Skip to content

[RISCV][VLOPT] Add vfsqrt/vfrsqrt7 instruction to isSupportInstr #127462

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

Merged
merged 3 commits into from
Feb 19, 2025
Merged

[RISCV][VLOPT] Add vfsqrt/vfrsqrt7 instruction to isSupportInstr #127462

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
c110e5b
[RISCV][VLOPT] Add fsqrt instruction to isSupportInstr
LiqinWeng Feb 17, 2025
33b7481
address the comments and add the tests
LiqinWeng Feb 18, 2025
5a5811f
address the comments
LiqinWeng Feb 19, 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
4 changes: 4 additions & 0 deletions llvm/lib/Target/RISCV/RISCVVLOptimizer.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1092,6 +1092,10 @@ static bool isSupportedInstr(const MachineInstr &MI) {
case RISCV::VFWNMSAC_VF:
case RISCV::VFWMACCBF16_VV:
case RISCV::VFWMACCBF16_VF:
// Vector Floating-Point Square-Root Instruction
case RISCV::VFSQRT_V:
// Vector Floating-Point Reciprocal Square-Root Estimate Instruction
case RISCV::VFRSQRT7_V:
// Vector Floating-Point MIN/MAX Instructions
case RISCV::VFMIN_VF:
case RISCV::VFMIN_VV:
Expand Down
6 changes: 2 additions & 4 deletions llvm/test/CodeGen/RISCV/rvv/fixed-vectors-fp.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1318,11 +1318,10 @@ define void @sqrt_v6bf16(ptr %x) {
; CHECK: # %bb.0:
; CHECK-NEXT: vsetivli zero, 6, e16, m1, ta, ma
; CHECK-NEXT: vle16.v v8, (a0)
; CHECK-NEXT: vsetivli zero, 8, e16, m1, ta, ma
; CHECK-NEXT: vfwcvtbf16.f.f.v v10, v8
; CHECK-NEXT: vsetvli zero, zero, e32, m2, ta, ma
; CHECK-NEXT: vfsqrt.v v8, v10
; CHECK-NEXT: vsetivli zero, 6, e16, m1, ta, ma
; CHECK-NEXT: vsetvli zero, zero, e16, m1, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v10, v8
; CHECK-NEXT: vse16.v v10, (a0)
; CHECK-NEXT: ret
Expand Down Expand Up @@ -1371,11 +1370,10 @@ define void @sqrt_v6f16(ptr %x) {
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: vsetivli zero, 6, e16, m1, ta, ma
; ZVFHMIN-NEXT: vle16.v v8, (a0)
; ZVFHMIN-NEXT: vsetivli zero, 8, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v8
; ZVFHMIN-NEXT: vsetvli zero, zero, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfsqrt.v v8, v10
; ZVFHMIN-NEXT: vsetivli zero, 6, e16, m1, ta, ma
; ZVFHMIN-NEXT: vsetvli zero, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v10, v8
; ZVFHMIN-NEXT: vse16.v v10, (a0)
; ZVFHMIN-NEXT: ret
Expand Down
48 changes: 48 additions & 0 deletions llvm/test/CodeGen/RISCV/rvv/vl-opt-instrs.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5069,3 +5069,51 @@ define <vscale x 4 x float> @vfwmaccbf16_vf(<vscale x 4 x float> %a, bfloat %b,
%2 = call <vscale x 4 x float> @llvm.riscv.vfadd(<vscale x 4 x float> poison, <vscale x 4 x float> %1, <vscale x 4 x float> %d, iXLen 7, iXLen %vl)
ret <vscale x 4 x float> %2
}

define <vscale x 4 x double> @vfsqrt(<vscale x 4 x float> %a) {
; NOVLOPT-LABEL: vfsqrt:
; NOVLOPT: # %bb.0:
; NOVLOPT-NEXT: fsrmi a0, 0
; NOVLOPT-NEXT: vsetivli zero, 7, e32, m2, ta, ma
; NOVLOPT-NEXT: vfsqrt.v v10, v8
; NOVLOPT-NEXT: fsrm a0
; NOVLOPT-NEXT: vsetivli zero, 6, e32, m2, ta, ma
; NOVLOPT-NEXT: vfwmacc.vv v12, v8, v10
; NOVLOPT-NEXT: vmv4r.v v8, v12
; NOVLOPT-NEXT: ret
;
; VLOPT-LABEL: vfsqrt:
; VLOPT: # %bb.0:
; VLOPT-NEXT: fsrmi a0, 0
; VLOPT-NEXT: vsetivli zero, 6, e32, m2, ta, ma
; VLOPT-NEXT: vfsqrt.v v10, v8
; VLOPT-NEXT: fsrm a0
; VLOPT-NEXT: vfwmacc.vv v12, v8, v10
; VLOPT-NEXT: vmv4r.v v8, v12
; VLOPT-NEXT: ret
%1 = call <vscale x 4 x float> @llvm.riscv.vfsqrt.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x float> %a, iXLen 0, iXLen 7)
%2 = call <vscale x 4 x double> @llvm.riscv.vfwmacc(<vscale x 4 x double> poison, <vscale x 4 x float> %a, <vscale x 4 x float> %1, iXLen 7, iXLen 6, iXLen 0)
ret <vscale x 4 x double> %2
}

