X86-LABEL: test_mul_spec:
+; X86: # BB#0:
+; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X86-NEXT: leal 42(%eax,%eax,8), %ecx
+; X86-NEXT: leal 2(%eax,%eax,4), %eax
+; X86-NEXT: imull %ecx, %eax
+; X86-NEXT: retl
+;
+; X64-HSW-LABEL: test_mul_spec:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-HSW-NEXT: leal (%rdi,%rdi,8), %ecx # sched: [1:0.50]
+; X64-HSW-NEXT: addl $42, %ecx # sched: [1:0.25]
+; X64-HSW-NEXT: leal (%rdi,%rdi,4), %eax # sched: [1:0.50]
+; X64-HSW-NEXT: addl $2, %eax # sched: [1:0.25]
+; X64-HSW-NEXT: imull %ecx, %eax # sched: [4:1.00]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_spec:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-JAG-NEXT: leal 42(%rdi,%rdi,8), %ecx # sched: [1:0.50]
+; X64-JAG-NEXT: leal 2(%rdi,%rdi,4), %eax # sched: [1:0.50]
+; X64-JAG-NEXT: imull %ecx, %eax # sched: [3:1.00]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_spec:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: leal 42(%eax,%eax,8), %ecx
+; X86-NOOPT-NEXT: leal 2(%eax,%eax,4), %eax
+; X86-NOOPT-NEXT: imull %ecx, %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_spec:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; HSW-NOOPT-NEXT: leal (%rdi,%rdi,8), %ecx # sched: [1:0.50]
+; HSW-NOOPT-NEXT: addl $42, %ecx # sched: [1:0.25]
+; HSW-NOOPT-NEXT: leal (%rdi,%rdi,4), %eax # sched: [1:0.50]
+; HSW-NOOPT-NEXT: addl $2, %eax # sched: [1:0.25]
+; HSW-NOOPT-NEXT: imull %ecx, %eax # sched: [4:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_spec:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; JAG-NOOPT-NEXT: leal 42(%rdi,%rdi,8), %ecx # sched: [1:0.50]
+; JAG-NOOPT-NEXT: leal 2(%rdi,%rdi,4), %eax # sched: [1:0.50]
+; JAG-NOOPT-NEXT: imull %ecx, %eax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_spec:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; X64-SLM-NEXT: leal 42(%rdi,%rdi,8), %ecx # sched: [1:1.00]
+; X64-SLM-NEXT: leal 2(%rdi,%rdi,4), %eax # sched: [1:1.00]
+; X64-SLM-NEXT: imull %ecx, %eax # sched: [3:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_spec:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: # kill: %EDI<def> %EDI<kill> %RDI<def>
+; SLM-NOOPT-NEXT: leal 42(%rdi,%rdi,8), %ecx # sched: [1:1.00]
+; SLM-NOOPT-NEXT: leal 2(%rdi,%rdi,4), %eax # sched: [1:1.00]
+; SLM-NOOPT-NEXT: imull %ecx, %eax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
+ %mul = mul nsw i32 %x, 9
+ %add = add nsw i32 %mul, 42
+ %mul2 = mul nsw i32 %x, 5
+ %add2 = add nsw i32 %mul2, 2
+ %mul3 = mul nsw i32 %add, %add2
+ ret i32 %mul3
+}
diff --git a/test/CodeGen/X86/mul-constant-i64.ll b/test/CodeGen/X86/mul-constant-i64.ll
index 8579179a8231..22eb0bdc6c3f 100644
--- a/test/CodeGen/X86/mul-constant-i64.ll
+++ b/test/CodeGen/X86/mul-constant-i64.ll
@@ -1,18 +1,55 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=i686-unknown | FileCheck %s --check-prefix=X86
-; RUN: llc < %s -mtriple=x86_64-unknown | FileCheck %s --check-prefix=X64
+; RUN: llc < %s -mtriple=x86_64-unknown -print-schedule=true -mcpu=haswell| FileCheck %s --check-prefix=X64-HSW
+; RUN: llc < %s -mtriple=x86_64-unknown -print-schedule=true -mcpu=btver2| FileCheck %s --check-prefix=X64-JAG
+; RUN: llc < %s -mtriple=i686-unknown -mul-constant-optimization=false | FileCheck %s --check-prefix=X86-NOOPT
+; RUN: llc < %s -mtriple=x86_64-unknown -mul-constant-optimization=false -print-schedule=true -mcpu=haswell| FileCheck %s --check-prefix=HSW-NOOPT
+; RUN: llc < %s -mtriple=x86_64-unknown -mul-constant-optimization=false -print-schedule=true -mcpu=btver2| FileCheck %s --check-prefix=JAG-NOOPT
+; RUN: llc < %s -mtriple=x86_64-unknown -print-schedule=true -mcpu=slm| FileCheck %s --check-prefix=X64-SLM
+; RUN: llc < %s -mtriple=x86_64-unknown -mul-constant-optimization=false -print-schedule=true -mcpu=slm| FileCheck %s --check-prefix=SLM-NOOPT
-define i64 @test_mul_by_1(i64 %x) {
+define i64 @test_mul_by_1(i64 %x) nounwind {
; X86-LABEL: test_mul_by_1:
; X86: # BB#0:
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
; X86-NEXT: movl {{[0-9]+}}(%esp), %edx
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_1:
-; X64: # BB#0:
-; X64-NEXT: movq %rdi, %rax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_1:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: movq %rdi, %rax # sched: [1:0.25]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_1:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: movq %rdi, %rax # sched: [1:0.17]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_1:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: movl {{[0-9]+}}(%esp), %edx
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_1:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: movq %rdi, %rax # sched: [1:0.25]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_1:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: movq %rdi, %rax # sched: [1:0.17]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_1:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: movq %rdi, %rax # sched: [1:0.50]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_1:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: movq %rdi, %rax # sched: [1:0.50]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i64 %x, 1
ret i64 %mul
}
@@ -26,10 +63,43 @@ define i64 @test_mul_by_2(i64 %x) {
; X86-NEXT: addl %eax, %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_2:
-; X64: # BB#0:
-; X64-NEXT: leaq (%rdi,%rdi), %rax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_2:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: leaq (%rdi,%rdi), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_2:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: leaq (%rdi,%rdi), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_2:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: movl {{[0-9]+}}(%esp), %edx
+; X86-NOOPT-NEXT: shldl $1, %eax, %edx
+; X86-NOOPT-NEXT: addl %eax, %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_2:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: leaq (%rdi,%rdi), %rax # sched: [1:0.50]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_2:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: leaq (%rdi,%rdi), %rax # sched: [1:0.50]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_2:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: leaq (%rdi,%rdi), %rax # sched: [1:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_2:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: leaq (%rdi,%rdi), %rax # sched: [1:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i64 %x, 2
ret i64 %mul
}
@@ -43,10 +113,43 @@ define i64 @test_mul_by_3(i64 %x) {
; X86-NEXT: addl %ecx, %edx
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_3:
-; X64: # BB#0:
-; X64-NEXT: leaq (%rdi,%rdi,2), %rax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_3:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: leaq (%rdi,%rdi,2), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_3:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: leaq (%rdi,%rdi,2), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_3:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl $3, %eax
+; X86-NOOPT-NEXT: mull {{[0-9]+}}(%esp)
+; X86-NOOPT-NEXT: imull $3, {{[0-9]+}}(%esp), %ecx
+; X86-NOOPT-NEXT: addl %ecx, %edx
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_3:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: leaq (%rdi,%rdi,2), %rax # sched: [1:0.50]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_3:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: leaq (%rdi,%rdi,2), %rax # sched: [1:0.50]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_3:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: leaq (%rdi,%rdi,2), %rax # sched: [1:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_3:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: leaq (%rdi,%rdi,2), %rax # sched: [1:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i64 %x, 3
ret i64 %mul
}
@@ -60,10 +163,43 @@ define i64 @test_mul_by_4(i64 %x) {
; X86-NEXT: shll $2, %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_4:
-; X64: # BB#0:
-; X64-NEXT: leaq (,%rdi,4), %rax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_4:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: leaq (,%rdi,4), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_4:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: leaq (,%rdi,4), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_4:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: movl {{[0-9]+}}(%esp), %edx
+; X86-NOOPT-NEXT: shldl $2, %eax, %edx
