replica
/
hanchenye-llvm-project
2
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

[SelectionDAG] Simplify extract_subvector( insert_subvector ( Vec, In, Idx ), Idx ) -> In 

If we are extracting a subvector that has just been inserted then we should just use the original inserted subvector.

This has come up in certain several x86 shuffle lowering cases where we are crossing 128-bit lanes.

Differential Revision: https://reviews.llvm.org/D24254

llvm-svn: 280715
This commit is contained in:
Simon Pilgrim 2016-09-06 16:42:05 +00:00
parent 6e2e0e8b8f
commit 1b4462b7c1
2 changed files with 10 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp
View File

@ -3750,6 +3750,12 @@ SDValue SelectionDAG::getNode(unsigned Opcode, const SDLoc &DL, EVT VT,
// Trivial extraction.
if (VT.getSimpleVT() == N1.getSimpleValueType())
return N1;
// EXTRACT_SUBVECTOR of INSERT_SUBVECTOR is often created
// during shuffle legalization.
if (N1.getOpcode() == ISD::INSERT_SUBVECTOR && N2 == N1.getOperand(2) &&
VT == N1.getOperand(1).getValueType())
return N1.getOperand(1);
}
break;
}

13
llvm/test/CodeGen/X86/vector-shuffle-256-v16.ll
View File

@ -3500,8 +3500,7 @@ define <16 x i16> @insert_dup_mem_v16i16_i32(i32* %ptr) {
; AVX1-LABEL: insert_dup_mem_v16i16_i32:
; AVX1: # BB#0:
; AVX1-NEXT: vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,0,0,0,4,5,6,7]
; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,0,1,1]
; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
; AVX1-NEXT: retq
;
@ -3521,8 +3520,7 @@ define <16 x i16> @insert_dup_mem_v16i16_sext_i16(i16* %ptr) {
; AVX1: # BB#0:
; AVX1-NEXT: movswl (%rdi), %eax
; AVX1-NEXT: vmovd %eax, %xmm0
; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,0,0,0,4,5,6,7]
; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,0,1,1]
; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
; AVX1-NEXT: retq
;
@ -3544,8 +3542,7 @@ define <16 x i16> @insert_dup_elt1_mem_v16i16_i32(i32* %ptr) #0 {
; AVX1-LABEL: insert_dup_elt1_mem_v16i16_i32:
; AVX1: # BB#0:
; AVX1-NEXT: vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[1,1,1,1,4,5,6,7]
; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,0,1,1]
; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[2,3,2,3,2,3,2,3,2,3,2,3,2,3,2,3]
; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
; AVX1-NEXT: retq
;
@ -3564,9 +3561,7 @@ define <16 x i16> @insert_dup_elt3_mem_v16i16_i32(i32* %ptr) #0 {
; AVX1-LABEL: insert_dup_elt3_mem_v16i16_i32:
; AVX1: # BB#0:
; AVX1-NEXT: vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,0,1,1]
; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[3,3,3,3,4,5,6,7]
; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,0,1,1]
; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[2,3,2,3,2,3,2,3,2,3,2,3,2,3,2,3]
; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
; AVX1-NEXT: retq
;