Skip to content

[RISCV] Don't use EVL/Mask for vid when lowering vp.reverse #123048

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.

Sign up for GitHub

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

Closed
wants to merge 2 commits into from
Closed

[RISCV] Don't use EVL/Mask for vid when lowering vp.reverse #123048

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
f9d4785
Precommit test
lukel97 Jan 15, 2025
275076d
[RISCV] Don't use EVL/Mask for vid when lowering vp.reverse
lukel97 Jan 15, 2025
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
6 changes: 5 additions & 1 deletion llvm/lib/Target/RISCV/RISCVISelLowering.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -12430,7 +12430,11 @@ RISCVTargetLowering::lowerVPReverseExperimental(SDValue Op,
GatherOpc = RISCVISD::VRGATHEREI16_VV_VL;
}

SDValue VID = DAG.getNode(RISCVISD::VID_VL, DL, IndicesVT, Mask, EVL);
// Don't use EVL or Mask for vid so it can be hoisted out of loops.
auto [TrueMask, VLMAX] =
getDefaultScalableVLOps(IndicesVT, DL, DAG, Subtarget);
SDValue VID = DAG.getNode(RISCVISD::VID_VL, DL, IndicesVT, TrueMask, VLMAX);

SDValue VecLen =
DAG.getNode(ISD::SUB, DL, XLenVT, EVL, DAG.getConstant(1, DL, XLenVT));
SDValue VecLenSplat = DAG.getNode(RISCVISD::VMV_V_X_VL, DL, IndicesVT,
Expand Down
12 changes: 6 additions & 6 deletions llvm/test/CodeGen/RISCV/rvv/vp-reverse-float-fixed-vectors.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5,10 +5,10 @@
define <2 x double> @test_vp_reverse_v2f64_masked(<2 x double> %src, <2 x i1> %mask, i32 zeroext %evl) {
; CHECK-LABEL: test_vp_reverse_v2f64_masked:
; CHECK: # %bb.0:
; CHECK-NEXT: addi a1, a0, -1
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vid.v v9, v0.t
; CHECK-NEXT: addi a0, a0, -1
; CHECK-NEXT: vrsub.vx v10, v9, a0, v0.t
; CHECK-NEXT: vid.v v9
; CHECK-NEXT: vrsub.vx v10, v9, a1, v0.t
; CHECK-NEXT: vrgather.vv v9, v8, v10, v0.t
; CHECK-NEXT: vmv.v.v v8, v9
; CHECK-NEXT: ret
Expand All @@ -34,10 +34,10 @@ define <2 x double> @test_vp_reverse_v2f64(<2 x double> %src, i32 zeroext %evl)
define <4 x float> @test_vp_reverse_v4f32_masked(<4 x float> %src, <4 x i1> %mask, i32 zeroext %evl) {
; CHECK-LABEL: test_vp_reverse_v4f32_masked:
; CHECK: # %bb.0:
; CHECK-NEXT: addi a1, a0, -1
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vid.v v9, v0.t
; CHECK-NEXT: addi a0, a0, -1
; CHECK-NEXT: vrsub.vx v10, v9, a0, v0.t
; CHECK-NEXT: vid.v v9
; CHECK-NEXT: vrsub.vx v10, v9, a1, v0.t
; CHECK-NEXT: vrgather.vv v9, v8, v10, v0.t
; CHECK-NEXT: vmv.v.v v8, v9
; CHECK-NEXT: ret
Expand Down
