diff options
Diffstat (limited to 'test/CodeGen/X86/avx-vzeroupper.ll')
| -rw-r--r-- | test/CodeGen/X86/avx-vzeroupper.ll | 108 |
1 files changed, 0 insertions, 108 deletions
diff --git a/test/CodeGen/X86/avx-vzeroupper.ll b/test/CodeGen/X86/avx-vzeroupper.ll deleted file mode 100644 index cf514d7aeb31..000000000000 --- a/test/CodeGen/X86/avx-vzeroupper.ll +++ /dev/null @@ -1,108 +0,0 @@ -; RUN: llc < %s -x86-use-vzeroupper -mtriple=x86_64-apple-darwin -mattr=+avx | FileCheck %s -; RUN: llc < %s -x86-use-vzeroupper -mtriple=x86_64-apple-darwin -mattr=+avx512f | FileCheck %s -; RUN: llc < %s -x86-use-vzeroupper -mtriple=x86_64-apple-darwin -mattr=+avx,+fast-partial-ymm-or-zmm-write | FileCheck --check-prefix=FAST-YMM-ZMM %s -; RUN: llc < %s -x86-use-vzeroupper -mtriple=x86_64-apple-darwin -mcpu=btver2 | FileCheck --check-prefix=BTVER2 %s - -; FAST-YMM-ZMM-NOT: vzeroupper -; BTVER2-NOT: vzeroupper - -declare i32 @foo() -declare <4 x float> @do_sse(<4 x float>) -declare <8 x float> @do_avx(<8 x float>) -declare <4 x float> @llvm.x86.avx.vextractf128.ps.256(<8 x float>, i8) nounwind readnone -@x = common global <4 x float> zeroinitializer, align 16 -@g = common global <8 x float> zeroinitializer, align 32 - -;; Basic checking - don't emit any vzeroupper instruction - -; CHECK: _test00 -define <4 x float> @test00(<4 x float> %a, <4 x float> %b) nounwind uwtable ssp { -entry: - ; CHECK-NOT: vzeroupper - %add.i = fadd <4 x float> %a, %b - %call3 = call <4 x float> @do_sse(<4 x float> %add.i) nounwind - ; CHECK: ret - ret <4 x float> %call3 -} - -;; Check parameter 256-bit parameter passing - -; CHECK: _test01 -define <8 x float> @test01(<4 x float> %a, <4 x float> %b, <8 x float> %c) nounwind uwtable ssp { -entry: - %tmp = load <4 x float>, <4 x float>* @x, align 16 - ; CHECK: vzeroupper - ; CHECK-NEXT: callq _do_sse - %call = tail call <4 x float> @do_sse(<4 x float> %tmp) nounwind - store <4 x float> %call, <4 x float>* @x, align 16 - ; CHECK-NOT: vzeroupper - ; CHECK: callq _do_sse - %call2 = tail call <4 x float> @do_sse(<4 x float> %call) nounwind - store <4 x float> %call2, <4 x float>* @x, align 16 - ; CHECK: ret - ret <8 x float> %c -} - -;; Check that vzeroupper is emitted for tail calls. - -; CHECK: _test02 -define <4 x float> @test02(<8 x float> %a, <8 x float> %b) nounwind uwtable ssp { -entry: - %add.i = fadd <8 x float> %a, %b - %add.low = call <4 x float> @llvm.x86.avx.vextractf128.ps.256(<8 x float> %add.i, i8 0) - ; CHECK: vzeroupper - ; CHECK: jmp _do_sse - %call3 = tail call <4 x float> @do_sse(<4 x float> %add.low) nounwind - ret <4 x float> %call3 -} - -;; Test the pass convergence and also that vzeroupper is only issued when necessary, -;; for this function it should be only once - -; CHECK: _test03 -define <4 x float> @test03(<4 x float> %a, <4 x float> %b) nounwind uwtable ssp { -entry: - %add.i = fadd <4 x float> %a, %b - br label %while.cond - -while.cond: - %call = tail call i32 @foo() - %tobool = icmp eq i32 %call, 0 - br i1 %tobool, label %for.body, label %while.cond - -for.body: - ; CHECK: LBB - ; CHECK-NOT: vzeroupper - %i.018 = phi i32 [ 0, %while.cond ], [ %1, %for.body ] - %c.017 = phi <4 x float> [ %add.i, %while.cond ], [ %call14, %for.body ] - ; CHECK: callq _do_sse - %call5 = tail call <4 x float> @do_sse(<4 x float> %c.017) nounwind - ; CHECK-NEXT: callq _do_sse - %call7 = tail call <4 x float> @do_sse(<4 x float> %call5) nounwind - %tmp11 = load <8 x float>, <8 x float>* @g, align 32 - %0 = tail call <4 x float> @llvm.x86.avx.vextractf128.ps.256(<8 x float> %tmp11, i8 1) nounwind - ; CHECK: vzeroupper - ; CHECK-NEXT: callq _do_sse - %call14 = tail call <4 x float> @do_sse(<4 x float> %0) nounwind - %1 = add nsw i32 %i.018, 1 - %exitcond = icmp eq i32 %1, 4 - br i1 %exitcond, label %for.end, label %for.body - -for.end: - ret <4 x float> %call14 -} - -;; Check that we also perform vzeroupper when we return from a function. - -; CHECK: _test04 -define <4 x float> @test04(<4 x float> %a, <4 x float> %b) nounwind uwtable ssp { -entry: - %shuf = shufflevector <4 x float> %a, <4 x float> %b, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7> - ; CHECK-NOT: vzeroupper - ; CHECK: call - %call = call <8 x float> @do_avx(<8 x float> %shuf) nounwind - %shuf2 = shufflevector <8 x float> %call, <8 x float> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3> - ; CHECK: vzeroupper - ; CHECK: ret - ret <4 x float> %shuf2 -} |
