https://github.com/jofrn updated https://github.com/llvm/llvm-project/pull/148899
>From ab47398f3abb2b2cbe6a98ab00c1c6e1e60a19e3 Mon Sep 17 00:00:00 2001 From: jofrn <[email protected]> Date: Tue, 15 Jul 2025 13:02:04 -0400 Subject: [PATCH] [X86] Cast atomic vectors in IR to support floats This commit casts floats to ints in an atomic load during AtomicExpand to support floating point types. It also is required to support 128 bit vectors in SSE/AVX. --- llvm/lib/Target/X86/X86ISelLowering.cpp | 7 + llvm/lib/Target/X86/X86ISelLowering.h | 2 + llvm/test/CodeGen/X86/atomic-load-store.ll | 508 ++++----------------- 3 files changed, 109 insertions(+), 408 deletions(-) diff --git a/llvm/lib/Target/X86/X86ISelLowering.cpp b/llvm/lib/Target/X86/X86ISelLowering.cpp index 49beadae63f03..e15f17281b958 100644 --- a/llvm/lib/Target/X86/X86ISelLowering.cpp +++ b/llvm/lib/Target/X86/X86ISelLowering.cpp @@ -32136,6 +32136,13 @@ X86TargetLowering::shouldExpandAtomicRMWInIR(AtomicRMWInst *AI) const { } } +TargetLowering::AtomicExpansionKind +X86TargetLowering::shouldCastAtomicLoadInIR(LoadInst *LI) const { + if (LI->getType()->getScalarType()->isFloatingPointTy()) + return AtomicExpansionKind::CastToInteger; + return AtomicExpansionKind::None; +} + LoadInst * X86TargetLowering::lowerIdempotentRMWIntoFencedLoad(AtomicRMWInst *AI) const { unsigned NativeWidth = Subtarget.is64Bit() ? 64 : 32; diff --git a/llvm/lib/Target/X86/X86ISelLowering.h b/llvm/lib/Target/X86/X86ISelLowering.h index b7151f65942b4..f9a8adbd7da0d 100644 --- a/llvm/lib/Target/X86/X86ISelLowering.h +++ b/llvm/lib/Target/X86/X86ISelLowering.h @@ -1841,6 +1841,8 @@ namespace llvm { shouldExpandAtomicRMWInIR(AtomicRMWInst *AI) const override; TargetLoweringBase::AtomicExpansionKind shouldExpandLogicAtomicRMWInIR(AtomicRMWInst *AI) const; + TargetLoweringBase::AtomicExpansionKind + shouldCastAtomicLoadInIR(LoadInst *LI) const override; void emitBitTestAtomicRMWIntrinsic(AtomicRMWInst *AI) const override; void emitCmpArithAtomicRMWIntrinsic(AtomicRMWInst *AI) const override; diff --git a/llvm/test/CodeGen/X86/atomic-load-store.ll b/llvm/test/CodeGen/X86/atomic-load-store.ll index d9ee379b4bd2f..4e8f637f34113 100644 --- a/llvm/test/CodeGen/X86/atomic-load-store.ll +++ b/llvm/test/CodeGen/X86/atomic-load-store.ll @@ -1,10 +1,10 @@ ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py -; RUN: llc < %s -mtriple=x86_64-- -verify-machineinstrs -mcpu=x86-64 | FileCheck %s --check-prefixes=CHECK,CHECK-O3,CHECK-SSE-O3 -; RUN: llc < %s -mtriple=x86_64-- -verify-machineinstrs -mcpu=x86-64-v2 | FileCheck %s --check-prefixes=CHECK,CHECK-O3,CHECK-SSE-O3 +; RUN: llc < %s -mtriple=x86_64-- -verify-machineinstrs -mcpu=x86-64 | FileCheck %s --check-prefixes=CHECK,CHECK-O3,CHECK-SSE-O3,CHECK-SSE1-O3 +; RUN: llc < %s -mtriple=x86_64-- -verify-machineinstrs -mcpu=x86-64-v2 | FileCheck %s --check-prefixes=CHECK,CHECK-O3,CHECK-SSE-O3,CHECK-SSE2-O3 ; RUN: llc < %s -mtriple=x86_64-- -verify-machineinstrs -mcpu=x86-64-v3 | FileCheck %s --check-prefixes=CHECK,CHECK-O3,CHECK-AVX-O3,CHECK-AVX2-O3 ; RUN: llc < %s -mtriple=x86_64-- -verify-machineinstrs -mcpu=x86-64-v4 | FileCheck %s --check-prefixes=CHECK,CHECK-O3,CHECK-AVX-O3,CHECK-AVX512-O3 -; RUN: llc < %s -mtriple=x86_64-- -verify-machineinstrs -O0 -mcpu=x86-64 | FileCheck %s --check-prefixes=CHECK,CHECK-O0,CHECK-SSE-O0 -; RUN: llc < %s -mtriple=x86_64-- -verify-machineinstrs -O0 -mcpu=x86-64-v2 | FileCheck %s --check-prefixes=CHECK,CHECK-O0,CHECK-SSE-O0 +; RUN: llc < %s -mtriple=x86_64-- -verify-machineinstrs -O0 -mcpu=x86-64 | FileCheck %s --check-prefixes=CHECK,CHECK-O0,CHECK-SSE-O0,CHECK-SSE1-O0 +; RUN: llc < %s -mtriple=x86_64-- -verify-machineinstrs -O0 -mcpu=x86-64-v2 | FileCheck %s --check-prefixes=CHECK,CHECK-O0,CHECK-SSE-O0,CHECK-SSE2-O0 ; RUN: llc < %s -mtriple=x86_64-- -verify-machineinstrs -O0 -mcpu=x86-64-v3 | FileCheck %s --check-prefixes=CHECK,CHECK-O0,CHECK-AVX-O0,CHECK-AVX2-O0 ; RUN: llc < %s -mtriple=x86_64-- -verify-machineinstrs -O0 -mcpu=x86-64-v4 | FileCheck %s --check-prefixes=CHECK,CHECK-O0,CHECK-AVX-O0,CHECK-AVX512-O0 @@ -119,13 +119,13 @@ define <1 x bfloat> @atomic_vec1_bfloat(ptr %x) { ; CHECK-SSE-O3-LABEL: atomic_vec1_bfloat: ; CHECK-SSE-O3: # %bb.0: ; CHECK-SSE-O3-NEXT: movzwl (%rdi), %eax -; CHECK-SSE-O3-NEXT: pinsrw $0, %eax, %xmm0 +; CHECK-SSE-O3-NEXT: movd %eax, %xmm0 ; CHECK-SSE-O3-NEXT: retq ; ; CHECK-AVX-O3-LABEL: atomic_vec1_bfloat: ; CHECK-AVX-O3: # %bb.0: ; CHECK-AVX-O3-NEXT: movzwl (%rdi), %eax -; CHECK-AVX-O3-NEXT: vpinsrw $0, %eax, %xmm0, %xmm0 +; CHECK-AVX-O3-NEXT: vmovd %eax, %xmm0 ; CHECK-AVX-O3-NEXT: retq ; ; CHECK-SSE-O0-LABEL: atomic_vec1_bfloat: @@ -133,8 +133,7 @@ define <1 x bfloat> @atomic_vec1_bfloat(ptr %x) { ; CHECK-SSE-O0-NEXT: movw (%rdi), %cx ; CHECK-SSE-O0-NEXT: # implicit-def: $eax ; CHECK-SSE-O0-NEXT: movw %cx, %ax -; CHECK-SSE-O0-NEXT: # implicit-def: $xmm0 -; CHECK-SSE-O0-NEXT: pinsrw $0, %eax, %xmm0 +; CHECK-SSE-O0-NEXT: movd %eax, %xmm0 ; CHECK-SSE-O0-NEXT: retq ; ; CHECK-AVX-O0-LABEL: atomic_vec1_bfloat: @@ -142,8 +141,7 @@ define <1 x bfloat> @atomic_vec1_bfloat(ptr %x) { ; CHECK-AVX-O0-NEXT: movw (%rdi), %cx ; CHECK-AVX-O0-NEXT: # implicit-def: $eax ; CHECK-AVX-O0-NEXT: movw %cx, %ax -; CHECK-AVX-O0-NEXT: # implicit-def: $xmm0 -; CHECK-AVX-O0-NEXT: vpinsrw $0, %eax, %xmm0, %xmm0 +; CHECK-AVX-O0-NEXT: vmovd %eax, %xmm0 ; CHECK-AVX-O0-NEXT: retq %ret = load atomic <1 x bfloat>, ptr %x acquire, align 2 ret <1 x bfloat> %ret @@ -296,22 +294,9 @@ define <2 x float> @atomic_vec2_float_align(ptr %x) { } define <2 x half> @atomic_vec2_half(ptr %x) { -; CHECK-O3-LABEL: atomic_vec2_half: -; CHECK-O3: # %bb.0: -; CHECK-O3-NEXT: movl (%rdi), %eax -; CHECK-O3-NEXT: pinsrw $0, %eax, %xmm0 -; CHECK-O3-NEXT: shrl $16, %eax -; CHECK-O3-NEXT: pinsrw $0, %eax, %xmm1 -; CHECK-O3-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3] -; CHECK-O3-NEXT: retq -; ; CHECK-SSE-O3-LABEL: atomic_vec2_half: ; CHECK-SSE-O3: # %bb.0: -; CHECK-SSE-O3-NEXT: movl (%rdi), %eax -; CHECK-SSE-O3-NEXT: pinsrw $0, %eax, %xmm0 -; CHECK-SSE-O3-NEXT: shrl $16, %eax -; CHECK-SSE-O3-NEXT: pinsrw $0, %eax, %xmm1 -; CHECK-SSE-O3-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3] +; CHECK-SSE-O3-NEXT: movss {{.*#+}} xmm0 = mem[0],zero,zero,zero ; CHECK-SSE-O3-NEXT: retq ; ; CHECK-AVX-O3-LABEL: atomic_vec2_half: @@ -319,153 +304,38 @@ define <2 x half> @atomic_vec2_half(ptr %x) { ; CHECK-AVX-O3-NEXT: vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero ; CHECK-AVX-O3-NEXT: retq ; -; CHECK-AVX512-O3-LABEL: atomic_vec2_half: -; CHECK-AVX512-O3: # %bb.0: -; CHECK-AVX512-O3-NEXT: vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero -; CHECK-AVX512-O3-NEXT: retq -; -; CHECK-O0-LABEL: atomic_vec2_half: -; CHECK-O0: # %bb.0: -; CHECK-O0-NEXT: movl (%rdi), %eax -; CHECK-O0-NEXT: movl %eax, %ecx -; CHECK-O0-NEXT: shrl $16, %ecx -; CHECK-O0-NEXT: movw %cx, %dx -; CHECK-O0-NEXT: # implicit-def: $ecx -; CHECK-O0-NEXT: movw %dx, %cx -; CHECK-O0-NEXT: # implicit-def: $xmm1 -; CHECK-O0-NEXT: pinsrw $0, %ecx, %xmm1 -; CHECK-O0-NEXT: movw %ax, %cx -; CHECK-O0-NEXT: # implicit-def: $eax -; CHECK-O0-NEXT: movw %cx, %ax -; CHECK-O0-NEXT: # implicit-def: $xmm0 -; CHECK-O0-NEXT: pinsrw $0, %eax, %xmm0 -; CHECK-O0-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3] -; CHECK-O0-NEXT: retq -; ; CHECK-SSE-O0-LABEL: atomic_vec2_half: ; CHECK-SSE-O0: # %bb.0: -; CHECK-SSE-O0-NEXT: movl (%rdi), %eax -; CHECK-SSE-O0-NEXT: movl %eax, %ecx -; CHECK-SSE-O0-NEXT: shrl $16, %ecx -; CHECK-SSE-O0-NEXT: movw %cx, %dx -; CHECK-SSE-O0-NEXT: # implicit-def: $ecx -; CHECK-SSE-O0-NEXT: movw %dx, %cx -; CHECK-SSE-O0-NEXT: # implicit-def: $xmm1 -; CHECK-SSE-O0-NEXT: pinsrw $0, %ecx, %xmm1 -; CHECK-SSE-O0-NEXT: movw %ax, %cx -; CHECK-SSE-O0-NEXT: # implicit-def: $eax -; CHECK-SSE-O0-NEXT: movw %cx, %ax -; CHECK-SSE-O0-NEXT: # implicit-def: $xmm0 -; CHECK-SSE-O0-NEXT: pinsrw $0, %eax, %xmm0 -; CHECK-SSE-O0-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3] +; CHECK-SSE-O0-NEXT: movd {{.*#+}} xmm0 = mem[0],zero,zero,zero ; CHECK-SSE-O0-NEXT: retq ; ; CHECK-AVX-O0-LABEL: atomic_vec2_half: ; CHECK-AVX-O0: # %bb.0: ; CHECK-AVX-O0-NEXT: vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero ; CHECK-AVX-O0-NEXT: retq -; -; CHECK-AVX512-O0-LABEL: atomic_vec2_half: -; CHECK-AVX512-O0: # %bb.0: -; CHECK-AVX512-O0-NEXT: vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero -; CHECK-AVX512-O0-NEXT: retq %ret = load atomic <2 x half>, ptr %x acquire, align 4 ret <2 x half> %ret } define <2 x bfloat> @atomic_vec2_bfloat(ptr %x) { -; CHECK-O3-LABEL: atomic_vec2_bfloat: -; CHECK-O3: # %bb.0: -; CHECK-O3-NEXT: movl (%rdi), %eax -; CHECK-O3-NEXT: pinsrw $0, %eax, %xmm0 -; CHECK-O3-NEXT: shrl $16, %eax -; CHECK-O3-NEXT: pinsrw $0, %eax, %xmm1 -; CHECK-O3-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3] -; CHECK-O3-NEXT: retq -; ; CHECK-SSE-O3-LABEL: atomic_vec2_bfloat: ; CHECK-SSE-O3: # %bb.0: -; CHECK-SSE-O3-NEXT: movl (%rdi), %eax -; CHECK-SSE-O3-NEXT: pinsrw $0, %eax, %xmm0 -; CHECK-SSE-O3-NEXT: shrl $16, %eax -; CHECK-SSE-O3-NEXT: pinsrw $0, %eax, %xmm1 -; CHECK-SSE-O3-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3] +; CHECK-SSE-O3-NEXT: movss {{.*#+}} xmm0 = mem[0],zero,zero,zero ; CHECK-SSE-O3-NEXT: retq ; ; CHECK-AVX-O3-LABEL: atomic_vec2_bfloat: ; CHECK-AVX-O3: # %bb.0: -; CHECK-AVX-O3-NEXT: movl (%rdi), %eax -; CHECK-AVX-O3-NEXT: movw %ax, -{{[0-9]+}}(%rsp) -; CHECK-AVX-O3-NEXT: shrl $16, %eax -; CHECK-AVX-O3-NEXT: movw %ax, -{{[0-9]+}}(%rsp) -; CHECK-AVX-O3-NEXT: vmovaps -{{[0-9]+}}(%rsp), %xmm0 +; CHECK-AVX-O3-NEXT: vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero ; CHECK-AVX-O3-NEXT: retq ; -; CHECK-AVX512-O3-LABEL: atomic_vec2_bfloat: -; CHECK-AVX512-O3: # %bb.0: -; CHECK-AVX512-O3-NEXT: movl (%rdi), %eax -; CHECK-AVX512-O3-NEXT: movw %ax, -{{[0-9]+}}(%rsp) -; CHECK-AVX512-O3-NEXT: shrl $16, %eax -; CHECK-AVX512-O3-NEXT: movw %ax, -{{[0-9]+}}(%rsp) -; CHECK-AVX512-O3-NEXT: vmovaps -{{[0-9]+}}(%rsp), %xmm0 -; CHECK-AVX512-O3-NEXT: retq -; -; CHECK-O0-LABEL: atomic_vec2_bfloat: -; CHECK-O0: # %bb.0: -; CHECK-O0-NEXT: movl (%rdi), %eax -; CHECK-O0-NEXT: movl %eax, %ecx -; CHECK-O0-NEXT: shrl $16, %ecx -; CHECK-O0-NEXT: # kill: def $cx killed $cx killed $ecx -; CHECK-O0-NEXT: movw %ax, %dx -; CHECK-O0-NEXT: # implicit-def: $eax -; CHECK-O0-NEXT: movw %dx, %ax -; CHECK-O0-NEXT: # implicit-def: $xmm0 -; CHECK-O0-NEXT: pinsrw $0, %eax, %xmm0 -; CHECK-O0-NEXT: # implicit-def: $eax -; CHECK-O0-NEXT: movw %cx, %ax -; CHECK-O0-NEXT: # implicit-def: $xmm1 -; CHECK-O0-NEXT: pinsrw $0, %eax, %xmm1 -; CHECK-O0-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3] -; CHECK-O0-NEXT: retq -; ; CHECK-SSE-O0-LABEL: atomic_vec2_bfloat: ; CHECK-SSE-O0: # %bb.0: -; CHECK-SSE-O0-NEXT: movl (%rdi), %eax -; CHECK-SSE-O0-NEXT: movl %eax, %ecx -; CHECK-SSE-O0-NEXT: shrl $16, %ecx -; CHECK-SSE-O0-NEXT: # kill: def $cx killed $cx killed $ecx -; CHECK-SSE-O0-NEXT: movw %ax, %dx -; CHECK-SSE-O0-NEXT: # implicit-def: $eax -; CHECK-SSE-O0-NEXT: movw %dx, %ax -; CHECK-SSE-O0-NEXT: # implicit-def: $xmm0 -; CHECK-SSE-O0-NEXT: pinsrw $0, %eax, %xmm0 -; CHECK-SSE-O0-NEXT: # implicit-def: $eax -; CHECK-SSE-O0-NEXT: movw %cx, %ax -; CHECK-SSE-O0-NEXT: # implicit-def: $xmm1 -; CHECK-SSE-O0-NEXT: pinsrw $0, %eax, %xmm1 -; CHECK-SSE-O0-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3] +; CHECK-SSE-O0-NEXT: movd {{.