48 changes: 24 additions & 24 deletions llvm/test/CodeGen/RISCV/rvv/vp-reverse-float.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,10 +4,10 @@
define <vscale x 1 x double> @test_vp_reverse_nxv1f64_masked(<vscale x 1 x double> %src, <vscale x 1 x i1> %mask, i32 zeroext %evl) {
; CHECK-LABEL: test_vp_reverse_nxv1f64_masked:
; CHECK: # %bb.0:
; CHECK-NEXT: addi a1, a0, -1
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vid.v v9, v0.t
; CHECK-NEXT: addi a0, a0, -1
; CHECK-NEXT: vrsub.vx v10, v9, a0, v0.t
; CHECK-NEXT: vid.v v9
; CHECK-NEXT: vrsub.vx v10, v9, a1, v0.t
; CHECK-NEXT: vrgather.vv v9, v8, v10, v0.t
; CHECK-NEXT: vmv.v.v v8, v9
; CHECK-NEXT: ret
Expand All @@ -33,10 +33,10 @@ define <vscale x 1 x double> @test_vp_reverse_nxv1f64(<vscale x 1 x double> %src
define <vscale x 2 x float> @test_vp_reverse_nxv2f32_masked(<vscale x 2 x float> %src, <vscale x 2 x i1> %mask, i32 zeroext %evl) {
; CHECK-LABEL: test_vp_reverse_nxv2f32_masked:
; CHECK: # %bb.0:
; CHECK-NEXT: addi a1, a0, -1
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vid.v v9, v0.t
; CHECK-NEXT: addi a0, a0, -1
; CHECK-NEXT: vrsub.vx v10, v9, a0, v0.t
; CHECK-NEXT: vid.v v9
; CHECK-NEXT: vrsub.vx v10, v9, a1, v0.t
; CHECK-NEXT: vrgather.vv v9, v8, v10, v0.t
; CHECK-NEXT: vmv.v.v v8, v9
; CHECK-NEXT: ret
Expand All @@ -62,10 +62,10 @@ define <vscale x 2 x float> @test_vp_reverse_nxv2f32(<vscale x 2 x float> %src,
define <vscale x 2 x double> @test_vp_reverse_nxv2f64_masked(<vscale x 2 x double> %src, <vscale x 2 x i1> %mask, i32 zeroext %evl) {
; CHECK-LABEL: test_vp_reverse_nxv2f64_masked:
; CHECK: # %bb.0:
; CHECK-NEXT: addi a1, a0, -1
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vid.v v10, v0.t
; CHECK-NEXT: addi a0, a0, -1
; CHECK-NEXT: vrsub.vx v12, v10, a0, v0.t
; CHECK-NEXT: vid.v v10
; CHECK-NEXT: vrsub.vx v12, v10, a1, v0.t
; CHECK-NEXT: vrgather.vv v10, v8, v12, v0.t
; CHECK-NEXT: vmv.v.v v8, v10
; CHECK-NEXT: ret
Expand All @@ -91,10 +91,10 @@ define <vscale x 2 x double> @test_vp_reverse_nxv2f64(<vscale x 2 x double> %src
define <vscale x 4 x float> @test_vp_reverse_nxv4f32_masked(<vscale x 4 x float> %src, <vscale x 4 x i1> %mask, i32 zeroext %evl) {
; CHECK-LABEL: test_vp_reverse_nxv4f32_masked:
; CHECK: # %bb.0:
; CHECK-NEXT: addi a1, a0, -1
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vid.v v10, v0.t
; CHECK-NEXT: addi a0, a0, -1
; CHECK-NEXT: vrsub.vx v12, v10, a0, v0.t
; CHECK-NEXT: vid.v v10
; CHECK-NEXT: vrsub.vx v12, v10, a1, v0.t
; CHECK-NEXT: vrgather.vv v10, v8, v12, v0.t
; CHECK-NEXT: vmv.v.v v8, v10
; CHECK-NEXT: ret