+; X86-NOOPT-NEXT: shll $2, %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_4:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: leaq (,%rdi,4), %rax # sched: [1:0.50]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_4:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: leaq (,%rdi,4), %rax # sched: [1:0.50]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_4:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: leaq (,%rdi,4), %rax # sched: [1:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_4:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: leaq (,%rdi,4), %rax # sched: [1:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i64 %x, 4
ret i64 %mul
}
@@ -77,10 +213,43 @@ define i64 @test_mul_by_5(i64 %x) {
; X86-NEXT: addl %ecx, %edx
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_5:
-; X64: # BB#0:
-; X64-NEXT: leaq (%rdi,%rdi,4), %rax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_5:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: leaq (%rdi,%rdi,4), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_5:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: leaq (%rdi,%rdi,4), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_5:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl $5, %eax
+; X86-NOOPT-NEXT: mull {{[0-9]+}}(%esp)
+; X86-NOOPT-NEXT: imull $5, {{[0-9]+}}(%esp), %ecx
+; X86-NOOPT-NEXT: addl %ecx, %edx
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_5:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: leaq (%rdi,%rdi,4), %rax # sched: [1:0.50]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_5:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: leaq (%rdi,%rdi,4), %rax # sched: [1:0.50]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_5:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: leaq (%rdi,%rdi,4), %rax # sched: [1:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_5:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: leaq (%rdi,%rdi,4), %rax # sched: [1:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i64 %x, 5
ret i64 %mul
}
@@ -95,11 +264,46 @@ define i64 @test_mul_by_6(i64 %x) {
; X86-NEXT: leal (%edx,%ecx,2), %edx
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_6:
-; X64: # BB#0:
-; X64-NEXT: addq %rdi, %rdi
-; X64-NEXT: leaq (%rdi,%rdi,2), %rax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_6:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: addq %rdi, %rdi # sched: [1:0.25]
+; X64-HSW-NEXT: leaq (%rdi,%rdi,2), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_6:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: addq %rdi, %rdi # sched: [1:0.50]
+; X64-JAG-NEXT: leaq (%rdi,%rdi,2), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_6:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl $6, %eax
+; X86-NOOPT-NEXT: mull {{[0-9]+}}(%esp)
+; X86-NOOPT-NEXT: imull $6, {{[0-9]+}}(%esp), %ecx
+; X86-NOOPT-NEXT: addl %ecx, %edx
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_6:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imulq $6, %rdi, %rax # sched: [3:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_6:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imulq $6, %rdi, %rax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_6:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: addq %rdi, %rdi # sched: [1:0.50]
+; X64-SLM-NEXT: leaq (%rdi,%rdi,2), %rax # sched: [1:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_6:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imulq $6, %rdi, %rax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i64 %x, 6
ret i64 %mul
}
@@ -115,11 +319,46 @@ define i64 @test_mul_by_7(i64 %x) {
; X86-NEXT: addl %ecx, %edx
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_7:
-; X64: # BB#0:
-; X64-NEXT: leaq (,%rdi,8), %rax
-; X64-NEXT: subq %rdi, %rax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_7:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: leaq (,%rdi,8), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: subq %rdi, %rax # sched: [1:0.25]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_7:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: leaq (,%rdi,8), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: subq %rdi, %rax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_7:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl $7, %eax
+; X86-NOOPT-NEXT: mull {{[0-9]+}}(%esp)
+; X86-NOOPT-NEXT: imull $7, {{[0-9]+}}(%esp), %ecx
+; X86-NOOPT-NEXT: addl %ecx, %edx
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_7:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imulq $7, %rdi, %rax # sched: [3:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_7:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imulq $7, %rdi, %rax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_7:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: leaq (,%rdi,8), %rax # sched: [1:1.00]
+; X64-SLM-NEXT: subq %rdi, %rax # sched: [1:0.50]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_7:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imulq $7, %rdi, %rax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i64 %x, 7
ret i64 %mul
}
@@ -133,10 +372,43 @@ define i64 @test_mul_by_8(i64 %x) {
; X86-NEXT: shll $3, %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_8:
-; X64: # BB#0:
-; X64-NEXT: leaq (,%rdi,8), %rax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_8:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: leaq (,%rdi,8), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_8:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: leaq (,%rdi,8), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_8:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: movl {{[0-9]+}}(%esp), %edx
+; X86-NOOPT-NEXT: shldl $3, %eax, %edx
+; X86-NOOPT-NEXT: shll $3, %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_8:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: leaq (,%rdi,8), %rax # sched: [1:0.50]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_8:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: leaq (,%rdi,8), %rax # sched: [1:0.50]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_8:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: leaq (,%rdi,8), %rax # sched: [1:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_8:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: leaq (,%rdi,8), %rax # sched: [1:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i64 %x, 8
ret i64 %mul
}
@@ -150,10 +422,43 @@ define i64 @test_mul_by_9(i64 %x) {
; X86-NEXT: addl %ecx, %edx
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_9:
-; X64: # BB#0:
-; X64-NEXT: leaq (%rdi,%rdi,8), %rax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_9:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: leaq (%rdi,%rdi,8), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_9:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: leaq (%rdi,%rdi,8), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_9:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl $9, %eax
+; X86-NOOPT-NEXT: mull {{[0-9]+}}(%esp)
+; X86-NOOPT-NEXT: imull $9, {{[0-9]+}}(%esp), %ecx
+; X86-NOOPT-NEXT: addl %ecx, %edx
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_9:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: leaq (%rdi,%rdi,8), %rax # sched: [1:0.50]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_9:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: leaq (%rdi,%rdi,8), %rax # sched: [1:0.50]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_9:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: leaq (%rdi,%rdi,8), %rax # sched: [1:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_9:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: leaq (%rdi,%rdi,8), %rax # sched: [1:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i64 %x, 9
ret i64 %mul
}
@@ -168,11 +473,46 @@ define i64 @test_mul_by_10(i64 %x) {
; X86-NEXT: leal (%edx,%ecx,2), %edx
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_10:
-; X64: # BB#0:
-; X64-NEXT: addq %rdi, %rdi
-; X64-NEXT: leaq (%rdi,%rdi,4), %rax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_10:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: addq %rdi, %rdi # sched: [1:0.25]
+; X64-HSW-NEXT: leaq (%rdi,%rdi,4), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_10:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: addq %rdi, %rdi # sched: [1:0.50]
+; X64-JAG-NEXT: leaq (%rdi,%rdi,4), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_10:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl $10, %eax
+; X86-NOOPT-NEXT: mull {{[0-9]+}}(%esp)
+; X86-NOOPT-NEXT: imull $10, {{[0-9]+}}(%esp), %ecx
+; X86-NOOPT-NEXT: addl %ecx, %edx