*#+}} xmm0 = mem[0],zero,zero,zero ; CHECK-SSE-O0-NEXT: retq ; ; CHECK-AVX-O0-LABEL: atomic_vec2_bfloat: ; CHECK-AVX-O0: # %bb.0: -; CHECK-AVX-O0-NEXT: movl (%rdi), %eax -; CHECK-AVX-O0-NEXT: movw %ax, %cx -; CHECK-AVX-O0-NEXT: movw %cx, -{{[0-9]+}}(%rsp) -; CHECK-AVX-O0-NEXT: shrl $16, %eax -; CHECK-AVX-O0-NEXT: # kill: def $ax killed $ax killed $eax -; CHECK-AVX-O0-NEXT: movw %ax, -{{[0-9]+}}(%rsp) -; CHECK-AVX-O0-NEXT: vmovaps -{{[0-9]+}}(%rsp), %xmm0 +; CHECK-AVX-O0-NEXT: vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero ; CHECK-AVX-O0-NEXT: retq -; -; CHECK-AVX512-O0-LABEL: atomic_vec2_bfloat: -; CHECK-AVX512-O0: # %bb.0: -; CHECK-AVX512-O0-NEXT: movl (%rdi), %eax -; CHECK-AVX512-O0-NEXT: movw %ax, %cx -; CHECK-AVX512-O0-NEXT: movw %cx, -{{[0-9]+}}(%rsp) -; CHECK-AVX512-O0-NEXT: shrl $16, %eax -; CHECK-AVX512-O0-NEXT: # kill: def $ax killed $ax killed $eax -; CHECK-AVX512-O0-NEXT: movw %ax, -{{[0-9]+}}(%rsp) -; CHECK-AVX512-O0-NEXT: vmovaps -{{[0-9]+}}(%rsp), %xmm0 -; CHECK-AVX512-O0-NEXT: retq %ret = load atomic <2 x bfloat>, ptr %x acquire, align 4 ret <2 x bfloat> %ret } @@ -501,13 +371,13 @@ define <1 x half> @atomic_vec1_half(ptr %x) { ; CHECK-SSE-O3-LABEL: atomic_vec1_half: ; CHECK-SSE-O3: # %bb.0: ; CHECK-SSE-O3-NEXT: movzwl (%rdi), %eax -; CHECK-SSE-O3-NEXT: pinsrw $0, %eax, %xmm0 +; CHECK-SSE-O3-NEXT: movd %eax, %xmm0 ; CHECK-SSE-O3-NEXT: retq ; ; CHECK-AVX-O3-LABEL: atomic_vec1_half: ; CHECK-AVX-O3: # %bb.0: ; CHECK-AVX-O3-NEXT: movzwl (%rdi), %eax -; CHECK-AVX-O3-NEXT: vpinsrw $0, %eax, %xmm0, %xmm0 +; CHECK-AVX-O3-NEXT: vmovd %eax, %xmm0 ; CHECK-AVX-O3-NEXT: retq ; ; CHECK-SSE-O0-LABEL: atomic_vec1_half: @@ -515,8 +385,7 @@ define <1 x half> @atomic_vec1_half(ptr %x) { ; CHECK-SSE-O0-NEXT: movw (%rdi), %cx ; CHECK-SSE-O0-NEXT: # implicit-def: $eax ; CHECK-SSE-O0-NEXT: movw %cx, %ax -; CHECK-SSE-O0-NEXT: # implicit-def: $xmm0 -; CHECK-SSE-O0-NEXT: pinsrw $0, %eax, %xmm0 +; CHECK-SSE-O0-NEXT: movd %eax, %xmm0 ; CHECK-SSE-O0-NEXT: retq ; ; CHECK-AVX-O0-LABEL: atomic_vec1_half: @@ -524,8 +393,7 @@ define <1 x half> @atomic_vec1_half(ptr %x) { ; CHECK-AVX-O0-NEXT: movw (%rdi), %cx ; CHECK-AVX-O0-NEXT: # implicit-def: $eax ; CHECK-AVX-O0-NEXT: movw %cx, %ax -; CHECK-AVX-O0-NEXT: # implicit-def: $xmm0 -; CHECK-AVX-O0-NEXT: vpinsrw $0, %eax, %xmm0, %xmm0 +; CHECK-AVX-O0-NEXT: vmovd %eax, %xmm0 ; CHECK-AVX-O0-NEXT: retq %ret = load atomic <1 x half>, ptr %x acquire, align 2 ret <1 x half> %ret @@ -760,38 +628,9 @@ define <4 x i16> @atomic_vec4_i16(ptr %x) nounwind { } define <4 x half> @atomic_vec4_half(ptr %x) nounwind { -; CHECK-O3-LABEL: atomic_vec4_half: -; CHECK-O3: # %bb.0: -; CHECK-O3-NEXT: movq (%rdi), %rax -; CHECK-O3-NEXT: movl %eax, %ecx -; CHECK-O3-NEXT: shrl $16, %ecx -; CHECK-O3-NEXT: pinsrw $0, %ecx, %xmm1 -; CHECK-O3-NEXT: pinsrw $0, %eax, %xmm0 -; CHECK-O3-NEXT: movq %rax, %rcx -; CHECK-O3-NEXT: shrq $32, %rcx -; CHECK-O3-NEXT: pinsrw $0, %ecx, %xmm2 -; CHECK-O3-NEXT: shrq $48, %rax -; CHECK-O3-NEXT: pinsrw $0, %eax, %xmm3 -; CHECK-O3-NEXT: punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1],xmm2[2],xmm3[2],xmm2[3],xmm3[3] -; CHECK-O3-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3] -; CHECK-O3-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1] -; CHECK-O3-NEXT: retq -; ; CHECK-SSE-O3-LABEL: atomic_vec4_half: ; CHECK-SSE-O3: # %bb.0: -; CHECK-SSE-O3-NEXT: movq (%rdi), %rax -; CHECK-SSE-O3-NEXT: movl %eax, %ecx -; CHECK-SSE-O3-NEXT: shrl $16, %ecx -; CHECK-SSE-O3-NEXT: pinsrw $0, %ecx, %xmm1 -; CHECK-SSE-O3-NEXT: pinsrw $0, %eax, %xmm0 -; CHECK-SSE-O3-NEXT: movq %rax, %rcx -; CHECK-SSE-O3-NEXT: shrq $32, %rcx -; CHECK-SSE-O3-NEXT: pinsrw $0, %ecx, %xmm2 -; CHECK-SSE-O3-NEXT: shrq $48, %rax -; CHECK-SSE-O3-NEXT: pinsrw $0, %eax, %xmm3 -; CHECK-SSE-O3-NEXT: punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1],xmm2[2],xmm3[2],xmm2[3],xmm3[3] -; CHECK-SSE-O3-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3] -; CHECK-SSE-O3-NEXT: insertps {{.*#+}} xmm0 = xmm0[0],xmm2[0],zero,zero +; CHECK-SSE-O3-NEXT: movq (%rdi), %xmm0 ; CHECK-SSE-O3-NEXT: retq ; ; CHECK-AVX-O3-LABEL: atomic_vec4_half: @@ -799,261 +638,38 @@ define <4 x half> @atomic_vec4_half(ptr %x) nounwind { ; CHECK-AVX-O3-NEXT: vmovq (%rdi), %xmm0 ; CHECK-AVX-O3-NEXT: retq ; -; CHECK-AVX512-O3-LABEL: atomic_vec4_half: -; CHECK-AVX512-O3: # %bb.0: -; CHECK-AVX512-O3-NEXT: vmovq (%rdi), %xmm0 -; CHECK-AVX512-O3-NEXT: retq -; -; CHECK-O0-LABEL: atomic_vec4_half: -; CHECK-O0: # %bb.0: -; CHECK-O0-NEXT: movq (%rdi), %rax -; CHECK-O0-NEXT: movl %eax, %ecx -; CHECK-O0-NEXT: shrl $16, %ecx -; CHECK-O0-NEXT: movw %cx, %dx -; CHECK-O0-NEXT: # implicit-def: $ecx -; CHECK-O0-NEXT: movw %dx, %cx -; CHECK-O0-NEXT: # implicit-def: $xmm2 -; CHECK-O0-NEXT: pinsrw $0, %ecx, %xmm2 -; CHECK-O0-NEXT: movw %ax, %dx -; CHECK-O0-NEXT: # implicit-def: $ecx -; CHECK-O0-NEXT: movw %dx, %cx -; CHECK-O0-NEXT: # implicit-def: $xmm0 -; CHECK-O0-NEXT: pinsrw $0, %ecx, %xmm0 -; CHECK-O0-NEXT: movq %rax, %rcx -; CHECK-O0-NEXT: shrq $32, %rcx -; CHECK-O0-NEXT: movw %cx, %dx -; CHECK-O0-NEXT: # implicit-def: $ecx -; CHECK-O0-NEXT: movw %dx, %cx -; CHECK-O0-NEXT: # implicit-def: $xmm1 -; CHECK-O0-NEXT: pinsrw $0, %ecx, %xmm1 -; CHECK-O0-NEXT: shrq $48, %rax -; CHECK-O0-NEXT: movw %ax, %cx -; CHECK-O0-NEXT: # implicit-def: $eax -; CHECK-O0-NEXT: movw %cx, %ax -; CHECK-O0-NEXT: # implicit-def: $xmm3 -; CHECK-O0-NEXT: pinsrw $0, %eax, %xmm3 -; CHECK-O0-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3] -; CHECK-O0-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3] -; CHECK-O0-NEXT: unpcklps {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] -; CHECK-O0-NEXT: retq -; ; CHECK-SSE-O0-LABEL: atomic_vec4_half: ; CHECK-SSE-O0: # %bb.0: -; CHECK-SSE-O0-NEXT: movq (%rdi), %rax -; CHECK-SSE-O0-NEXT: movl %eax, %ecx -; CHECK-SSE-O0-NEXT: shrl $16, %ecx -; CHECK-SSE-O0-NEXT: movw %cx, %dx -; CHECK-SSE-O0-NEXT: # implicit-def: $ecx -; CHECK-SSE-O0-NEXT: movw %dx, %cx -; CHECK-SSE-O0-NEXT: # implicit-def: $xmm2 -; CHECK-SSE-O0-NEXT: pinsrw $0, %ecx, %xmm2 -; CHECK-SSE-O0-NEXT: movw %ax, %dx -; CHECK-SSE-O0-NEXT: # implicit-def: $ecx -; CHECK-SSE-O0-NEXT: movw %dx, %cx -; CHECK-SSE-O0-NEXT: # implicit-def: $xmm0 -; CHECK-SSE-O0-NEXT: pinsrw $0, %ecx, %xmm0 -; CHECK-SSE-O0-NEXT: movq %rax, %rcx -; CHECK-SSE-O0-NEXT: shrq $32, %rcx -; CHECK-SSE-O0-NEXT: movw %cx, %dx -; CHECK-SSE-O0-NEXT: # implicit-def: $ecx -; CHECK-SSE-O0-NEXT: movw %dx, %cx -; CHECK-SSE-O0-NEXT: # implicit-def: $xmm1 -; CHECK-SSE-O0-NEXT: pinsrw $0, %ecx, %xmm1 -; CHECK-SSE-O0-NEXT: shrq $48, %rax -; CHECK-SSE-O0-NEXT: movw %ax, %cx -; CHECK-SSE-O0-NEXT: # implicit-def: $eax -; CHECK-SSE-O0-NEXT: movw %cx, %ax -; CHECK-SSE-O0-NEXT: # implicit-def: $xmm3 -; CHECK-SSE-O0-NEXT: pinsrw $0, %eax, %xmm3 -; CHECK-SSE-O0-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3] -; CHECK-SSE-O0-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3] -; CHECK-SSE-O0-NEXT: insertps {{.