Skip to content

[RISCV] Check only demanded VTYPE fields in needVSETVLIPHI #90168

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 1 commit into from
Closed

[RISCV] Check only demanded VTYPE fields in needVSETVLIPHI #90168

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
22 changes: 14 additions & 8 deletions llvm/lib/Target/RISCV/RISCVInsertVSETVLI.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -808,8 +808,8 @@ class RISCVInsertVSETVLI : public MachineFunctionPass {
private:
bool needVSETVLI(const MachineInstr &MI, const VSETVLIInfo &Require,
const VSETVLIInfo &CurInfo) const;
bool needVSETVLIPHI(const VSETVLIInfo &Require,
const MachineBasicBlock &MBB) const;
bool needVSETVLIPHI(const VSETVLIInfo &Require, const MachineBasicBlock &MBB,
const DemandedFields &Used) const;
void insertVSETVLI(MachineBasicBlock &MBB, MachineInstr &MI,
const VSETVLIInfo &Info, const VSETVLIInfo &PrevInfo);
void insertVSETVLI(MachineBasicBlock &MBB,
Expand Down Expand Up @@ -1318,7 +1318,8 @@ void RISCVInsertVSETVLI::computeIncomingVLVTYPE(const MachineBasicBlock &MBB) {
// be unneeded if the AVL is a phi node where all incoming values are VL
// outputs from the last VSETVLI in their respective basic blocks.
bool RISCVInsertVSETVLI::needVSETVLIPHI(const VSETVLIInfo &Require,
const MachineBasicBlock &MBB) const {
const MachineBasicBlock &MBB,
const DemandedFields &Used) const {
if (DisableInsertVSETVLPHIOpt)
return true;

Expand Down Expand Up @@ -1350,10 +1351,14 @@ bool RISCVInsertVSETVLI::needVSETVLIPHI(const VSETVLIInfo &Require,
if (DefInfo != PBBExit)
return true;

// Require has the same VL as PBBExit, so if the exit from the
// predecessor has the VTYPE we are looking for we might be able
// to avoid a VSETVLI.
if (PBBExit.isUnknown() || !PBBExit.hasSameVTYPE(Require))
// Make sure Require has the same VLMAX as the predecessor block,
// otherwise the VL might be different.
if (PBBExit.isUnknown() || !PBBExit.hasSameVLMAX(Require))
return true;

// If the exit from the predecessor has the VTYPE we are looking
// for we might be able to avoid a VSETVLI.
if (!PBBExit.hasCompatibleVTYPE(Used, Require))
return true;
}

Expand Down Expand Up @@ -1392,7 +1397,8 @@ void RISCVInsertVSETVLI::emitVSETVLIs(MachineBasicBlock &MBB) {
// wouldn't be used and VL/VTYPE registers are correct. Note that
// we *do* need to model the state as if it changed as while the
// register contents are unchanged, the abstract model can change.
if (!PrefixTransparent || needVSETVLIPHI(CurInfo, MBB))
if (!PrefixTransparent ||
needVSETVLIPHI(CurInfo, MBB, getDemanded(MI, MRI, ST)))
insertVSETVLI(MBB, MI, CurInfo, PrevInfo);
PrefixTransparent = false;
}
Expand Down
1 change: 0 additions & 1 deletion llvm/test/CodeGen/RISCV/rvv/vsetvli-insert-crossbb.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -494,7 +494,6 @@ define void @saxpy_vec_demanded_fields(i64 %n, float %a, ptr nocapture readonly
; CHECK-NEXT: beqz a3, .LBB9_2
; CHECK-NEXT: .LBB9_1: # %for.body
; CHECK-NEXT: # =>This Inner Loop Header: Depth=1
; CHECK-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; CHECK-NEXT: vle32.v v8, (a1)
; CHECK-NEXT: vle32.v v16, (a2)
; CHECK-NEXT: slli a4, a3, 2
Expand Down