Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[DAG] Add select_cc -> abd folds #102137

Merged
merged 2 commits into from
Aug 21, 2024
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 2 additions & 0 deletions llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -27919,6 +27919,8 @@ SDValue DAGCombiner::SimplifySelectCC(const SDLoc &DL, SDValue N0, SDValue N1,
return S;
if (SDValue S = PerformUMinFpToSatCombine(N0, N1, N2, N3, CC, DAG))
return S;
if (SDValue ABD = foldSelectToABD(N0, N1, N2, N3, CC, DL))
return ABD;

return SDValue();
}
Expand Down
17 changes: 8 additions & 9 deletions llvm/test/CodeGen/AArch64/abds-neg.ll
Original file line number Diff line number Diff line change
Expand Up @@ -377,7 +377,7 @@ define i64 @abd_cmp_i64(i64 %a, i64 %b) nounwind {
; CHECK: // %bb.0:
; CHECK-NEXT: sub x8, x1, x0
; CHECK-NEXT: subs x9, x0, x1
; CHECK-NEXT: csel x0, x8, x9, lt
; CHECK-NEXT: csel x0, x9, x8, gt
; CHECK-NEXT: ret
%cmp = icmp slt i64 %a, %b
%ab = sub i64 %a, %b
Expand All @@ -389,14 +389,13 @@ define i64 @abd_cmp_i64(i64 %a, i64 %b) nounwind {
define i128 @abd_cmp_i128(i128 %a, i128 %b) nounwind {
; CHECK-LABEL: abd_cmp_i128:
; CHECK: // %bb.0:
; CHECK-NEXT: cmp x0, x2
; CHECK-NEXT: sbc x8, x1, x3
; CHECK-NEXT: subs x9, x2, x0
; CHECK-NEXT: sbc x10, x3, x1
; CHECK-NEXT: subs x11, x0, x2
; CHECK-NEXT: sbcs xzr, x1, x3
; CHECK-NEXT: csel x0, x9, x11, lt
; CHECK-NEXT: csel x1, x10, x8, lt
; CHECK-NEXT: subs x8, x0, x2
; CHECK-NEXT: sbc x9, x1, x3
; CHECK-NEXT: subs x10, x2, x0
; CHECK-NEXT: sbc x11, x3, x1
; CHECK-NEXT: sbcs xzr, x3, x1
; CHECK-NEXT: csel x0, x8, x10, lt
; CHECK-NEXT: csel x1, x9, x11, lt
; CHECK-NEXT: ret
%cmp = icmp slt i128 %a, %b
%ab = sub i128 %a, %b
Expand Down
16 changes: 7 additions & 9 deletions llvm/test/CodeGen/AArch64/abds.ll
Original file line number Diff line number Diff line change
Expand Up @@ -298,10 +298,9 @@ define i8 @abd_cmp_i8(i8 %a, i8 %b) nounwind {
; CHECK-LABEL: abd_cmp_i8:
; CHECK: // %bb.0:
; CHECK-NEXT: sxtb w8, w0
; CHECK-NEXT: sub w9, w0, w1
; CHECK-NEXT: sub w10, w1, w0
; CHECK-NEXT: cmp w8, w1, sxtb
; CHECK-NEXT: csel w0, w9, w10, gt
; CHECK-NEXT: sub w8, w8, w1, sxtb
; CHECK-NEXT: cmp w8, #0
; CHECK-NEXT: cneg w0, w8, mi
; CHECK-NEXT: ret
%cmp = icmp sgt i8 %a, %b
%ab = sub i8 %a, %b
Expand All @@ -314,10 +313,9 @@ define i16 @abd_cmp_i16(i16 %a, i16 %b) nounwind {