*#+}} xmm0 = xmm0[0],xmm1[0],zero,zero +; CHECK-SSE-O0-NEXT: movq (%rdi), %xmm0 ; CHECK-SSE-O0-NEXT: retq ; ; CHECK-AVX-O0-LABEL: atomic_vec4_half: ; CHECK-AVX-O0: # %bb.0: ; CHECK-AVX-O0-NEXT: vmovq (%rdi), %xmm0 ; CHECK-AVX-O0-NEXT: retq -; -; CHECK-AVX512-O0-LABEL: atomic_vec4_half: -; CHECK-AVX512-O0: # %bb.0: -; CHECK-AVX512-O0-NEXT: vmovq (%rdi), %xmm0 -; CHECK-AVX512-O0-NEXT: retq %ret = load atomic <4 x half>, ptr %x acquire, align 8 ret <4 x half> %ret } define <4 x bfloat> @atomic_vec4_bfloat(ptr %x) nounwind { -; CHECK-O3-LABEL: atomic_vec4_bfloat: -; CHECK-O3: # %bb.0: -; CHECK-O3-NEXT: movq (%rdi), %rax -; CHECK-O3-NEXT: movq %rax, %rcx -; CHECK-O3-NEXT: movq %rax, %rdx -; CHECK-O3-NEXT: pinsrw $0, %eax, %xmm0 -; CHECK-O3-NEXT: # kill: def $eax killed $eax killed $rax -; CHECK-O3-NEXT: shrl $16, %eax -; CHECK-O3-NEXT: shrq $32, %rcx -; CHECK-O3-NEXT: shrq $48, %rdx -; CHECK-O3-NEXT: pinsrw $0, %edx, %xmm1 -; CHECK-O3-NEXT: pinsrw $0, %ecx, %xmm2 -; CHECK-O3-NEXT: pinsrw $0, %eax, %xmm3 -; CHECK-O3-NEXT: punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3] -; CHECK-O3-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3] -; CHECK-O3-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1] -; CHECK-O3-NEXT: retq -; ; CHECK-SSE-O3-LABEL: atomic_vec4_bfloat: ; CHECK-SSE-O3: # %bb.0: -; CHECK-SSE-O3-NEXT: movq (%rdi), %rax -; CHECK-SSE-O3-NEXT: movq %rax, %rcx -; CHECK-SSE-O3-NEXT: movq %rax, %rdx -; CHECK-SSE-O3-NEXT: pinsrw $0, %eax, %xmm0 -; CHECK-SSE-O3-NEXT: # kill: def $eax killed $eax killed $rax -; CHECK-SSE-O3-NEXT: shrl $16, %eax -; CHECK-SSE-O3-NEXT: shrq $32, %rcx -; CHECK-SSE-O3-NEXT: shrq $48, %rdx -; CHECK-SSE-O3-NEXT: pinsrw $0, %edx, %xmm1 -; CHECK-SSE-O3-NEXT: pinsrw $0, %ecx, %xmm2 -; CHECK-SSE-O3-NEXT: pinsrw $0, %eax, %xmm3 -; CHECK-SSE-O3-NEXT: punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3] -; CHECK-SSE-O3-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3] -; CHECK-SSE-O3-NEXT: insertps {{.*#+}} xmm0 = xmm0[0],xmm2[0],zero,zero +; CHECK-SSE-O3-NEXT: movq (%rdi), %xmm0 ; CHECK-SSE-O3-NEXT: retq ; ; CHECK-AVX-O3-LABEL: atomic_vec4_bfloat: ; CHECK-AVX-O3: # %bb.0: -; CHECK-AVX-O3-NEXT: movq (%rdi), %rax -; CHECK-AVX-O3-NEXT: movw %ax, -{{[0-9]+}}(%rsp) -; CHECK-AVX-O3-NEXT: movq %rax, %rcx -; CHECK-AVX-O3-NEXT: shrq $48, %rcx -; CHECK-AVX-O3-NEXT: movw %cx, -{{[0-9]+}}(%rsp) -; CHECK-AVX-O3-NEXT: movq %rax, %rcx -; CHECK-AVX-O3-NEXT: shrq $32, %rcx -; CHECK-AVX-O3-NEXT: movw %cx, -{{[0-9]+}}(%rsp) -; CHECK-AVX-O3-NEXT: shrl $16, %eax -; CHECK-AVX-O3-NEXT: movw %ax, -{{[0-9]+}}(%rsp) -; CHECK-AVX-O3-NEXT: vmovaps -{{[0-9]+}}(%rsp), %xmm0 +; CHECK-AVX-O3-NEXT: vmovq (%rdi), %xmm0 ; CHECK-AVX-O3-NEXT: retq ; -; CHECK-AVX512-O3-LABEL: atomic_vec4_bfloat: -; CHECK-AVX512-O3: # %bb.0: -; CHECK-AVX512-O3-NEXT: movq (%rdi), %rax -; CHECK-AVX512-O3-NEXT: movw %ax, -{{[0-9]+}}(%rsp) -; CHECK-AVX512-O3-NEXT: movq %rax, %rcx -; CHECK-AVX512-O3-NEXT: shrq $48, %rcx -; CHECK-AVX512-O3-NEXT: movw %cx, -{{[0-9]+}}(%rsp) -; CHECK-AVX512-O3-NEXT: movq %rax, %rcx -; CHECK-AVX512-O3-NEXT: shrq $32, %rcx -; CHECK-AVX512-O3-NEXT: movw %cx, -{{[0-9]+}}(%rsp) -; CHECK-AVX512-O3-NEXT: shrl $16, %eax -; CHECK-AVX512-O3-NEXT: movw %ax, -{{[0-9]+}}(%rsp) -; CHECK-AVX512-O3-NEXT: vmovaps -{{[0-9]+}}(%rsp), %xmm0 -; CHECK-AVX512-O3-NEXT: retq -; -; CHECK-O0-LABEL: atomic_vec4_bfloat: -; CHECK-O0: # %bb.0: -; CHECK-O0-NEXT: movq (%rdi), %rax -; CHECK-O0-NEXT: movl %eax, %ecx -; CHECK-O0-NEXT: shrl $16, %ecx -; CHECK-O0-NEXT: # kill: def $cx killed $cx killed $ecx -; CHECK-O0-NEXT: movw %ax, %dx -; CHECK-O0-NEXT: movq %rax, %rsi -; CHECK-O0-NEXT: shrq $32, %rsi -; CHECK-O0-NEXT: # kill: def $si killed $si killed $rsi -; CHECK-O0-NEXT: shrq $48, %rax -; CHECK-O0-NEXT: movw %ax, %di -; CHECK-O0-NEXT: # implicit-def: $eax -; CHECK-O0-NEXT: movw %di, %ax -; CHECK-O0-NEXT: # implicit-def: $xmm0 -; CHECK-O0-NEXT: pinsrw $0, %eax, %xmm0 -; CHECK-O0-NEXT: # implicit-def: $eax -; CHECK-O0-NEXT: movw %si, %ax -; CHECK-O0-NEXT: # implicit-def: $xmm1 -; CHECK-O0-NEXT: pinsrw $0, %eax, %xmm1 -; CHECK-O0-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3] -; CHECK-O0-NEXT: # implicit-def: $eax -; CHECK-O0-NEXT: movw %dx, %ax -; CHECK-O0-NEXT: # implicit-def: $xmm0 -; CHECK-O0-NEXT: pinsrw $0, %eax, %xmm0 -; CHECK-O0-NEXT: # implicit-def: $eax -; CHECK-O0-NEXT: movw %cx, %ax -; CHECK-O0-NEXT: # implicit-def: $xmm2 -; CHECK-O0-NEXT: pinsrw $0, %eax, %xmm2 -; CHECK-O0-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3] -; CHECK-O0-NEXT: unpcklps {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] -; CHECK-O0-NEXT: retq -; ; CHECK-SSE-O0-LABEL: atomic_vec4_bfloat: ; CHECK-SSE-O0: # %bb.0: -; CHECK-SSE-O0-NEXT: movq (%rdi), %rax -; CHECK-SSE-O0-NEXT: movl %eax, %ecx -; CHECK-SSE-O0-NEXT: shrl $16, %ecx -; CHECK-SSE-O0-NEXT: # kill: def $cx killed $cx killed $ecx -; CHECK-SSE-O0-NEXT: movw %ax, %dx -; CHECK-SSE-O0-NEXT: movq %rax, %rsi -; CHECK-SSE-O0-NEXT: shrq $32, %rsi -; CHECK-SSE-O0-NEXT: # kill: def $si killed $si killed $rsi -; CHECK-SSE-O0-NEXT: shrq $48, %rax -; CHECK-SSE-O0-NEXT: movw %ax, %di -; CHECK-SSE-O0-NEXT: # implicit-def: $eax -; CHECK-SSE-O0-NEXT: movw %di, %ax -; CHECK-SSE-O0-NEXT: # implicit-def: $xmm0 -; CHECK-SSE-O0-NEXT: pinsrw $0, %eax, %xmm0 -; CHECK-SSE-O0-NEXT: # implicit-def: $eax -; CHECK-SSE-O0-NEXT: movw %si, %ax -; CHECK-SSE-O0-NEXT: # implicit-def: $xmm1 -; CHECK-SSE-O0-NEXT: pinsrw $0, %eax, %xmm1 -; CHECK-SSE-O0-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3] -; CHECK-SSE-O0-NEXT: # implicit-def: $eax -; CHECK-SSE-O0-NEXT: movw %dx, %ax -; CHECK-SSE-O0-NEXT: # implicit-def: $xmm0 -; CHECK-SSE-O0-NEXT: pinsrw $0, %eax, %xmm0 -; CHECK-SSE-O0-NEXT: # implicit-def: $eax -; CHECK-SSE-O0-NEXT: movw %cx, %ax -; CHECK-SSE-O0-NEXT: # implicit-def: $xmm2 -; CHECK-SSE-O0-NEXT: pinsrw $0, %eax, %xmm2 -; CHECK-SSE-O0-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3] -; CHECK-SSE-O0-NEXT: insertps {{.*#+}} xmm0 = xmm0[0],xmm1[0],zero,zero +; CHECK-SSE-O0-NEXT: movq (%rdi), %xmm0 ; CHECK-SSE-O0-NEXT: retq ; ; CHECK-AVX-O0-LABEL: atomic_vec4_bfloat: ; CHECK-AVX-O0: # %bb.0: -; CHECK-AVX-O0-NEXT: movq (%rdi), %rax -; CHECK-AVX-O0-NEXT: movq %rax, %rcx -; CHECK-AVX-O0-NEXT: shrq $48, %rcx -; CHECK-AVX-O0-NEXT: # kill: def $cx killed $cx killed $rcx -; CHECK-AVX-O0-NEXT: movw %cx, -{{[0-9]+}}(%rsp) -; CHECK-AVX-O0-NEXT: movq %rax, %rcx -; CHECK-AVX-O0-NEXT: shrq $32, %rcx -; CHECK-AVX-O0-NEXT: # kill: def $cx killed $cx killed $rcx -; CHECK-AVX-O0-NEXT: movw %cx, -{{[0-9]+}}(%rsp) -; CHECK-AVX-O0-NEXT: movw %ax, %cx -; CHECK-AVX-O0-NEXT: movw %cx, -{{[0-9]+}}(%rsp) -; CHECK-AVX-O0-NEXT: # kill: def $eax killed $eax killed $rax -; CHECK-AVX-O0-NEXT: shrl $16, %eax -; CHECK-AVX-O0-NEXT: # kill: def $ax killed $ax killed $eax -; CHECK-AVX-O0-NEXT: movw %ax, -{{[0-9]+}}(%rsp) -; CHECK-AVX-O0-NEXT: vmovaps -{{[0-9]+}}(%rsp), %xmm0 +; CHECK-AVX-O0-NEXT: vmovq (%rdi), %xmm0 ; CHECK-AVX-O0-NEXT: retq -; -; CHECK-AVX512-O0-LABEL: atomic_vec4_bfloat: -; CHECK-AVX512-O0: # %bb.0: -; CHECK-AVX512-O0-NEXT: movq (%rdi), %rax -; CHECK-AVX512-O0-NEXT: movq %rax, %rcx -; CHECK-AVX512-O0-NEXT: shrq $48, %rcx -; CHECK-AVX512-O0-NEXT: # kill: def $cx killed $cx killed $rcx -; CHECK-AVX512-O0-NEXT: movw %cx, -{{[0-9]+}}(%rsp) -; CHECK-AVX512-O0-NEXT: movq %rax, %rcx -; CHECK-AVX512-O0-NEXT: shrq $32, %rcx -; CHECK-AVX512-O0-NEXT: # kill: def $cx killed $cx killed $rcx -; CHECK-AVX512-O0-NEXT: movw %cx, -{{[0-9]+}}(%rsp) -; CHECK-AVX512-O0-NEXT: movw %ax, %cx -; CHECK-AVX512-O0-NEXT: movw %cx, -{{[0-9]+}}(%rsp) -; CHECK-AVX512-O0-NEXT: # kill: def $eax killed $eax killed $rax -; CHECK-AVX512-O0-NEXT: shrl $16, %eax -; CHECK-AVX512-O0-NEXT: # kill: def $ax killed $ax killed $eax -; CHECK-AVX512-O0-NEXT: movw %ax, -{{[0-9]+}}(%rsp) -; CHECK-AVX512-O0-NEXT: vmovaps -{{[0-9]+}}(%rsp), %xmm0 -; CHECK-AVX512-O0-NEXT: retq %ret = load atomic <4 x bfloat>, ptr %x acquire, align 8 ret <4 x bfloat> %ret } @@ -1110,6 +726,82 @@ define <4 x float> @atomic_vec4_float(ptr %x) nounwind { ret <4 x float> %ret } +define <4 x float> @atomic_vec4_float_align(ptr %x) nounwind { +; CHECK-SSE1-O-LABEL: atomic_vec4_float_align: +; CHECK-SSE1-O: # %bb.0: +; CHECK-SSE1-O-NEXT: pushq %rax +; CHECK-SSE1-O-NEXT: movl $2, %esi +; CHECK-SSE1-O-NEXT: callq __atomic_load_16@PLT +; CHECK-SSE1-O-NEXT: movq %rdx, %xmm1 +; CHECK-SSE1-O-NEXT: movq %rax, %xmm0 +; CHECK-SSE1-O-NEXT: punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0] +; CHECK-SSE1-O-NEXT: popq %rax +; CHECK-SSE1-O-NEXT: retq +; +; CHECK-SSE1-O3-LABEL: atomic_vec4_float_align: +; CHECK-SSE1-O3: # %bb.0: +; CHECK-SSE1-O3-NEXT: pushq %rax +; CHECK-SSE1-O3-NEXT: movl $2, %esi +; CHECK-SSE1-O3-NEXT: callq __atomic_load_16@PLT +; CHECK-SSE1-O3-NEXT: movq %rdx, %xmm1 +; CHECK-SSE1-O3-NEXT: movq %rax, %xmm0 +; CHECK-SSE1-O3-NEXT: punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0] +; CHECK-SSE1-O3-NEXT: popq %rax +; CHECK-SSE1-O3-NEXT: retq +; +; CHECK-SSE2-O3-LABEL: atomic_vec4_float_align: +; CHECK-SSE2-O3: # %bb.0: +; CHECK-SSE2-O3-NEXT: pushq %rbx +; CHECK-SSE2-O3-NEXT: xorl %eax, %eax +; CHECK-SSE2-O3-NEXT: xorl %edx, %edx +; CHECK-SSE2-O3-NEXT: xorl %ecx, %ecx +; CHECK-SSE2-O3-NEXT: xorl %ebx, %ebx +; CHECK-SSE2-O3-NEXT: lock cmpxchg16b (%rdi) +; CHECK-SSE2-O3-NEXT: movq %rdx, %xmm1 +; CHECK-SSE2-O3-NEXT: movq %rax, %xmm0 +; CHECK-SSE2-O3-NEXT: punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0] +; CHECK-SSE2-O3-NEXT: popq %rbx +; CHECK-SSE2-O3-NEXT: retq +; +; CHECK-AVX-O3-LABEL: atomic_vec4_float_align: +; CHECK-AVX-O3: # %bb.0: +; CHECK-AVX-O3-NEXT: vmovaps (%rdi), %xmm0 +; CHECK-AVX-O3-NEXT: retq +; +; CHECK-SSE1-O0-LABEL: atomic_vec4_float_align: +; CHECK-SSE1-O0: # %bb.0: +; CHECK-SSE1-O0-NEXT: pushq %rax +; CHECK-SSE1-O0-NEXT: movl $2, %esi +; CHECK-SSE1-O0-NEXT: callq __atomic_load_16@PLT +; CHECK-SSE1-O0-NEXT: movq %rdx, %xmm1 +; CHECK-SSE1-O0-NEXT: movq %rax, %xmm0 +; CHECK-SSE1-O0-NEXT: punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0] +; CHECK-SSE1-O0-NEXT: popq %rax +; CHECK-SSE1-O0-NEXT: retq +; +; CHECK-SSE2-O0-LABEL: atomic_vec4_float_align: +; CHECK-SSE2-O0: # %bb.0: +; CHECK-SSE2-O0-NEXT: pushq %rbx +; CHECK-SSE2-O0-NEXT: xorl %eax, %eax +; CHECK-SSE2-O0-NEXT: movl %eax, %ebx +; CHECK-SSE2-O0-NEXT: movq %rbx, %rax +; CHECK-SSE2-O0-NEXT: movq %rbx, %rdx +; CHECK-SSE2-O0-NEXT: movq %rbx, %rcx +; CHECK-SSE2-O0-NEXT: lock cmpxchg16b (%rdi) +; CHECK-SSE2-O0-NEXT: movq %rdx, %xmm1 +; CHECK-SSE2-O0-NEXT: movq %rax, %xmm0 +; CHECK-SSE2-O0-NEXT: punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0] +; CHECK-SSE2-O0-NEXT: popq %rbx +; CHECK-SSE2-O0-NEXT: retq +; +; CHECK-AVX-O0-LABEL: atomic_vec4_float_align: +; CHECK-AVX-O0: # %bb.0: +; CHECK-AVX-O0-NEXT: vmovaps (%rdi), %xmm0 +; CHECK-AVX-O0-NEXT: retq + %ret = load atomic <4 x float>, ptr %x acquire, align 16 + ret <4 x float> %ret +} + define <8 x double> @atomic_vec8_double(ptr %x) nounwind { ; CHECK-SSE-O3-LABEL: atomic_vec8_double: ; CHECK-SSE-O3: # %bb.0: _______________________________________________ llvm-branch-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
