[llvm] r325729 - [X86][MMX] Add MMX_MOVD64rr build vector tests showing undef elements in the lower half
Simon Pilgrim via llvm-commits
llvm-commits at lists.llvm.org
Wed Feb 21 14:10:48 PST 2018
Author: rksimon
Date: Wed Feb 21 14:10:48 2018
New Revision: 325729
URL: http://llvm.org/viewvc/llvm-project?rev=325729&view=rev
Log:
[X86][MMX] Add MMX_MOVD64rr build vector tests showing undef elements in the lower half
Modified:
llvm/trunk/test/CodeGen/X86/mmx-build-vector.ll
Modified: llvm/trunk/test/CodeGen/X86/mmx-build-vector.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/mmx-build-vector.ll?rev=325729&r1=325728&r2=325729&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/mmx-build-vector.ll (original)
+++ llvm/trunk/test/CodeGen/X86/mmx-build-vector.ll Wed Feb 21 14:10:48 2018
@@ -543,6 +543,104 @@ define void @build_v4i16_01zz(x86_mmx *%
ret void
}
+define void @build_v4i16_0uuz(x86_mmx *%p0, i16 %a0, i16 %a1, i16 %a2, i16 %a3) nounwind {
+; X86-MMX-LABEL: build_v4i16_0uuz:
+; X86-MMX: # %bb.0:
+; X86-MMX-NEXT: pushl %ebp
+; X86-MMX-NEXT: movl %esp, %ebp
+; X86-MMX-NEXT: andl $-8, %esp
+; X86-MMX-NEXT: subl $8, %esp
+; X86-MMX-NEXT: movl 8(%ebp), %eax
+; X86-MMX-NEXT: movzwl %ax, %ecx
+; X86-MMX-NEXT: movl %ecx, {{[0-9]+}}(%esp)
+; X86-MMX-NEXT: shll $16, %ecx
+; X86-MMX-NEXT: movzwl 12(%ebp), %edx
+; X86-MMX-NEXT: orl %ecx, %edx
+; X86-MMX-NEXT: movl %edx, (%esp)
+; X86-MMX-NEXT: movq (%esp), %mm0
+; X86-MMX-NEXT: paddd %mm0, %mm0
+; X86-MMX-NEXT: movq %mm0, (%eax)
+; X86-MMX-NEXT: movl %ebp, %esp
+; X86-MMX-NEXT: popl %ebp
+; X86-MMX-NEXT: retl
+;
+; X86-SSE2-LABEL: build_v4i16_0uuz:
+; X86-SSE2: # %bb.0:
+; X86-SSE2-NEXT: pushl %ebp
+; X86-SSE2-NEXT: movl %esp, %ebp
+; X86-SSE2-NEXT: andl $-8, %esp
+; X86-SSE2-NEXT: subl $8, %esp
+; X86-SSE2-NEXT: movl 8(%ebp), %eax
+; X86-SSE2-NEXT: pxor %xmm0, %xmm0
+; X86-SSE2-NEXT: movd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; X86-SSE2-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
+; X86-SSE2-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
+; X86-SSE2-NEXT: movq %xmm1, (%esp)
+; X86-SSE2-NEXT: movq (%esp), %mm0
+; X86-SSE2-NEXT: paddd %mm0, %mm0
+; X86-SSE2-NEXT: movq %mm0, (%eax)
+; X86-SSE2-NEXT: movl %ebp, %esp
+; X86-SSE2-NEXT: popl %ebp
+; X86-SSE2-NEXT: retl
+;
+; X86-SSSE3-LABEL: build_v4i16_0uuz:
+; X86-SSSE3: # %bb.0:
+; X86-SSSE3-NEXT: pushl %ebp
+; X86-SSSE3-NEXT: movl %esp, %ebp
+; X86-SSSE3-NEXT: andl $-8, %esp
+; X86-SSSE3-NEXT: subl $8, %esp
+; X86-SSSE3-NEXT: movl 8(%ebp), %eax
+; X86-SSSE3-NEXT: movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; X86-SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[0,1],zero,zero,zero,zero,zero,zero,xmm0[2,3],zero,zero,zero,zero,zero,zero
+; X86-SSSE3-NEXT: movq %xmm0, (%esp)
+; X86-SSSE3-NEXT: movq (%esp), %mm0
+; X86-SSSE3-NEXT: paddd %mm0, %mm0
+; X86-SSSE3-NEXT: movq %mm0, (%eax)
+; X86-SSSE3-NEXT: movl %ebp, %esp
+; X86-SSSE3-NEXT: popl %ebp
+; X86-SSSE3-NEXT: retl