; CHECK-LABEL: abd_cmp_i16:
; CHECK: // %bb.0:
; CHECK-NEXT: sxth w8, w0
; CHECK-NEXT: sub w9, w0, w1
; CHECK-NEXT: sub w10, w1, w0
; CHECK-NEXT: cmp w8, w1, sxth
; CHECK-NEXT: csel w0, w9, w10, ge
; CHECK-NEXT: sub w8, w8, w1, sxth
; CHECK-NEXT: cmp w8, #0
; CHECK-NEXT: cneg w0, w8, mi
; CHECK-NEXT: ret
%cmp = icmp sge i16 %a, %b
%ab = sub i16 %a, %b
Expand All @@ -331,7 +329,7 @@ define i32 @abd_cmp_i32(i32 %a, i32 %b) nounwind {
; CHECK: // %bb.0:
; CHECK-NEXT: sub w8, w1, w0
; CHECK-NEXT: subs w9, w0, w1
; CHECK-NEXT: csel w0, w8, w9, lt
; CHECK-NEXT: csel w0, w9, w8, gt
; CHECK-NEXT: ret
%cmp = icmp slt i32 %a, %b
%ab = sub i32 %a, %b
Expand Down
18 changes: 9 additions & 9 deletions llvm/test/CodeGen/AArch64/abdu-neg.ll
Original file line number Diff line number Diff line change
Expand Up @@ -379,7 +379,7 @@ define i64 @abd_cmp_i64(i64 %a, i64 %b) nounwind {
; CHECK: // %bb.0:
; CHECK-NEXT: sub x8, x1, x0
; CHECK-NEXT: subs x9, x0, x1
; CHECK-NEXT: csel x0, x8, x9, lo
; CHECK-NEXT: csel x0, x9, x8, hi
; CHECK-NEXT: ret
%cmp = icmp ult i64 %a, %b
%ab = sub i64 %a, %b
Expand All @@ -391,14 +391,14 @@ define i64 @abd_cmp_i64(i64 %a, i64 %b) nounwind {
define i128 @abd_cmp_i128(i128 %a, i128 %b) nounwind {
; CHECK-LABEL: abd_cmp_i128:
; CHECK: // %bb.0:
; CHECK-NEXT: cmp x0, x2
; CHECK-NEXT: sbc x8, x1, x3
; CHECK-NEXT: subs x9, x2, x0
; CHECK-NEXT: sbc x10, x3, x1
; CHECK-NEXT: subs x11, x0, x2
; CHECK-NEXT: sbcs xzr, x1, x3
; CHECK-NEXT: csel x0, x9, x11, lo
; CHECK-NEXT: csel x1, x10, x8, lo
; CHECK-NEXT: subs x8, x0, x2
; CHECK-NEXT: sbcs x9, x1, x3
; CHECK-NEXT: cset w10, lo
; CHECK-NEXT: sbfx x10, x10, #0, #1
; CHECK-NEXT: eor x8, x8, x10
; CHECK-NEXT: eor x9, x9, x10
; CHECK-NEXT: subs x0, x8, x10
; CHECK-NEXT: sbc x1, x9, x10
; CHECK-NEXT: ret
%cmp = icmp ult i128 %a, %b
%ab = sub i128 %a, %b
Expand Down
16 changes: 7 additions & 9 deletions llvm/test/CodeGen/AArch64/abdu.ll
Original file line number Diff line number Diff line change
Expand Up @@ -301,10 +301,9 @@ define i8 @abd_cmp_i8(i8 %a, i8 %b) nounwind {
; CHECK-LABEL: abd_cmp_i8:
; CHECK: // %bb.0:
; CHECK-NEXT: and w8, w0, #0xff
; CHECK-NEXT: sub w9, w0, w1
; CHECK-NEXT: sub w10, w1, w0
; CHECK-NEXT: cmp w8, w1, uxtb
; CHECK-NEXT: csel w0, w9, w10, hi
; CHECK-NEXT: sub w8, w8, w1, uxtb
; CHECK-NEXT: cmp w8, #0
; CHECK-NEXT: cneg w0, w8, mi
; CHECK-NEXT: ret
%cmp = icmp ugt i8 %a, %b