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_10:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imulq $10, %rdi, %rax # sched: [3:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_10:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imulq $10, %rdi, %rax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_10:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: addq %rdi, %rdi # sched: [1:0.50]
+; X64-SLM-NEXT: leaq (%rdi,%rdi,4), %rax # sched: [1:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_10:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imulq $10, %rdi, %rax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i64 %x, 10
ret i64 %mul
}
@@ -180,16 +520,53 @@ define i64 @test_mul_by_10(i64 %x) {
define i64 @test_mul_by_11(i64 %x) {
; X86-LABEL: test_mul_by_11:
; X86: # BB#0:
+; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X86-NEXT: leal (%eax,%eax,4), %ecx
+; X86-NEXT: leal (%eax,%ecx,2), %ecx
; X86-NEXT: movl $11, %eax
; X86-NEXT: mull {{[0-9]+}}(%esp)
-; X86-NEXT: imull $11, {{[0-9]+}}(%esp), %ecx
; X86-NEXT: addl %ecx, %edx
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_11:
-; X64: # BB#0:
-; X64-NEXT: imulq $11, %rdi, %rax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_11:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: leaq (%rdi,%rdi,4), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: leaq (%rdi,%rax,2), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_11:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: leaq (%rdi,%rdi,4), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: leaq (%rdi,%rax,2), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_11:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl $11, %eax
+; X86-NOOPT-NEXT: mull {{[0-9]+}}(%esp)
+; X86-NOOPT-NEXT: imull $11, {{[0-9]+}}(%esp), %ecx
+; X86-NOOPT-NEXT: addl %ecx, %edx
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_11:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imulq $11, %rdi, %rax # sched: [3:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_11:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imulq $11, %rdi, %rax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_11:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: imulq $11, %rdi, %rax # sched: [3:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_11:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imulq $11, %rdi, %rax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i64 %x, 11
ret i64 %mul
}
@@ -204,11 +581,46 @@ define i64 @test_mul_by_12(i64 %x) {
; X86-NEXT: leal (%edx,%ecx,4), %edx
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_12:
-; X64: # BB#0:
-; X64-NEXT: shlq $2, %rdi
-; X64-NEXT: leaq (%rdi,%rdi,2), %rax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_12:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: shlq $2, %rdi # sched: [1:0.50]
+; X64-HSW-NEXT: leaq (%rdi,%rdi,2), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_12:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: shlq $2, %rdi # sched: [1:0.50]
+; X64-JAG-NEXT: leaq (%rdi,%rdi,2), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_12:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl $12, %eax
+; X86-NOOPT-NEXT: mull {{[0-9]+}}(%esp)
+; X86-NOOPT-NEXT: imull $12, {{[0-9]+}}(%esp), %ecx
+; X86-NOOPT-NEXT: addl %ecx, %edx
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_12:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imulq $12, %rdi, %rax # sched: [3:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_12:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imulq $12, %rdi, %rax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_12:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: shlq $2, %rdi # sched: [1:1.00]
+; X64-SLM-NEXT: leaq (%rdi,%rdi,2), %rax # sched: [1:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_12:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imulq $12, %rdi, %rax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i64 %x, 12
ret i64 %mul
}
@@ -216,16 +628,53 @@ define i64 @test_mul_by_12(i64 %x) {
define i64 @test_mul_by_13(i64 %x) {
; X86-LABEL: test_mul_by_13:
; X86: # BB#0:
+; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X86-NEXT: leal (%eax,%eax,2), %ecx
+; X86-NEXT: leal (%eax,%ecx,4), %ecx
; X86-NEXT: movl $13, %eax
; X86-NEXT: mull {{[0-9]+}}(%esp)
-; X86-NEXT: imull $13, {{[0-9]+}}(%esp), %ecx
; X86-NEXT: addl %ecx, %edx
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_13:
-; X64: # BB#0:
-; X64-NEXT: imulq $13, %rdi, %rax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_13:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: leaq (%rdi,%rdi,2), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: leaq (%rdi,%rax,4), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_13:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: leaq (%rdi,%rdi,2), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: leaq (%rdi,%rax,4), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_13:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl $13, %eax
+; X86-NOOPT-NEXT: mull {{[0-9]+}}(%esp)
+; X86-NOOPT-NEXT: imull $13, {{[0-9]+}}(%esp), %ecx
+; X86-NOOPT-NEXT: addl %ecx, %edx
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_13:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imulq $13, %rdi, %rax # sched: [3:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_13:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imulq $13, %rdi, %rax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_13:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: imulq $13, %rdi, %rax # sched: [3:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_13:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imulq $13, %rdi, %rax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i64 %x, 13
ret i64 %mul
}
@@ -233,16 +682,56 @@ define i64 @test_mul_by_13(i64 %x) {
define i64 @test_mul_by_14(i64 %x) {
; X86-LABEL: test_mul_by_14:
; X86: # BB#0:
+; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X86-NEXT: leal (%eax,%eax,2), %ecx
+; X86-NEXT: leal (%eax,%ecx,4), %ecx
+; X86-NEXT: addl %eax, %ecx
; X86-NEXT: movl $14, %eax
; X86-NEXT: mull {{[0-9]+}}(%esp)
-; X86-NEXT: imull $14, {{[0-9]+}}(%esp), %ecx
; X86-NEXT: addl %ecx, %edx
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_14:
-; X64: # BB#0:
-; X64-NEXT: imulq $14, %rdi, %rax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_14:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: leaq (%rdi,%rdi,2), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: leaq (%rdi,%rax,4), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: addq %rdi, %rax # sched: [1:0.25]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_14:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: leaq (%rdi,%rdi,2), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: leaq (%rdi,%rax,4), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: addq %rdi, %rax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_14:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl $14, %eax
+; X86-NOOPT-NEXT: mull {{[0-9]+}}(%esp)
+; X86-NOOPT-NEXT: imull $14, {{[0-9]+}}(%esp), %ecx
+; X86-NOOPT-NEXT: addl %ecx, %edx
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_14:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imulq $14, %rdi, %rax # sched: [3:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_14:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imulq $14, %rdi, %rax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_14:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: imulq $14, %rdi, %rax # sched: [3:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_14:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imulq $14, %rdi, %rax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i64 %x, 14
ret i64 %mul
}
@@ -258,11 +747,46 @@ define i64 @test_mul_by_15(i64 %x) {
; X86-NEXT: addl %ecx, %edx
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_15:
-; X64: # BB#0:
-; X64-NEXT: leaq (%rdi,%rdi,4), %rax
-; X64-NEXT: leaq (%rax,%rax,2), %rax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_15:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: leaq (%rdi,%rdi,4), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: leaq (%rax,%rax,2), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_15:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: leaq (%rdi,%rdi,4), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: leaq (%rax,%rax,2), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_15:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl $15, %eax
+; X86-NOOPT-NEXT: mull {{[0-9]+}}(%esp)
+; X86-NOOPT-NEXT: imull $15, {{[0-9]+}}(%esp), %ecx
+; X86-NOOPT-NEXT: addl %ecx, %edx