Expand All @@ -120,10 +120,10 @@ define <vscale x 4 x float> @test_vp_reverse_nxv4f32(<vscale x 4 x float> %src,
define <vscale x 4 x double> @test_vp_reverse_nxv4f64_masked(<vscale x 4 x double> %src, <vscale x 4 x i1> %mask, i32 zeroext %evl) {
; CHECK-LABEL: test_vp_reverse_nxv4f64_masked:
; CHECK: # %bb.0:
; CHECK-NEXT: addi a1, a0, -1
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vid.v v12, v0.t
; CHECK-NEXT: addi a0, a0, -1
; CHECK-NEXT: vrsub.vx v16, v12, a0, v0.t
; CHECK-NEXT: vid.v v12
; CHECK-NEXT: vrsub.vx v16, v12, a1, v0.t
; CHECK-NEXT: vrgather.vv v12, v8, v16, v0.t
; CHECK-NEXT: vmv.v.v v8, v12
; CHECK-NEXT: ret
Expand All @@ -149,10 +149,10 @@ define <vscale x 4 x double> @test_vp_reverse_nxv4f64(<vscale x 4 x double> %src
define <vscale x 8 x float> @test_vp_reverse_nxv8f32_masked(<vscale x 8 x float> %src, <vscale x 8 x i1> %mask, i32 zeroext %evl) {
; CHECK-LABEL: test_vp_reverse_nxv8f32_masked:
; CHECK: # %bb.0:
; CHECK-NEXT: addi a1, a0, -1
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vid.v v12, v0.t
; CHECK-NEXT: addi a0, a0, -1
; CHECK-NEXT: vrsub.vx v16, v12, a0, v0.t
; CHECK-NEXT: vid.v v12
; CHECK-NEXT: vrsub.vx v16, v12, a1, v0.t
; CHECK-NEXT: vrgather.vv v12, v8, v16, v0.t
; CHECK-NEXT: vmv.v.v v8, v12
; CHECK-NEXT: ret
Expand All @@ -178,10 +178,10 @@ define <vscale x 8 x float> @test_vp_reverse_nxv8f32(<vscale x 8 x float> %src,
define <vscale x 8 x double> @test_vp_reverse_nxv8f64_masked(<vscale x 8 x double> %src, <vscale x 8 x i1> %mask, i32 zeroext %evl) {
; CHECK-LABEL: test_vp_reverse_nxv8f64_masked:
; CHECK: # %bb.0:
; CHECK-NEXT: addi a1, a0, -1
; CHECK-NEXT: vsetvli zero, a0, e64, m8, ta, ma
; CHECK-NEXT: vid.v v16, v0.t
; CHECK-NEXT: addi a0, a0, -1
; CHECK-NEXT: vrsub.vx v24, v16, a0, v0.t
; CHECK-NEXT: vid.v v16
; CHECK-NEXT: vrsub.vx v24, v16, a1, v0.t
; CHECK-NEXT: vrgather.vv v16, v8, v24, v0.t
; CHECK-NEXT: vmv.v.v v8, v16
; CHECK-NEXT: ret
Expand All @@ -207,10 +207,10 @@ define <vscale x 8 x double> @test_vp_reverse_nxv8f64(<vscale x 8 x double> %src
define <vscale x 16 x float> @test_vp_reverse_nxv16f32_masked(<vscale x 16 x float> %src, <vscale x 16 x i1> %mask, i32 zeroext %evl) {
; CHECK-LABEL: test_vp_reverse_nxv16f32_masked:
; CHECK: # %bb.0:
; CHECK-NEXT: addi a1, a0, -1
; CHECK-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; CHECK-NEXT: vid.v v16, v0.t
; CHECK-NEXT: addi a0, a0, -1
; CHECK-NEXT: vrsub.vx v24, v16, a0, v0.t
; CHECK-NEXT: vid.v v16
; CHECK-NEXT: vrsub.vx v24, v16, a1, v0.t
; CHECK-NEXT: vrgather.vv v16, v8, v24, v0.t
; CHECK-NEXT: vmv.v.v v8, v16