define <vscale x 4 x double> @vfrsqrt7(<vscale x 4 x float> %a) {
; NOVLOPT-LABEL: vfrsqrt7:
; NOVLOPT: # %bb.0:
; NOVLOPT-NEXT: vsetivli zero, 7, e32, m2, ta, ma
; NOVLOPT-NEXT: vfrsqrt7.v v10, v8
; NOVLOPT-NEXT: vsetivli zero, 6, e32, m2, ta, ma
; NOVLOPT-NEXT: vfwmacc.vv v12, v8, v10
; NOVLOPT-NEXT: vmv4r.v v8, v12
; NOVLOPT-NEXT: ret
;
; VLOPT-LABEL: vfrsqrt7:
; VLOPT: # %bb.0:
; VLOPT-NEXT: vsetivli zero, 6, e32, m2, ta, ma
; VLOPT-NEXT: vfrsqrt7.v v10, v8
; VLOPT-NEXT: vfwmacc.vv v12, v8, v10
; VLOPT-NEXT: vmv4r.v v8, v12
; VLOPT-NEXT: ret
%1 = call <vscale x 4 x float> @llvm.riscv.vfrsqrt7.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x float> %a, iXLen 7)
%2 = call <vscale x 4 x double> @llvm.riscv.vfwmacc(<vscale x 4 x double> poison, <vscale x 4 x float> %a, <vscale x 4 x float> %1, iXLen 7, iXLen 6, iXLen 0)
ret <vscale x 4 x double> %2
}
40 changes: 40 additions & 0 deletions llvm/test/CodeGen/RISCV/rvv/vl-opt.mir
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -141,6 +141,46 @@ body: |
%y:vr = PseudoVADD_VV_M1 $noreg, %x, $noreg, 1, 4 /* e16 */, 0
...
---
name: vfsqrt_nofpexcept
body: |
bb.0:
; CHECK-LABEL: name: vfsqrt_nofpexcept
; CHECK: %x:vrm2 = nofpexcept PseudoVFSQRT_V_M2_E32 $noreg, $noreg, 7, 6, 5 /* e32 */, 3 /* ta, ma */, implicit $frm
; CHECK-NEXT: early-clobber %y:vr = nofpexcept PseudoVFNCVTBF16_F_F_W_M1_E16 $noreg, %x, 7, 6, 4 /* e16 */, 3 /* ta, ma */, implicit $frm
%x:vrm2 = nofpexcept PseudoVFSQRT_V_M2_E32 $noreg, $noreg, 7, 8, 5, 3, implicit $frm
early-clobber %y:vr = nofpexcept PseudoVFNCVTBF16_F_F_W_M1_E16 $noreg, %x, 7, 6, 4, 3, implicit $frm
...
---
name: vfsqrt_fpexcept
body: |
bb.0:
; CHECK-LABEL: name: vfsqrt_fpexcept
; CHECK: %x:vrm2 = PseudoVFSQRT_V_M2_E32 $noreg, $noreg, 7, 8, 5 /* e32 */, 3 /* ta, ma */, implicit $frm
; CHECK-NEXT: early-clobber %y:vr = nofpexcept PseudoVFNCVTBF16_F_F_W_M1_E16 $noreg, %x, 7, 6, 4 /* e16 */, 3 /* ta, ma */, implicit $frm
%x:vrm2 = PseudoVFSQRT_V_M2_E32 $noreg, $noreg, 7, 8, 5, 3, implicit $frm
early-clobber %y:vr = nofpexcept PseudoVFNCVTBF16_F_F_W_M1_E16 $noreg, %x, 7, 6, 4, 3, implicit $frm
...
---
name: vfrsqrt7_nofpexcept
body: |
bb.0:
; CHECK-LABEL: name: vfrsqrt7_nofpexcept
; CHECK: %x:vrm2 = nofpexcept PseudoVFRSQRT7_V_M2_E32 $noreg, $noreg, 1, 5 /* e32 */, 0 /* tu, mu */
; CHECK-NEXT: %y:vrm2 = PseudoVADD_VV_M2 $noreg, %x, $noreg, 1, 5 /* e32 */, 0 /* tu, mu */
%x:vrm2 = nofpexcept PseudoVFRSQRT7_V_M2_E32 $noreg, $noreg, 7, 5, 0
%y:vrm2 = PseudoVADD_VV_M2 $noreg, %x, $noreg, 1, 5 /* e32 */, 0
...
---
name: vfrsqrt7_fpexcept
body: |
bb.0:
; CHECK-LABEL: name: vfrsqrt7_fpexcept
; CHECK: %x:vrm2 = PseudoVFRSQRT7_V_M2_E32 $noreg, $noreg, 7, 5 /* e32 */, 0 /* tu, mu */
; CHECK-NEXT: %y:vrm2 = PseudoVADD_VV_M2 $noreg, %x, $noreg, 1, 5 /* e32 */, 0 /* tu, mu */
%x:vrm2 = PseudoVFRSQRT7_V_M2_E32 $noreg, $noreg, 7, 5, 0
%y:vrm2 = PseudoVADD_VV_M2 $noreg, %x, $noreg, 1, 5 /* e32 */, 0
...
---
name: vwadd_tied_vs1
body: |
bb.0:
Expand Down