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_15:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imulq $15, %rdi, %rax # sched: [3:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_15:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imulq $15, %rdi, %rax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_15:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: leaq (%rdi,%rdi,4), %rax # sched: [1:1.00]
+; X64-SLM-NEXT: leaq (%rax,%rax,2), %rax # sched: [1:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_15:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imulq $15, %rdi, %rax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i64 %x, 15
ret i64 %mul
}
@@ -276,11 +800,49 @@ define i64 @test_mul_by_16(i64 %x) {
; X86-NEXT: shll $4, %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_16:
-; X64: # BB#0:
-; X64-NEXT: shlq $4, %rdi
-; X64-NEXT: movq %rdi, %rax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_16:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: shlq $4, %rdi # sched: [1:0.50]
+; X64-HSW-NEXT: movq %rdi, %rax # sched: [1:0.25]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_16:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: shlq $4, %rdi # sched: [1:0.50]
+; X64-JAG-NEXT: movq %rdi, %rax # sched: [1:0.17]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_16:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: movl {{[0-9]+}}(%esp), %edx
+; X86-NOOPT-NEXT: shldl $4, %eax, %edx
+; X86-NOOPT-NEXT: shll $4, %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_16:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: shlq $4, %rdi # sched: [1:0.50]
+; HSW-NOOPT-NEXT: movq %rdi, %rax # sched: [1:0.25]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_16:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: shlq $4, %rdi # sched: [1:0.50]
+; JAG-NOOPT-NEXT: movq %rdi, %rax # sched: [1:0.17]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_16:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: shlq $4, %rdi # sched: [1:1.00]
+; X64-SLM-NEXT: movq %rdi, %rax # sched: [1:0.50]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_16:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: shlq $4, %rdi # sched: [1:1.00]
+; SLM-NOOPT-NEXT: movq %rdi, %rax # sched: [1:0.50]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i64 %x, 16
ret i64 %mul
}
@@ -297,12 +859,49 @@ define i64 @test_mul_by_17(i64 %x) {
; X86-NEXT: addl %ecx, %edx
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_17:
-; X64: # BB#0:
-; X64-NEXT: movq %rdi, %rax
-; X64-NEXT: shlq $4, %rax
-; X64-NEXT: leaq (%rax,%rdi), %rax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_17:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: movq %rdi, %rax # sched: [1:0.25]
+; X64-HSW-NEXT: shlq $4, %rax # sched: [1:0.50]
+; X64-HSW-NEXT: leaq (%rax,%rdi), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_17:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: movq %rdi, %rax # sched: [1:0.17]
+; X64-JAG-NEXT: shlq $4, %rax # sched: [1:0.50]
+; X64-JAG-NEXT: leaq (%rax,%rdi), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_17:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl $17, %eax
+; X86-NOOPT-NEXT: mull {{[0-9]+}}(%esp)
+; X86-NOOPT-NEXT: imull $17, {{[0-9]+}}(%esp), %ecx
+; X86-NOOPT-NEXT: addl %ecx, %edx
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_17:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imulq $17, %rdi, %rax # sched: [3:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_17:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imulq $17, %rdi, %rax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_17:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: movq %rdi, %rax # sched: [1:0.50]
+; X64-SLM-NEXT: shlq $4, %rax # sched: [1:1.00]
+; X64-SLM-NEXT: addq %rdi, %rax # sched: [1:0.50]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_17:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imulq $17, %rdi, %rax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i64 %x, 17
ret i64 %mul
}
@@ -317,11 +916,46 @@ define i64 @test_mul_by_18(i64 %x) {
; X86-NEXT: leal (%edx,%ecx,2), %edx
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_18:
-; X64: # BB#0:
-; X64-NEXT: addq %rdi, %rdi
-; X64-NEXT: leaq (%rdi,%rdi,8), %rax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_18:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: addq %rdi, %rdi # sched: [1:0.25]
+; X64-HSW-NEXT: leaq (%rdi,%rdi,8), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_18:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: addq %rdi, %rdi # sched: [1:0.50]
+; X64-JAG-NEXT: leaq (%rdi,%rdi,8), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_18:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl $18, %eax
+; X86-NOOPT-NEXT: mull {{[0-9]+}}(%esp)
+; X86-NOOPT-NEXT: imull $18, {{[0-9]+}}(%esp), %ecx
+; X86-NOOPT-NEXT: addl %ecx, %edx
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_18:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imulq $18, %rdi, %rax # sched: [3:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_18:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imulq $18, %rdi, %rax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_18:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: addq %rdi, %rdi # sched: [1:0.50]
+; X64-SLM-NEXT: leaq (%rdi,%rdi,8), %rax # sched: [1:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_18:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imulq $18, %rdi, %rax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i64 %x, 18
ret i64 %mul
}
@@ -329,16 +963,58 @@ define i64 @test_mul_by_18(i64 %x) {
define i64 @test_mul_by_19(i64 %x) {
; X86-LABEL: test_mul_by_19:
; X86: # BB#0:
+; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx
+; X86-NEXT: leal (%ecx,%ecx,4), %eax
+; X86-NEXT: shll $2, %eax
+; X86-NEXT: subl %eax, %ecx
; X86-NEXT: movl $19, %eax
; X86-NEXT: mull {{[0-9]+}}(%esp)
-; X86-NEXT: imull $19, {{[0-9]+}}(%esp), %ecx
; X86-NEXT: addl %ecx, %edx
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_19:
-; X64: # BB#0:
-; X64-NEXT: imulq $19, %rdi, %rax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_19:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: leaq (%rdi,%rdi,4), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: shlq $2, %rax # sched: [1:0.50]
+; X64-HSW-NEXT: subq %rax, %rdi # sched: [1:0.25]
+; X64-HSW-NEXT: movq %rdi, %rax # sched: [1:0.25]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_19:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: leaq (%rdi,%rdi,4), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: shlq $2, %rax # sched: [1:0.50]
+; X64-JAG-NEXT: subq %rax, %rdi # sched: [1:0.50]
+; X64-JAG-NEXT: movq %rdi, %rax # sched: [1:0.17]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_19:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl $19, %eax
+; X86-NOOPT-NEXT: mull {{[0-9]+}}(%esp)
+; X86-NOOPT-NEXT: imull $19, {{[0-9]+}}(%esp), %ecx
+; X86-NOOPT-NEXT: addl %ecx, %edx
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_19:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imulq $19, %rdi, %rax # sched: [3:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_19:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imulq $19, %rdi, %rax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_19:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: imulq $19, %rdi, %rax # sched: [3:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_19:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imulq $19, %rdi, %rax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i64 %x, 19
ret i64 %mul
}
@@ -353,11 +1029,46 @@ define i64 @test_mul_by_20(i64 %x) {
; X86-NEXT: leal (%edx,%ecx,4), %edx
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_20:
-; X64: # BB#0:
-; X64-NEXT: shlq $2, %rdi
-; X64-NEXT: leaq (%rdi,%rdi,4), %rax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_20:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: shlq $2, %rdi # sched: [1:0.50]
+; X64-HSW-NEXT: leaq (%rdi,%rdi,4), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_20:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: shlq $2, %rdi # sched: [1:0.50]
+; X64-JAG-NEXT: leaq (%rdi,%rdi,4), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_20:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl $20, %eax
+; X86-NOOPT-NEXT: mull {{[0-9]+}}(%esp)
+; X86-NOOPT-NEXT: imull $20, {{[0-9]+}}(%esp), %ecx
+; X86-NOOPT-NEXT: addl %ecx, %edx