; CHECK-NEXT: ret
Expand Down
24 changes: 12 additions & 12 deletions llvm/test/CodeGen/RISCV/rvv/vp-reverse-int-fixed-vectors.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5,10 +5,10 @@
define <2 x i64> @test_vp_reverse_v2i64_masked(<2 x i64> %src, <2 x i1> %mask, i32 zeroext %evl) {
; CHECK-LABEL: test_vp_reverse_v2i64_masked:
; CHECK: # %bb.0:
; CHECK-NEXT: addi a1, a0, -1
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vid.v v9, v0.t
; CHECK-NEXT: addi a0, a0, -1
; CHECK-NEXT: vrsub.vx v10, v9, a0, v0.t
; CHECK-NEXT: vid.v v9
; CHECK-NEXT: vrsub.vx v10, v9, a1, v0.t
; CHECK-NEXT: vrgather.vv v9, v8, v10, v0.t
; CHECK-NEXT: vmv.v.v v8, v9
; CHECK-NEXT: ret
Expand All @@ -34,10 +34,10 @@ define <2 x i64> @test_vp_reverse_v2i64(<2 x i64> %src, i32 zeroext %evl) {
define <4 x i32> @test_vp_reverse_v4i32_masked(<4 x i32> %src, <4 x i1> %mask, i32 zeroext %evl) {
; CHECK-LABEL: test_vp_reverse_v4i32_masked:
; CHECK: # %bb.0:
; CHECK-NEXT: addi a1, a0, -1
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vid.v v9, v0.t
; CHECK-NEXT: addi a0, a0, -1
; CHECK-NEXT: vrsub.vx v10, v9, a0, v0.t
; CHECK-NEXT: vid.v v9
; CHECK-NEXT: vrsub.vx v10, v9, a1, v0.t
; CHECK-NEXT: vrgather.vv v9, v8, v10, v0.t
; CHECK-NEXT: vmv.v.v v8, v9
; CHECK-NEXT: ret
Expand All @@ -63,10 +63,10 @@ define <4 x i32> @test_vp_reverse_v4i32(<4 x i32> %src, i32 zeroext %evl) {
define <8 x i16> @test_vp_reverse_v8i16_masked(<8 x i16> %src, <8 x i1> %mask, i32 zeroext %evl) {
; CHECK-LABEL: test_vp_reverse_v8i16_masked:
; CHECK: # %bb.0:
; CHECK-NEXT: addi a1, a0, -1
; CHECK-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; CHECK-NEXT: vid.v v9, v0.t
; CHECK-NEXT: addi a0, a0, -1
; CHECK-NEXT: vrsub.vx v10, v9, a0, v0.t
; CHECK-NEXT: vid.v v9
; CHECK-NEXT: vrsub.vx v10, v9, a1, v0.t
; CHECK-NEXT: vrgather.vv v9, v8, v10, v0.t
; CHECK-NEXT: vmv.v.v v8, v9
; CHECK-NEXT: ret
Expand All @@ -92,10 +92,10 @@ define <8 x i16> @test_vp_reverse_v8i16(<8 x i16> %src, i32 zeroext %evl) {
define <16 x i8> @test_vp_reverse_v16i8_masked(<16 x i8> %src, <16 x i1> %mask, i32 zeroext %evl) {
; CHECK-LABEL: test_vp_reverse_v16i8_masked:
; CHECK: # %bb.0:
; CHECK-NEXT: addi a1, a0, -1
; CHECK-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; CHECK-NEXT: vid.v v10, v0.t
; CHECK-NEXT: addi a0, a0, -1
; CHECK-NEXT: vrsub.vx v10, v10, a0, v0.t
; CHECK-NEXT: vid.v v10
; CHECK-NEXT: vrsub.vx v10, v10, a1, v0.t
; CHECK-NEXT: vsetvli zero, zero, e8, m1, ta, ma
; CHECK-NEXT: vrgatherei16.vv v9, v8, v10, v0.t
; CHECK-NEXT: vmv.v.v v8, v9
Expand Down
Loading
Loading