+;
+; X64-SSE2-LABEL: build_v4i16_0uuz:
+; X64-SSE2: # %bb.0:
+; X64-SSE2-NEXT: movd %esi, %xmm0
+; X64-SSE2-NEXT: pxor %xmm1, %xmm1
+; X64-SSE2-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
+; X64-SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; X64-SSE2-NEXT: movq %xmm0, -{{[0-9]+}}(%rsp)
+; X64-SSE2-NEXT: movq -{{[0-9]+}}(%rsp), %mm0
+; X64-SSE2-NEXT: paddd %mm0, %mm0
+; X64-SSE2-NEXT: movq %mm0, (%rdi)
+; X64-SSE2-NEXT: retq
+;
+; X64-SSSE3-LABEL: build_v4i16_0uuz:
+; X64-SSSE3: # %bb.0:
+; X64-SSSE3-NEXT: movd %esi, %xmm0
+; X64-SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[0,1],zero,zero,zero,zero,zero,zero,xmm0[2,3],zero,zero,zero,zero,zero,zero
+; X64-SSSE3-NEXT: movq %xmm0, -{{[0-9]+}}(%rsp)
+; X64-SSSE3-NEXT: movq -{{[0-9]+}}(%rsp), %mm0
+; X64-SSSE3-NEXT: paddd %mm0, %mm0
+; X64-SSSE3-NEXT: movq %mm0, (%rdi)
+; X64-SSSE3-NEXT: retq
+;
+; X64-AVX-LABEL: build_v4i16_0uuz:
+; X64-AVX: # %bb.0:
+; X64-AVX-NEXT: vmovd %esi, %xmm0
+; X64-AVX-NEXT: vpmovzxwq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero
+; X64-AVX-NEXT: vmovq %xmm0, -{{[0-9]+}}(%rsp)
+; X64-AVX-NEXT: movq -{{[0-9]+}}(%rsp), %mm0
+; X64-AVX-NEXT: paddd %mm0, %mm0
+; X64-AVX-NEXT: movq %mm0, (%rdi)
+; X64-AVX-NEXT: retq
+ %1 = insertelement <4 x i16> undef, i16 %a0, i32 0
+ %2 = insertelement <4 x i16> %1, i16 undef, i32 1
+ %3 = insertelement <4 x i16> %2, i16 undef, i32 2
+ %4 = insertelement <4 x i16> %3, i16 0, i32 3
+ %5 = bitcast <4 x i16> %4 to x86_mmx
+ %6 = tail call x86_mmx @llvm.x86.mmx.padd.d(x86_mmx %5, x86_mmx %5)
+ store x86_mmx %6, x86_mmx *%p0
+ ret void
+}
+
define void @build_v4i16_0zuz(x86_mmx *%p0, i16 %a0, i16 %a1, i16 %a2, i16 %a3) nounwind {
; X86-MMX-LABEL: build_v4i16_0zuz:
; X86-MMX: # %bb.0:
@@ -1292,6 +1390,112 @@ define void @build_v8i8_0123zzzu(x86_mmx
%9 = bitcast <8 x i8> %8 to x86_mmx
%10 = tail call x86_mmx @llvm.x86.mmx.padd.d(x86_mmx %9, x86_mmx %9)
store x86_mmx %10, x86_mmx *%p0
+ ret void
+}
+
+define void @build_v8i8_0uuuuzzz(x86_mmx *%p0, i8 %a0, i8 %a1, i8 %a2, i8 %a3, i8 %a4, i8 %a5, i8 %a6, i8 %a7) nounwind {
+; X86-MMX-LABEL: build_v8i8_0uuuuzzz:
+; X86-MMX: # %bb.0:
+; X86-MMX-NEXT: pushl %ebp
+; X86-MMX-NEXT: movl %esp, %ebp
+; X86-MMX-NEXT: andl $-8, %esp
+; X86-MMX-NEXT: subl $8, %esp
+; X86-MMX-NEXT: movl 8(%ebp), %eax
+; X86-MMX-NEXT: movzbl 12(%ebp), %ecx
+; X86-MMX-NEXT: movl %ecx, (%esp)
+; X86-MMX-NEXT: movl $0, {{[0-9]+}}(%esp)
+; X86-MMX-NEXT: movq (%esp), %mm0
+; X86-MMX-NEXT: paddd %mm0, %mm0
+; X86-MMX-NEXT: movq %mm0, (%eax)
+; X86-MMX-NEXT: movl %ebp, %esp
+; X86-MMX-NEXT: popl %ebp
+; X86-MMX-NEXT: retl
+;
+; X86-SSE2-LABEL: build_v8i8_0uuuuzzz:
+; X86-SSE2: # %bb.0:
+; X86-SSE2-NEXT: pushl %ebp
+; X86-SSE2-NEXT: movl %esp, %ebp
+; X86-SSE2-NEXT: andl $-8, %esp
+; X86-SSE2-NEXT: subl $8, %esp
+; X86-SSE2-NEXT: movl 8(%ebp), %eax
+; X86-SSE2-NEXT: movzbl 12(%ebp), %ecx
+; X86-SSE2-NEXT: movd %ecx, %xmm0
+; X86-SSE2-NEXT: pxor %xmm1, %xmm1