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_20:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imulq $20, %rdi, %rax # sched: [3:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_20:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imulq $20, %rdi, %rax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_20:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: shlq $2, %rdi # sched: [1:1.00]
+; X64-SLM-NEXT: leaq (%rdi,%rdi,4), %rax # sched: [1:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_20:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imulq $20, %rdi, %rax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i64 %x, 20
ret i64 %mul
}
@@ -365,16 +1076,53 @@ define i64 @test_mul_by_20(i64 %x) {
define i64 @test_mul_by_21(i64 %x) {
; X86-LABEL: test_mul_by_21:
; X86: # BB#0:
+; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X86-NEXT: leal (%eax,%eax,4), %ecx
+; X86-NEXT: leal (%eax,%ecx,4), %ecx
; X86-NEXT: movl $21, %eax
; X86-NEXT: mull {{[0-9]+}}(%esp)
-; X86-NEXT: imull $21, {{[0-9]+}}(%esp), %ecx
; X86-NEXT: addl %ecx, %edx
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_21:
-; X64: # BB#0:
-; X64-NEXT: imulq $21, %rdi, %rax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_21:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: leaq (%rdi,%rdi,4), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: leaq (%rdi,%rax,4), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_21:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: leaq (%rdi,%rdi,4), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: leaq (%rdi,%rax,4), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_21:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl $21, %eax
+; X86-NOOPT-NEXT: mull {{[0-9]+}}(%esp)
+; X86-NOOPT-NEXT: imull $21, {{[0-9]+}}(%esp), %ecx
+; X86-NOOPT-NEXT: addl %ecx, %edx
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_21:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imulq $21, %rdi, %rax # sched: [3:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_21:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imulq $21, %rdi, %rax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_21:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: imulq $21, %rdi, %rax # sched: [3:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_21:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imulq $21, %rdi, %rax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i64 %x, 21
ret i64 %mul
}
@@ -382,16 +1130,56 @@ define i64 @test_mul_by_21(i64 %x) {
define i64 @test_mul_by_22(i64 %x) {
; X86-LABEL: test_mul_by_22:
; X86: # BB#0:
+; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X86-NEXT: leal (%eax,%eax,4), %ecx
+; X86-NEXT: leal (%eax,%ecx,4), %ecx
+; X86-NEXT: addl %eax, %ecx
; X86-NEXT: movl $22, %eax
; X86-NEXT: mull {{[0-9]+}}(%esp)
-; X86-NEXT: imull $22, {{[0-9]+}}(%esp), %ecx
; X86-NEXT: addl %ecx, %edx
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_22:
-; X64: # BB#0:
-; X64-NEXT: imulq $22, %rdi, %rax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_22:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: leaq (%rdi,%rdi,4), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: leaq (%rdi,%rax,4), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: addq %rdi, %rax # sched: [1:0.25]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_22:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: leaq (%rdi,%rdi,4), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: leaq (%rdi,%rax,4), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: addq %rdi, %rax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_22:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl $22, %eax
+; X86-NOOPT-NEXT: mull {{[0-9]+}}(%esp)
+; X86-NOOPT-NEXT: imull $22, {{[0-9]+}}(%esp), %ecx
+; X86-NOOPT-NEXT: addl %ecx, %edx
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_22:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imulq $22, %rdi, %rax # sched: [3:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_22:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imulq $22, %rdi, %rax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_22:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: imulq $22, %rdi, %rax # sched: [3:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_22:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imulq $22, %rdi, %rax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i64 %x, 22
ret i64 %mul
}
@@ -399,16 +1187,58 @@ define i64 @test_mul_by_22(i64 %x) {
define i64 @test_mul_by_23(i64 %x) {
; X86-LABEL: test_mul_by_23:
; X86: # BB#0:
+; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx
+; X86-NEXT: leal (%ecx,%ecx,2), %eax
+; X86-NEXT: shll $3, %eax
+; X86-NEXT: subl %eax, %ecx
; X86-NEXT: movl $23, %eax
; X86-NEXT: mull {{[0-9]+}}(%esp)
-; X86-NEXT: imull $23, {{[0-9]+}}(%esp), %ecx
; X86-NEXT: addl %ecx, %edx
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_23:
-; X64: # BB#0:
-; X64-NEXT: imulq $23, %rdi, %rax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_23:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: leaq (%rdi,%rdi,2), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: shlq $3, %rax # sched: [1:0.50]
+; X64-HSW-NEXT: subq %rax, %rdi # sched: [1:0.25]
+; X64-HSW-NEXT: movq %rdi, %rax # sched: [1:0.25]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_23:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: leaq (%rdi,%rdi,2), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: shlq $3, %rax # sched: [1:0.50]
+; X64-JAG-NEXT: subq %rax, %rdi # sched: [1:0.50]
+; X64-JAG-NEXT: movq %rdi, %rax # sched: [1:0.17]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_23:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl $23, %eax
+; X86-NOOPT-NEXT: mull {{[0-9]+}}(%esp)
+; X86-NOOPT-NEXT: imull $23, {{[0-9]+}}(%esp), %ecx
+; X86-NOOPT-NEXT: addl %ecx, %edx
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_23:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imulq $23, %rdi, %rax # sched: [3:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_23:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imulq $23, %rdi, %rax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_23:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: imulq $23, %rdi, %rax # sched: [3:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_23:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imulq $23, %rdi, %rax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i64 %x, 23
ret i64 %mul
}
@@ -423,11 +1253,46 @@ define i64 @test_mul_by_24(i64 %x) {
; X86-NEXT: leal (%edx,%ecx,8), %edx
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_24:
-; X64: # BB#0:
-; X64-NEXT: shlq $3, %rdi
-; X64-NEXT: leaq (%rdi,%rdi,2), %rax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_24:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: shlq $3, %rdi # sched: [1:0.50]
+; X64-HSW-NEXT: leaq (%rdi,%rdi,2), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_24:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: shlq $3, %rdi # sched: [1:0.50]
+; X64-JAG-NEXT: leaq (%rdi,%rdi,2), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_24:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl $24, %eax
+; X86-NOOPT-NEXT: mull {{[0-9]+}}(%esp)
+; X86-NOOPT-NEXT: imull $24, {{[0-9]+}}(%esp), %ecx
+; X86-NOOPT-NEXT: addl %ecx, %edx
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_24:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imulq $24, %rdi, %rax # sched: [3:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_24:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imulq $24, %rdi, %rax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_24:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: shlq $3, %rdi # sched: [1:1.00]
+; X64-SLM-NEXT: leaq (%rdi,%rdi,2), %rax # sched: [1:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_24:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imulq $24, %rdi, %rax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i64 %x, 24
ret i64 %mul
}
@@ -443,11 +1308,46 @@ define i64 @test_mul_by_25(i64 %x) {
; X86-NEXT: addl %ecx, %edx
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_25:
-; X64: # BB#0:
-; X64-NEXT: leaq (%rdi,%rdi,4), %rax
-; X64-NEXT: leaq (%rax,%rax,4), %rax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_25:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: leaq (%rdi,%rdi,4), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: leaq (%rax,%rax,4), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_25:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: leaq (%rdi,%rdi,4), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: leaq (%rax,%rax,4), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_25:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl $25, %eax
