Skip to content

[RISCV] Add vcpop.m/vfirst.m to RISCVMaskedPseudosTable #115162

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 4 commits into from
Nov 7, 2024
Merged

[RISCV] Add vcpop.m/vfirst.m to RISCVMaskedPseudosTable #115162

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
d66d8b6
[𝘀𝗽𝗿] changes to main this commit is based on
wangpc-pp Nov 6, 2024
7e72631
[𝘀𝗽𝗿] initial version
wangpc-pp Nov 6, 2024
224be13
[𝘀𝗽𝗿] changes introduced through rebase
wangpc-pp Nov 7, 2024
1478ea7
Move to near def
wangpc-pp Nov 7, 2024
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
14 changes: 12 additions & 2 deletions llvm/lib/Target/RISCV/RISCVISelDAGToDAG.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3708,6 +3708,15 @@ static bool isImplicitDef(SDValue V) {
return V.getMachineOpcode() == TargetOpcode::IMPLICIT_DEF;
}

static bool hasGPROut(unsigned Opc) {
switch (RISCV::getRVVMCOpcode(Opc)) {
case RISCV::VCPOP_M:
case RISCV::VFIRST_M:
return true;
}
return false;
}

// Optimize masked RVV pseudo instructions with a known all-ones mask to their
// corresponding "unmasked" pseudo versions. The mask we're interested in will
// take the form of a V0 physical register operand, with a glued
Expand Down Expand Up @@ -3737,8 +3746,9 @@ bool RISCVDAGToDAGISel::doPeepholeMaskedRVV(MachineSDNode *N) {
#endif

SmallVector<SDValue, 8> Ops;
// Skip the passthru operand at index 0 if !UseTUPseudo.
for (unsigned I = !UseTUPseudo, E = N->getNumOperands(); I != E; I++) {
// Skip the passthru operand at index 0 if !UseTUPseudo and no GPR out.
bool ShouldSkip = !UseTUPseudo && !hasGPROut(Opc);
for (unsigned I = ShouldSkip, E = N->getNumOperands(); I != E; I++) {
// Skip the mask, and the Glue.
SDValue Op = N->getOperand(I);
if (I == MaskOpIdx || Op.getValueType() == MVT::Glue)
Expand Down
2 changes: 2 additions & 0 deletions llvm/lib/Target/RISCV/RISCVInstrInfoVPseudos.td
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1983,6 +1983,7 @@ multiclass VPseudoVPOP_M {
def "_M_" # mti.BX : VPseudoUnaryNoMaskGPROut,
SchedBinary<"WriteVMPopV", "ReadVMPopV", "ReadVMPopV", mx>;
def "_M_" # mti.BX # "_MASK" : VPseudoUnaryMaskGPROut,
RISCVMaskedPseudo<MaskIdx=1>,
SchedBinary<"WriteVMPopV", "ReadVMPopV", "ReadVMPopV", mx>;
}
}
Expand All @@ -1995,6 +1996,7 @@ multiclass VPseudoV1ST_M {
def "_M_" #mti.BX : VPseudoUnaryNoMaskGPROut,
SchedBinary<"WriteVMFFSV", "ReadVMFFSV", "ReadVMFFSV", mx>;
def "_M_" # mti.BX # "_MASK" : VPseudoUnaryMaskGPROut,
RISCVMaskedPseudo<MaskIdx=1>,
SchedBinary<"WriteVMFFSV", "ReadVMFFSV", "ReadVMFFSV", mx>;
}
}
Expand Down
10 changes: 2 additions & 8 deletions llvm/test/CodeGen/RISCV/rvv/fixed-vectors-reduction-fp.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1797,11 +1797,8 @@ define float @vreduce_fminimum_v7f32(ptr %x) {
; CHECK: # %bb.0:
; CHECK-NEXT: vsetivli zero, 7, e32, m2, ta, ma
; CHECK-NEXT: vle32.v v8, (a0)
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
; CHECK-NEXT: vmset.m v0
; CHECK-NEXT: vsetivli zero, 7, e32, m2, ta, ma
; CHECK-NEXT: vmfne.vv v10, v8, v8
; CHECK-NEXT: vcpop.m a0, v10, v0.t
; CHECK-NEXT: vcpop.m a0, v10
; CHECK-NEXT: beqz a0, .LBB111_2
; CHECK-NEXT: # %bb.1:
; CHECK-NEXT: lui a0, 523264
Expand Down Expand Up @@ -2558,11 +2555,8 @@ define float @vreduce_fmaximum_v7f32(ptr %x) {
; CHECK: # %bb.0:
; CHECK-NEXT: vsetivli zero, 7, e32, m2, ta, ma
; CHECK-NEXT: vle32.v v8, (a0)
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
; CHECK-NEXT: vmset.m v0
; CHECK-NEXT: vsetivli zero, 7, e32, m2, ta, ma
; CHECK-NEXT: vmfne.vv v10, v8, v8
; CHECK-NEXT: vcpop.m a0, v10, v0.t
; CHECK-NEXT: vcpop.m a0, v10
; CHECK-NEXT: beqz a0, .LBB139_2
; CHECK-NEXT: # %bb.1:
; CHECK-NEXT: lui a0, 523264
Expand Down
Loading