%ab = sub i8 %a, %b
Expand All @@ -317,10 +316,9 @@ define i16 @abd_cmp_i16(i16 %a, i16 %b) nounwind {
; CHECK-LABEL: abd_cmp_i16:
; CHECK: // %bb.0:
; CHECK-NEXT: and w8, w0, #0xffff
; CHECK-NEXT: sub w9, w0, w1
; CHECK-NEXT: sub w10, w1, w0
; CHECK-NEXT: cmp w8, w1, uxth
; CHECK-NEXT: csel w0, w9, w10, hs
; CHECK-NEXT: sub w8, w8, w1, uxth
; CHECK-NEXT: cmp w8, #0
; CHECK-NEXT: cneg w0, w8, mi
; CHECK-NEXT: ret
%cmp = icmp uge i16 %a, %b
%ab = sub i16 %a, %b
Expand All @@ -334,7 +332,7 @@ define i32 @abd_cmp_i32(i32 %a, i32 %b) nounwind {
; CHECK: // %bb.0:
; CHECK-NEXT: sub w8, w1, w0
; CHECK-NEXT: subs w9, w0, w1
; CHECK-NEXT: csel w0, w8, w9, lo
; CHECK-NEXT: csel w0, w9, w8, hi
; CHECK-NEXT: ret
%cmp = icmp ult i32 %a, %b
%ab = sub i32 %a, %b
Expand Down
21 changes: 8 additions & 13 deletions llvm/test/CodeGen/AMDGPU/sad.ll
Original file line number Diff line number Diff line change
Expand Up @@ -258,10 +258,9 @@ define amdgpu_kernel void @v_sad_u32_multi_use_select_pat2(ptr addrspace(1) %out
; GCN-NEXT: s_add_u32 s16, s16, s13
; GCN-NEXT: s_addc_u32 s17, s17, 0
; GCN-NEXT: s_waitcnt lgkmcnt(0)
; GCN-NEXT: s_sub_i32 s3, s0, s1
; GCN-NEXT: s_sub_i32 s6, s1, s0
; GCN-NEXT: s_cmp_gt_u32 s0, s1
; GCN-NEXT: s_cselect_b32 s0, s3, s6
; GCN-NEXT: s_min_u32 s3, s0, s1
; GCN-NEXT: s_max_u32 s0, s0, s1
; GCN-NEXT: s_sub_i32 s0, s0, s3
; GCN-NEXT: v_mov_b32_e32 v0, s4
; GCN-NEXT: v_mov_b32_e32 v2, s0
; GCN-NEXT: s_add_i32 s0, s0, s2
Expand Down Expand Up @@ -477,18 +476,14 @@ define amdgpu_kernel void @s_sad_u32_i8_pat2(ptr addrspace(1) %out, i8 zeroext %
; GCN-NEXT: s_load_dword s2, s[6:7], 0x2
; GCN-NEXT: s_load_dwordx2 s[0:1], s[6:7], 0x0
; GCN-NEXT: s_waitcnt lgkmcnt(0)
; GCN-NEXT: s_lshr_b32 s4, s2, 8
; GCN-NEXT: s_and_b32 s3, s2, 0xff
; GCN-NEXT: s_bfe_u32 s5, s2, 0x80008
; GCN-NEXT: s_lshr_b32 s6, s2, 16
; GCN-NEXT: s_sub_i32 s7, s2, s4
; GCN-NEXT: s_sub_i32 s2, s4, s2
; GCN-NEXT: s_cmp_gt_u32 s3, s5
; GCN-NEXT: s_cselect_b32 s2, s7, s2
; GCN-NEXT: s_add_i32 s2, s2, s6
; GCN-NEXT: s_bfe_u32 s4, s2, 0x80008
; GCN-NEXT: s_lshr_b32 s2, s2, 16
; GCN-NEXT: v_mov_b32_e32 v0, s4
; GCN-NEXT: v_mov_b32_e32 v1, s2
; GCN-NEXT: v_sad_u32 v2, s3, v0, v1
; GCN-NEXT: v_mov_b32_e32 v0, s0
; GCN-NEXT: v_mov_b32_e32 v1, s1
; GCN-NEXT: v_mov_b32_e32 v2, s2
; GCN-NEXT: flat_store_byte v[0:1], v2
; GCN-NEXT: s_endpgm
%icmp0 = icmp ugt i8 %a, %b
Expand Down
Loading
Loading