+; X86-NOOPT-NEXT: mull {{[0-9]+}}(%esp)
+; X86-NOOPT-NEXT: imull $25, {{[0-9]+}}(%esp), %ecx
+; X86-NOOPT-NEXT: addl %ecx, %edx
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_25:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imulq $25, %rdi, %rax # sched: [3:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_25:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imulq $25, %rdi, %rax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_25:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: leaq (%rdi,%rdi,4), %rax # sched: [1:1.00]
+; X64-SLM-NEXT: leaq (%rax,%rax,4), %rax # sched: [1:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_25:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imulq $25, %rdi, %rax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i64 %x, 25
ret i64 %mul
}
@@ -455,16 +1355,58 @@ define i64 @test_mul_by_25(i64 %x) {
define i64 @test_mul_by_26(i64 %x) {
; X86-LABEL: test_mul_by_26:
; X86: # BB#0:
+; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx
+; X86-NEXT: leal (%ecx,%ecx,8), %eax
+; X86-NEXT: leal (%eax,%eax,2), %eax
+; X86-NEXT: subl %eax, %ecx
; X86-NEXT: movl $26, %eax
; X86-NEXT: mull {{[0-9]+}}(%esp)
-; X86-NEXT: imull $26, {{[0-9]+}}(%esp), %ecx
; X86-NEXT: addl %ecx, %edx
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_26:
-; X64: # BB#0:
-; X64-NEXT: imulq $26, %rdi, %rax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_26:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: leaq (%rdi,%rdi,8), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: leaq (%rax,%rax,2), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: subq %rax, %rdi # sched: [1:0.25]
+; X64-HSW-NEXT: movq %rdi, %rax # sched: [1:0.25]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_26:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: leaq (%rdi,%rdi,8), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: leaq (%rax,%rax,2), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: subq %rax, %rdi # sched: [1:0.50]
+; X64-JAG-NEXT: movq %rdi, %rax # sched: [1:0.17]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_26:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl $26, %eax
+; X86-NOOPT-NEXT: mull {{[0-9]+}}(%esp)
+; X86-NOOPT-NEXT: imull $26, {{[0-9]+}}(%esp), %ecx
+; X86-NOOPT-NEXT: addl %ecx, %edx
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_26:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imulq $26, %rdi, %rax # sched: [3:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_26:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imulq $26, %rdi, %rax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_26:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: imulq $26, %rdi, %rax # sched: [3:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_26:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imulq $26, %rdi, %rax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i64 %x, 26
ret i64 %mul
}
@@ -480,11 +1422,46 @@ define i64 @test_mul_by_27(i64 %x) {
; X86-NEXT: addl %ecx, %edx
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_27:
-; X64: # BB#0:
-; X64-NEXT: leaq (%rdi,%rdi,8), %rax
-; X64-NEXT: leaq (%rax,%rax,2), %rax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_27:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: leaq (%rdi,%rdi,8), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: leaq (%rax,%rax,2), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_27:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: leaq (%rdi,%rdi,8), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: leaq (%rax,%rax,2), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_27:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl $27, %eax
+; X86-NOOPT-NEXT: mull {{[0-9]+}}(%esp)
+; X86-NOOPT-NEXT: imull $27, {{[0-9]+}}(%esp), %ecx
+; X86-NOOPT-NEXT: addl %ecx, %edx
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_27:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imulq $27, %rdi, %rax # sched: [3:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_27:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imulq $27, %rdi, %rax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_27:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: leaq (%rdi,%rdi,8), %rax # sched: [1:1.00]
+; X64-SLM-NEXT: leaq (%rax,%rax,2), %rax # sched: [1:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_27:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imulq $27, %rdi, %rax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i64 %x, 27
ret i64 %mul
}
@@ -492,16 +1469,56 @@ define i64 @test_mul_by_27(i64 %x) {
define i64 @test_mul_by_28(i64 %x) {
; X86-LABEL: test_mul_by_28:
; X86: # BB#0:
+; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X86-NEXT: leal (%eax,%eax,8), %ecx
+; X86-NEXT: leal (%ecx,%ecx,2), %ecx
+; X86-NEXT: addl %eax, %ecx
; X86-NEXT: movl $28, %eax
; X86-NEXT: mull {{[0-9]+}}(%esp)
-; X86-NEXT: imull $28, {{[0-9]+}}(%esp), %ecx
; X86-NEXT: addl %ecx, %edx
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_28:
-; X64: # BB#0:
-; X64-NEXT: imulq $28, %rdi, %rax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_28:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: leaq (%rdi,%rdi,8), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: leaq (%rax,%rax,2), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: addq %rdi, %rax # sched: [1:0.25]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_28:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: leaq (%rdi,%rdi,8), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: leaq (%rax,%rax,2), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: addq %rdi, %rax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_28:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl $28, %eax
+; X86-NOOPT-NEXT: mull {{[0-9]+}}(%esp)
+; X86-NOOPT-NEXT: imull $28, {{[0-9]+}}(%esp), %ecx
+; X86-NOOPT-NEXT: addl %ecx, %edx
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_28:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imulq $28, %rdi, %rax # sched: [3:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_28:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imulq $28, %rdi, %rax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_28:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: imulq $28, %rdi, %rax # sched: [3:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_28:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imulq $28, %rdi, %rax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i64 %x, 28
ret i64 %mul
}
@@ -509,16 +1526,59 @@ define i64 @test_mul_by_28(i64 %x) {
define i64 @test_mul_by_29(i64 %x) {
; X86-LABEL: test_mul_by_29:
; X86: # BB#0:
+; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X86-NEXT: leal (%eax,%eax,8), %ecx
+; X86-NEXT: leal (%ecx,%ecx,2), %ecx
+; X86-NEXT: addl %eax, %ecx
+; X86-NEXT: addl %eax, %ecx
; X86-NEXT: movl $29, %eax
; X86-NEXT: mull {{[0-9]+}}(%esp)
-; X86-NEXT: imull $29, {{[0-9]+}}(%esp), %ecx
; X86-NEXT: addl %ecx, %edx
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_29:
-; X64: # BB#0:
-; X64-NEXT: imulq $29, %rdi, %rax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_29:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: leaq (%rdi,%rdi,8), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: leaq (%rax,%rax,2), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: addq %rdi, %rax # sched: [1:0.25]
+; X64-HSW-NEXT: addq %rdi, %rax # sched: [1:0.25]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_29:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: leaq (%rdi,%rdi,8), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: leaq (%rax,%rax,2), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: addq %rdi, %rax # sched: [1:0.50]
+; X64-JAG-NEXT: addq %rdi, %rax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_29:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl $29, %eax
+; X86-NOOPT-NEXT: mull {{[0-9]+}}(%esp)
+; X86-NOOPT-NEXT: imull $29, {{[0-9]+}}(%esp), %ecx
+; X86-NOOPT-NEXT: addl %ecx, %edx
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_29:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imulq $29, %rdi, %rax # sched: [3:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_29:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imulq $29, %rdi, %rax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_29:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: imulq $29, %rdi, %rax # sched: [3:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_29:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imulq $29, %rdi, %rax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i64 %x, 29
ret i64 %mul
}
@@ -526,16 +1586,60 @@ define i64 @test_mul_by_29(i64 %x) {