+; X86-SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
+; X86-SSE2-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
+; X86-SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; X86-SSE2-NEXT: movq %xmm0, (%esp)
+; X86-SSE2-NEXT: movq (%esp), %mm0
+; X86-SSE2-NEXT: paddd %mm0, %mm0
+; X86-SSE2-NEXT: movq %mm0, (%eax)
+; X86-SSE2-NEXT: movl %ebp, %esp
+; X86-SSE2-NEXT: popl %ebp
+; X86-SSE2-NEXT: retl
+;
+; X86-SSSE3-LABEL: build_v8i8_0uuuuzzz:
+; X86-SSSE3: # %bb.0:
+; X86-SSSE3-NEXT: pushl %ebp
+; X86-SSSE3-NEXT: movl %esp, %ebp
+; X86-SSSE3-NEXT: andl $-8, %esp
+; X86-SSSE3-NEXT: subl $8, %esp
+; X86-SSSE3-NEXT: movl 8(%ebp), %eax
+; X86-SSSE3-NEXT: movzbl 12(%ebp), %ecx
+; X86-SSSE3-NEXT: movd %ecx, %xmm0
+; X86-SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero
+; X86-SSSE3-NEXT: movq %xmm0, (%esp)
+; X86-SSSE3-NEXT: movq (%esp), %mm0
+; X86-SSSE3-NEXT: paddd %mm0, %mm0
+; X86-SSSE3-NEXT: movq %mm0, (%eax)
+; X86-SSSE3-NEXT: movl %ebp, %esp
+; X86-SSSE3-NEXT: popl %ebp
+; X86-SSSE3-NEXT: retl
+;
+; X64-SSE2-LABEL: build_v8i8_0uuuuzzz:
+; X64-SSE2: # %bb.0:
+; X64-SSE2-NEXT: movzwl %si, %eax
+; X64-SSE2-NEXT: movd %eax, %xmm0
+; X64-SSE2-NEXT: pxor %xmm1, %xmm1
+; X64-SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
+; X64-SSE2-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
+; X64-SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; X64-SSE2-NEXT: movq %xmm0, -{{[0-9]+}}(%rsp)
+; X64-SSE2-NEXT: movq -{{[0-9]+}}(%rsp), %mm0
+; X64-SSE2-NEXT: paddd %mm0, %mm0
+; X64-SSE2-NEXT: movq %mm0, (%rdi)
+; X64-SSE2-NEXT: retq
+;
+; X64-SSSE3-LABEL: build_v8i8_0uuuuzzz:
+; X64-SSSE3: # %bb.0:
+; X64-SSSE3-NEXT: movzwl %si, %eax
+; X64-SSSE3-NEXT: movd %eax, %xmm0
+; X64-SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero
+; X64-SSSE3-NEXT: movq %xmm0, -{{[0-9]+}}(%rsp)
+; X64-SSSE3-NEXT: movq -{{[0-9]+}}(%rsp), %mm0
+; X64-SSSE3-NEXT: paddd %mm0, %mm0
+; X64-SSSE3-NEXT: movq %mm0, (%rdi)
+; X64-SSSE3-NEXT: retq
+;
+; X64-AVX-LABEL: build_v8i8_0uuuuzzz:
+; X64-AVX: # %bb.0:
+; X64-AVX-NEXT: movzwl %si, %eax
+; X64-AVX-NEXT: vmovd %eax, %xmm0
+; X64-AVX-NEXT: vpmovzxbq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero
+; X64-AVX-NEXT: vmovq %xmm0, -{{[0-9]+}}(%rsp)
+; X64-AVX-NEXT: movq -{{[0-9]+}}(%rsp), %mm0
+; X64-AVX-NEXT: paddd %mm0, %mm0
+; X64-AVX-NEXT: movq %mm0, (%rdi)
+; X64-AVX-NEXT: retq
+ %1 = insertelement <8 x i8> undef, i8 %a0, i32 0
+ %2 = insertelement <8 x i8> %1, i8 undef, i32 1
+ %3 = insertelement <8 x i8> %2, i8 undef, i32 2
+ %4 = insertelement <8 x i8> %3, i8 undef, i32 3
+ %5 = insertelement <8 x i8> %4, i8 undef, i32 4
+ %6 = insertelement <8 x i8> %5, i8 0, i32 5
+ %7 = insertelement <8 x i8> %6, i8 0, i32 6
+ %8 = insertelement <8 x i8> %7, i8 0, i32 7
+ %9 = bitcast <8 x i8> %8 to x86_mmx
+ %10 = tail call x86_mmx @llvm.x86.mmx.padd.d(x86_mmx %9, x86_mmx %9)
+ store x86_mmx %10, x86_mmx *%p0
ret void
}
More information about the llvm-commits
mailing list