define i64 @test_mul_by_30(i64 %x) {
; X86-LABEL: test_mul_by_30:
; X86: # BB#0:
+; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx
+; X86-NEXT: shll $5, %ecx
; X86-NEXT: movl $30, %eax
; X86-NEXT: mull {{[0-9]+}}(%esp)
-; X86-NEXT: imull $30, {{[0-9]+}}(%esp), %ecx
; X86-NEXT: addl %ecx, %edx
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_30:
-; X64: # BB#0:
-; X64-NEXT: imulq $30, %rdi, %rax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_30:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: movq %rdi, %rax # sched: [1:0.25]
+; X64-HSW-NEXT: shlq $5, %rax # sched: [1:0.50]
+; X64-HSW-NEXT: movq %rdi, %rcx # sched: [1:0.25]
+; X64-HSW-NEXT: subq %rax, %rcx # sched: [1:0.25]
+; X64-HSW-NEXT: subq %rcx, %rdi # sched: [1:0.25]
+; X64-HSW-NEXT: movq %rdi, %rax # sched: [1:0.25]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_30:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: movq %rdi, %rax # sched: [1:0.17]
+; X64-JAG-NEXT: movq %rdi, %rcx # sched: [1:0.17]
+; X64-JAG-NEXT: shlq $5, %rax # sched: [1:0.50]
+; X64-JAG-NEXT: subq %rax, %rcx # sched: [1:0.50]
+; X64-JAG-NEXT: subq %rcx, %rdi # sched: [1:0.50]
+; X64-JAG-NEXT: movq %rdi, %rax # sched: [1:0.17]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_30:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl $30, %eax
+; X86-NOOPT-NEXT: mull {{[0-9]+}}(%esp)
+; X86-NOOPT-NEXT: imull $30, {{[0-9]+}}(%esp), %ecx
+; X86-NOOPT-NEXT: addl %ecx, %edx
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_30:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imulq $30, %rdi, %rax # sched: [3:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_30:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imulq $30, %rdi, %rax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_30:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: imulq $30, %rdi, %rax # sched: [3:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_30:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imulq $30, %rdi, %rax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i64 %x, 30
ret i64 %mul
}
@@ -552,12 +1656,49 @@ define i64 @test_mul_by_31(i64 %x) {
; X86-NEXT: addl %ecx, %edx
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_31:
-; X64: # BB#0:
-; X64-NEXT: movq %rdi, %rax
-; X64-NEXT: shlq $5, %rax
-; X64-NEXT: subq %rdi, %rax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_31:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: movq %rdi, %rax # sched: [1:0.25]
+; X64-HSW-NEXT: shlq $5, %rax # sched: [1:0.50]
+; X64-HSW-NEXT: subq %rdi, %rax # sched: [1:0.25]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_31:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: movq %rdi, %rax # sched: [1:0.17]
+; X64-JAG-NEXT: shlq $5, %rax # sched: [1:0.50]
+; X64-JAG-NEXT: subq %rdi, %rax # sched: [1:0.50]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_31:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl $31, %eax
+; X86-NOOPT-NEXT: mull {{[0-9]+}}(%esp)
+; X86-NOOPT-NEXT: imull $31, {{[0-9]+}}(%esp), %ecx
+; X86-NOOPT-NEXT: addl %ecx, %edx
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_31:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: imulq $31, %rdi, %rax # sched: [3:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_31:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: imulq $31, %rdi, %rax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_31:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: movq %rdi, %rax # sched: [1:0.50]
+; X64-SLM-NEXT: shlq $5, %rax # sched: [1:1.00]
+; X64-SLM-NEXT: subq %rdi, %rax # sched: [1:0.50]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_31:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: imulq $31, %rdi, %rax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i64 %x, 31
ret i64 %mul
}
@@ -571,11 +1712,168 @@ define i64 @test_mul_by_32(i64 %x) {
; X86-NEXT: shll $5, %eax
; X86-NEXT: retl
;
-; X64-LABEL: test_mul_by_32:
-; X64: # BB#0:
-; X64-NEXT: shlq $5, %rdi
-; X64-NEXT: movq %rdi, %rax
-; X64-NEXT: retq
+; X64-HSW-LABEL: test_mul_by_32:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: shlq $5, %rdi # sched: [1:0.50]
+; X64-HSW-NEXT: movq %rdi, %rax # sched: [1:0.25]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_by_32:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: shlq $5, %rdi # sched: [1:0.50]
+; X64-JAG-NEXT: movq %rdi, %rax # sched: [1:0.17]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_by_32:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X86-NOOPT-NEXT: movl {{[0-9]+}}(%esp), %edx
+; X86-NOOPT-NEXT: shldl $5, %eax, %edx
+; X86-NOOPT-NEXT: shll $5, %eax
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_by_32:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: shlq $5, %rdi # sched: [1:0.50]
+; HSW-NOOPT-NEXT: movq %rdi, %rax # sched: [1:0.25]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_by_32:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: shlq $5, %rdi # sched: [1:0.50]
+; JAG-NOOPT-NEXT: movq %rdi, %rax # sched: [1:0.17]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_by_32:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: shlq $5, %rdi # sched: [1:1.00]
+; X64-SLM-NEXT: movq %rdi, %rax # sched: [1:0.50]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_by_32:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: shlq $5, %rdi # sched: [1:1.00]
+; SLM-NOOPT-NEXT: movq %rdi, %rax # sched: [1:0.50]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
%mul = mul nsw i64 %x, 32
ret i64 %mul
}
+
+; (x*9+42)*(x*5+2)
+define i64 @test_mul_spec(i64 %x) nounwind {
+; X86-LABEL: test_mul_spec:
+; X86: # BB#0:
+; X86-NEXT: pushl %ebx
+; X86-NEXT: pushl %edi
+; X86-NEXT: pushl %esi
+; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx
+; X86-NEXT: movl {{[0-9]+}}(%esp), %edi
+; X86-NEXT: movl $9, %edx
+; X86-NEXT: movl %ecx, %eax
+; X86-NEXT: mull %edx
+; X86-NEXT: movl %eax, %esi
+; X86-NEXT: leal (%edi,%edi,8), %ebx
+; X86-NEXT: addl $42, %esi
+; X86-NEXT: adcl %edx, %ebx
+; X86-NEXT: movl $5, %edx
+; X86-NEXT: movl %ecx, %eax
+; X86-NEXT: mull %edx
+; X86-NEXT: movl %eax, %ecx
+; X86-NEXT: leal (%edi,%edi,4), %edi
+; X86-NEXT: addl $2, %ecx
+; X86-NEXT: adcl %edx, %edi
+; X86-NEXT: movl %esi, %eax
+; X86-NEXT: mull %ecx
+; X86-NEXT: imull %esi, %edi
+; X86-NEXT: addl %edi, %edx
+; X86-NEXT: imull %ebx, %ecx
+; X86-NEXT: addl %ecx, %edx
+; X86-NEXT: popl %esi
+; X86-NEXT: popl %edi
+; X86-NEXT: popl %ebx
+; X86-NEXT: retl
+;
+; X64-HSW-LABEL: test_mul_spec:
+; X64-HSW: # BB#0:
+; X64-HSW-NEXT: leaq (%rdi,%rdi,8), %rcx # sched: [1:0.50]
+; X64-HSW-NEXT: addq $42, %rcx # sched: [1:0.25]
+; X64-HSW-NEXT: leaq (%rdi,%rdi,4), %rax # sched: [1:0.50]
+; X64-HSW-NEXT: addq $2, %rax # sched: [1:0.25]
+; X64-HSW-NEXT: imulq %rcx, %rax # sched: [3:1.00]
+; X64-HSW-NEXT: retq # sched: [1:1.00]
+;
+; X64-JAG-LABEL: test_mul_spec:
+; X64-JAG: # BB#0:
+; X64-JAG-NEXT: leaq 42(%rdi,%rdi,8), %rcx # sched: [1:0.50]
+; X64-JAG-NEXT: leaq 2(%rdi,%rdi,4), %rax # sched: [1:0.50]
+; X64-JAG-NEXT: imulq %rcx, %rax # sched: [3:1.00]
+; X64-JAG-NEXT: retq # sched: [4:1.00]
+;
+; X86-NOOPT-LABEL: test_mul_spec:
+; X86-NOOPT: # BB#0:
+; X86-NOOPT-NEXT: pushl %ebx
+; X86-NOOPT-NEXT: pushl %edi
+; X86-NOOPT-NEXT: pushl %esi
+; X86-NOOPT-NEXT: movl {{[0-9]+}}(%esp), %ecx
+; X86-NOOPT-NEXT: movl {{[0-9]+}}(%esp), %edi
+; X86-NOOPT-NEXT: movl $9, %edx
+; X86-NOOPT-NEXT: movl %ecx, %eax
+; X86-NOOPT-NEXT: mull %edx
+; X86-NOOPT-NEXT: movl %eax, %esi
+; X86-NOOPT-NEXT: leal (%edi,%edi,8), %ebx
+; X86-NOOPT-NEXT: addl $42, %esi
+; X86-NOOPT-NEXT: adcl %edx, %ebx
+; X86-NOOPT-NEXT: movl $5, %edx
+; X86-NOOPT-NEXT: movl %ecx, %eax
+; X86-NOOPT-NEXT: mull %edx
+; X86-NOOPT-NEXT: movl %eax, %ecx
+; X86-NOOPT-NEXT: leal (%edi,%edi,4), %edi
+; X86-NOOPT-NEXT: addl $2, %ecx
+; X86-NOOPT-NEXT: adcl %edx, %edi
+; X86-NOOPT-NEXT: movl %esi, %eax
+; X86-NOOPT-NEXT: mull %ecx
+; X86-NOOPT-NEXT: imull %esi, %edi
+; X86-NOOPT-NEXT: addl %edi, %edx
+; X86-NOOPT-NEXT: imull %ebx, %ecx
+; X86-NOOPT-NEXT: addl %ecx, %edx
+; X86-NOOPT-NEXT: popl %esi
+; X86-NOOPT-NEXT: popl %edi
+; X86-NOOPT-NEXT: popl %ebx
+; X86-NOOPT-NEXT: retl
+;
+; HSW-NOOPT-LABEL: test_mul_spec:
+; HSW-NOOPT: # BB#0:
+; HSW-NOOPT-NEXT: leaq (%rdi,%rdi,8), %rcx # sched: [1:0.50]
+; HSW-NOOPT-NEXT: addq $42, %rcx # sched: [1:0.25]
+; HSW-NOOPT-NEXT: leaq (%rdi,%rdi,4), %rax # sched: [1:0.50]
+; HSW-NOOPT-NEXT: addq $2, %rax # sched: [1:0.25]
+; HSW-NOOPT-NEXT: imulq %rcx, %rax # sched: [3:1.00]
+; HSW-NOOPT-NEXT: retq # sched: [1:1.00]
+;
+; JAG-NOOPT-LABEL: test_mul_spec:
+; JAG-NOOPT: # BB#0:
+; JAG-NOOPT-NEXT: leaq 42(%rdi,%rdi,8), %rcx # sched: [1:0.50]
+; JAG-NOOPT-NEXT: leaq 2(%rdi,%rdi,4), %rax # sched: [1:0.50]
+; JAG-NOOPT-NEXT: imulq %rcx, %rax # sched: [3:1.00]
+; JAG-NOOPT-NEXT: retq # sched: [4:1.00]
+;
+; X64-SLM-LABEL: test_mul_spec:
+; X64-SLM: # BB#0:
+; X64-SLM-NEXT: leaq 42(%rdi,%rdi,8), %rcx # sched: [1:1.00]
+; X64-SLM-NEXT: leaq 2(%rdi,%rdi,4), %rax # sched: [1:1.00]
+; X64-SLM-NEXT: imulq %rcx, %rax # sched: [3:1.00]
+; X64-SLM-NEXT: retq # sched: [4:1.00]
+;
+; SLM-NOOPT-LABEL: test_mul_spec:
+; SLM-NOOPT: # BB#0:
+; SLM-NOOPT-NEXT: leaq 42(%rdi,%rdi,8), %rcx # sched: [1:1.00]
+; SLM-NOOPT-NEXT: leaq 2(%rdi,%rdi,4), %rax # sched: [1:1.00]
+; SLM-NOOPT-NEXT: imulq %rcx, %rax # sched: [3:1.00]
+; SLM-NOOPT-NEXT: retq # sched: [4:1.00]
+ %mul = mul nsw i64 %x, 9
+ %add = add nsw i64 %mul, 42
+ %mul2 = mul nsw i64 %x, 5
+ %add2 = add nsw i64 %mul2, 2
+ %mul3 = mul nsw i64 %add, %add2
+ ret i64 %mul3
+}
diff --git a/test/CodeGen/X86/setcc-lowering.ll b/test/CodeGen/X86/setcc-lowering.ll
index 391f1cc9fb43..1b8f8e7ae559 100644
--- a/test/CodeGen/X86/setcc-lowering.ll
+++ b/test/CodeGen/X86/setcc-lowering.ll
@@ -41,14 +41,67 @@ entry:
ret <8 x i16> %3
}
-define void @pr26232(i64 %a) {
+define void @pr26232(i64 %a, <16 x i1> %b) {
; AVX-LABEL: pr26232:
; AVX: # BB#0: # %for_loop599.preheader
+; AVX-NEXT: vpxor %xmm1, %xmm1, %xmm1
; AVX-NEXT: .p2align 4, 0x90
; AVX-NEXT: .LBB1_1: # %for_loop599
; AVX-NEXT: # =>This Inner Loop Header: Depth=1
+; AVX-NEXT: xorl %eax, %eax
; AVX-NEXT: cmpq $65536, %rdi # imm = 0x10000
-; AVX-NEXT: setl -{{[0-9]+}}(%rsp)
+; AVX-NEXT: setl %al
+; AVX-NEXT: vmovd %eax, %xmm2
+; AVX-NEXT: vpshufb %xmm1, %xmm2, %xmm2
+; AVX-NEXT: vpand %xmm0, %xmm2, %xmm2
+; AVX-NEXT: vpextrb $15, %xmm2, %eax
+; AVX-NEXT: andb $1, %al
+; AVX-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX-NEXT: vpextrb $14, %xmm2, %eax
+; AVX-NEXT: andb $1, %al
+; AVX-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX-NEXT: vpextrb $13, %xmm2, %eax
+; AVX-NEXT: andb $1, %al
+; AVX-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX-NEXT: vpextrb $12, %xmm2, %eax
+; AVX-NEXT: andb $1, %al
+; AVX-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX-NEXT: vpextrb $11, %xmm2, %eax
+; AVX-NEXT: andb $1, %al
+; AVX-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX-NEXT: vpextrb $10, %xmm2, %eax
+; AVX-NEXT: andb $1, %al
+; AVX-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX-NEXT: vpextrb $9, %xmm2, %eax
+; AVX-NEXT: andb $1, %al
+; AVX-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX-NEXT: vpextrb $8, %xmm2, %eax
+; AVX-NEXT: andb $1, %al
+; AVX-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX-NEXT: vpextrb $7, %xmm2, %eax
+; AVX-NEXT: andb $1, %al
+; AVX-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX-NEXT: vpextrb $6, %xmm2, %eax
+; AVX-NEXT: andb $1, %al
+; AVX-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX-NEXT: vpextrb $5, %xmm2, %eax
+; AVX-NEXT: andb $1, %al
+; AVX-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX-NEXT: vpextrb $4, %xmm2, %eax
+; AVX-NEXT: andb $1, %al
+; AVX-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX-NEXT: vpextrb $3, %xmm2, %eax
+; AVX-NEXT: andb $1, %al
+; AVX-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX-NEXT: vpextrb $2, %xmm2, %eax
+; AVX-NEXT: andb $1, %al
+; AVX-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX-NEXT: vpextrb $1, %xmm2, %eax
+; AVX-NEXT: andb $1, %al
+; AVX-NEXT: movb %al, -{{[0-9]+}}(%rsp)
+; AVX-NEXT: vpextrb $0, %xmm2, %eax
+; AVX-NEXT: andb $1, %al
+; AVX-NEXT: movb %al, -{{[0-9]+}}(%rsp)
; AVX-NEXT: cmpw $0, -{{[0-9]+}}(%rsp)
; AVX-NEXT: jne .LBB1_1
; AVX-NEXT: # BB#2: # %for_exit600
@@ -61,6 +114,9 @@ define void @pr26232(i64 %a) {
; KNL-32-NEXT: .cfi_def_cfa_offset 8
; KNL-32-NEXT: .Lcfi1:
; KNL-32-NEXT: .cfi_offset %esi, -8
+; KNL-32-NEXT: vpmovsxbd %xmm0, %zmm0
+; KNL-32-NEXT: vpslld $31, %zmm0, %zmm0
+; KNL-32-NEXT: vptestmd %zmm0, %zmm0, %k0
; KNL-32-NEXT: movl {{[0-9]+}}(%esp), %eax
; KNL-32-NEXT: movl {{[0-9]+}}(%esp), %ecx
; KNL-32-NEXT: movw $-1, %dx
@@ -72,6 +128,9 @@ define void @pr26232(i64 %a) {
; KNL-32-NEXT: sbbl $0, %esi
; KNL-32-NEXT: movl $0, %esi
; KNL-32-NEXT: cmovlw %dx, %si
+; KNL-32-NEXT: kmovw %esi, %k1
+; KNL-32-NEXT: kandw %k0, %k1, %k1
+; KNL-32-NEXT: kmovw %k1, %esi
; KNL-32-NEXT: testw %si, %si
; KNL-32-NEXT: jne .LBB1_1
; KNL-32-NEXT: # BB#2: # %for_exit600
@@ -87,7 +146,7 @@ for_loop599: ; preds = %for_loop599, %for_t
%less_i_load605_ = icmp slt i64 %a, 65536
%less_i_load605__broadcast_init = insertelement <16 x i1> undef, i1 %less_i_load605_, i32 0
%less_i_load605__broadcast = shufflevector <16 x i1> %less_i_load605__broadcast_init, <16 x i1> undef, <16 x i32> zeroinitializer
- %"oldMask&test607" = and <16 x i1> %less_i_load605__broadcast, undef
+ %"oldMask&test607" = and <16 x i1> %less_i_load605__broadcast, %b
%intmask.i894 = bitcast <16 x i1> %"oldMask&test607" to i16
%res.i895 = icmp eq i16 %intmask.i894, 0
br i1 %res.i895, label %for_exit600, label %for_loop599
diff --git a/test/CodeGen/X86/vector-sext.ll b/test/CodeGen/X86/vector-sext.ll
index 8cc1d8c765ac..53e471d6f175 100644
--- a/test/CodeGen/X86/vector-sext.ll
+++ b/test/CodeGen/X86/vector-sext.ll
@@ -1749,6 +1749,62 @@ entry:
ret <4 x i64> %Y
}
+define <2 x i64> @load_sext_4i8_to_4i64_extract(<4 x i8> *%ptr) {
+; SSE2-LABEL: load_sext_4i8_to_4i64_extract:
+; SSE2: # BB#0:
+; SSE2-NEXT: movsbq 3(%rdi), %rax
+; SSE2-NEXT: movq %rax, %xmm1
+; SSE2-NEXT: movsbq 2(%rdi), %rax
+; SSE2-NEXT: movq %rax, %xmm0
+; SSE2-NEXT: punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; SSE2-NEXT: retq
+;
+; SSSE3-LABEL: load_sext_4i8_to_4i64_extract:
+; SSSE3: # BB#0:
+; SSSE3-NEXT: movsbq 3(%rdi), %rax
+; SSSE3-NEXT: movq %rax, %xmm1
+; SSSE3-NEXT: movsbq 2(%rdi), %rax
+; SSSE3-NEXT: movq %rax, %xmm0
+; SSSE3-NEXT: punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; SSSE3-NEXT: retq
+;
+; SSE41-LABEL: load_sext_4i8_to_4i64_extract:
+; SSE41: # BB#0:
+; SSE41-NEXT: pmovsxbq 2(%rdi), %xmm0
+; SSE41-NEXT: retq
+;
+; AVX1-LABEL: load_sext_4i8_to_4i64_extract:
+; AVX1: # BB#0:
+; AVX1-NEXT: vpmovsxbd (%rdi), %xmm0
+; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
+; AVX1-NEXT: vpmovsxdq %xmm0, %xmm0
+; AVX1-NEXT: retq
+;
+; AVX2-LABEL: load_sext_4i8_to_4i64_extract:
+; AVX2: # BB#0:
+; AVX2-NEXT: vpmovsxbq (%rdi), %ymm0
+; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm0
+; AVX2-NEXT: vzeroupper
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: load_sext_4i8_to_4i64_extract:
+; AVX512: # BB#0:
+; AVX512-NEXT: vpmovsxbq (%rdi), %ymm0
+; AVX512-NEXT: vextracti128 $1, %ymm0, %xmm0
+; AVX512-NEXT: vzeroupper
+; AVX512-NEXT: retq
+;
+; X32-SSE41-LABEL: load_sext_4i8_to_4i64_extract:
+; X32-SSE41: # BB#0:
+; X32-SSE41-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X32-SSE41-NEXT: pmovsxbq 2(%eax), %xmm0
+; X32-SSE41-NEXT: retl
+ %ld = load <4 x i8>, <4 x i8>* %ptr
+ %sext = sext <4 x i8> %ld to <4 x i64>
+ %extract = shufflevector <4 x i64> %sext, <4 x i64> undef, <2 x i32> <i32 2, i32 3>
+ ret <2 x i64> %extract
+}
+
define <8 x i16> @load_sext_8i1_to_8i16(<8 x i1> *%ptr) {
; SSE2-LABEL: load_sext_8i1_to_8i16:
; SSE2: # BB#0: # %entry
diff --git a/test/CodeGen/X86/xchg-nofold.ll b/test/CodeGen/X86/xchg-nofold.ll
new file mode 100644
index 000000000000..fddc7906e08f
--- /dev/null
+++ b/test/CodeGen/X86/xchg-nofold.ll
@@ -0,0 +1,37 @@
+; RUN: llc -mtriple=x86_64-linux-gnu < %s | FileCheck %s
+
+%"struct.std::atomic" = type { %"struct.std::atomic_bool" }
+%"struct.std::atomic_bool" = type { %"struct.std::__atomic_base" }
+%"struct.std::__atomic_base" = type { i8 }
+
+; CHECK-LABEL: _Z3fooRSt6atomicIbEb
+define zeroext i1 @_Z3fooRSt6atomicIbEb(%"struct.std::atomic"* nocapture dereferenceable(1) %a, i1 returned zeroext %b) nounwind {
+entry:
+ %frombool.i.i = zext i1 %b to i8
+ %_M_i.i.i = getelementptr inbounds %"struct.std::atomic", %"struct.std::atomic"* %a, i64 0, i32 0, i32 0, i32 0
+ %0 = ptrtoint i8* %_M_i.i.i to i64
+ %1 = lshr i64 %0, 3
+ %2 = add i64 %1, 2147450880
+ %3 = inttoptr i64 %2 to i8*
+ %4 = load i8, i8* %3
+ %5 = icmp ne i8 %4, 0
+ br i1 %5, label %6, label %11
+
+; <label>:6: ; preds = %entry
+ %7 = and i64 %0, 7
+ %8 = trunc i64 %7 to i8
+ %9 = icmp sge i8 %8, %4
+ br i1 %9, label %10, label %11
+
+; <label>:10: ; preds = %6
+ call void @__asan_report_store1(i64 %0)
+ call void asm sideeffect "", ""()
+ unreachable
+
+; <label>:11: ; preds = %6, %entry
+ store atomic i8 %frombool.i.i, i8* %_M_i.i.i seq_cst, align 1
+; CHECK: xchgb %{{.*}}, (%{{.*}})
+ ret i1 %b
+}
+
+declare void @__asan_report_store1(i64)
generated by cgit v1.3 (git 2.53.0) at 2026-08-10 22:27:11 +0000