Skip to content

[SPARC] Use FMA instructions when we have UA2007 #148434

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

Open
wants to merge 1 commit into
base: main
Choose a base branch
from
+168 −2
Open

[SPARC] Use FMA instructions when we have UA2007 #148434

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
15 changes: 13 additions & 2 deletions llvm/lib/Target/Sparc/SparcISelLowering.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1799,12 +1799,14 @@ SparcTargetLowering::SparcTargetLowering(const TargetMachine &TM,
setOperationAction(ISD::FCOS , MVT::f64, Expand);
setOperationAction(ISD::FSINCOS, MVT::f64, Expand);
setOperationAction(ISD::FREM , MVT::f64, Expand);
setOperationAction(ISD::FMA , MVT::f64, Expand);
setOperationAction(ISD::FMA, MVT::f64,
Subtarget->isUA2007() ? Legal : Expand);
setOperationAction(ISD::FSIN , MVT::f32, Expand);
setOperationAction(ISD::FCOS , MVT::f32, Expand);
setOperationAction(ISD::FSINCOS, MVT::f32, Expand);
setOperationAction(ISD::FREM , MVT::f32, Expand);
setOperationAction(ISD::FMA, MVT::f32, Expand);
setOperationAction(ISD::FMA, MVT::f32,
Subtarget->isUA2007() ? Legal : Expand);
setOperationAction(ISD::ROTL , MVT::i32, Expand);
setOperationAction(ISD::ROTR , MVT::i32, Expand);
setOperationAction(ISD::BSWAP, MVT::i32, Expand);
Expand Down Expand Up @@ -3570,6 +3572,15 @@ bool SparcTargetLowering::isCheapToSpeculateCttz(Type *Ty) const {
return isCheapToSpeculateCtlz(Ty);
}

bool SparcTargetLowering::isFMAFasterThanFMulAndFAdd(const MachineFunction &MF,
EVT VT) const {
return Subtarget->isUA2007();
}

bool SparcTargetLowering::enableAggressiveFMAFusion(EVT VT) const {
return Subtarget->isUA2007();
}

Comment on lines +3575 to +3583
Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Would it be safe if I change those to return a constant true?
Given that on targets without UA2007 FMAs are going to be expanded anyway...

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Probably yes, I think this is a modifier that only matters if isFMAFasterThanFMulAndFAdd was true anyway

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

But you should just add combine tests that show the effect

// Override to disable global variable loading on Linux.
void SparcTargetLowering::insertSSPDeclarations(Module &M) const {
if (!Subtarget->isTargetLinux())
Expand Down
5 changes: 5 additions & 0 deletions llvm/lib/Target/Sparc/SparcISelLowering.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -177,6 +177,11 @@ namespace llvm {

bool isCheapToSpeculateCttz(Type *Ty) const override;

bool enableAggressiveFMAFusion(EVT VT) const override;

bool isFMAFasterThanFMulAndFAdd(const MachineFunction &MF,
EVT VT) const override;

bool shouldInsertFencesForAtomic(const Instruction *I) const override {
// FIXME: We insert fences for each atomics and generate
// sub-optimal code for PSO/TSO. (Approximately nobody uses any
Expand Down
12 changes: 12 additions & 0 deletions llvm/lib/Target/Sparc/SparcInstrUAOSA.td
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -66,3 +66,15 @@ defm CXBCOND : F2_56<"cxb", 1>;
def FPMADDX : FourOp<"fpmaddx", 0b110111, 0b0000, DFPRegs>;
def FPMADDXHI : FourOp<"fpmaddxhi", 0b110111, 0b0100, DFPRegs>;
} // Predicates = [HasOSA2011]

// UA2007 instruction patterns.
let Predicates = [HasUA2007] in {
def : Pat<(f32 (any_fma f32:$rs1, f32:$rs2, f32:$add)), (FMADDS $rs1, $rs2, $add)>;
def : Pat<(f64 (any_fma f64:$rs1, f64:$rs2, f64:$add)), (FMADDD $rs1, $rs2, $add)>;
def : Pat<(f32 (any_fma f32:$rs1, f32:$rs2, (fneg f32:$sub))), (FMSUBS $rs1, $rs2, $sub)>;
def : Pat<(f64 (any_fma f64:$rs1, f64:$rs2, (fneg f64:$sub))), (FMSUBD $rs1, $rs2, $sub)>;
def : Pat<(f32 (fneg (any_fma f32:$rs1, f32:$rs2, f32:$add))), (FNMADDS $rs1, $rs2, $add)>;
def : Pat<(f64 (fneg (any_fma f64:$rs1, f64:$rs2, f64:$add))), (FNMADDD $rs1, $rs2, $add)>;
def : Pat<(f32 (fneg (any_fma f32:$rs1, f32:$rs2, (fneg f32:$sub)))), (FNMSUBS $rs1, $rs2, $sub)>;
def : Pat<(f64 (fneg (any_fma f64:$rs1, f64:$rs2, (fneg f64:$sub)))), (FNMSUBD $rs1, $rs2, $sub)>;
} // Predicates = [HasUA2007]
138 changes: 138 additions & 0 deletions llvm/test/CodeGen/SPARC/float-ua2007.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,138 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
; RUN: llc -mtriple=sparc64 -mattr=-ua2007 < %s | FileCheck %s -check-prefix=NO-UA2007
; RUN: llc -mtriple=sparc64 -mattr=+ua2007 < %s | FileCheck %s -check-prefix=UA2007

define float @fmadds(float %a, float %b, float %c) nounwind {
; NO-UA2007-LABEL: fmadds:
; NO-UA2007: ! %bb.0:
; NO-UA2007-NEXT: fmuls %f1, %f3, %f0
; NO-UA2007-NEXT: retl
; NO-UA2007-NEXT: fadds %f0, %f5, %f0
;
; UA2007-LABEL: fmadds:
; UA2007: ! %bb.0:
; UA2007-NEXT: retl
; UA2007-NEXT: fmadds %f1, %f3, %f5, %f0
%ret = call float @llvm.fmuladd.f32(float %a, float %b, float %c)
ret float %ret
}

define double @fmaddd(double %a, double %b, double %c) nounwind {
; NO-UA2007-LABEL: fmaddd:
; NO-UA2007: ! %bb.0:
; NO-UA2007-NEXT: fmuld %f0, %f2, %f0
; NO-UA2007-NEXT: retl
; NO-UA2007-NEXT: faddd %f0, %f4, %f0
;
; UA2007-LABEL: fmaddd:
; UA2007: ! %bb.0:
; UA2007-NEXT: retl
; UA2007-NEXT: fmaddd %f0, %f2, %f4, %f0
%ret = call double @llvm.fmuladd.f64(double %a, double %b, double %c)
ret double %ret
}

define float @fmsubs(float %a, float %b, float %c) nounwind {
; NO-UA2007-LABEL: fmsubs:
; NO-UA2007: ! %bb.0:
; NO-UA2007-NEXT: fmuls %f1, %f3, %f0
; NO-UA2007-NEXT: retl
; NO-UA2007-NEXT: fsubs %f0, %f5, %f0
;
; UA2007-LABEL: fmsubs:
; UA2007: ! %bb.0:
; UA2007-NEXT: retl
; UA2007-NEXT: fmsubs %f1, %f3, %f5, %f0
%neg = fneg float %c
%ret = call float @llvm.fmuladd.f32(float %a, float %b, float %neg)
ret float %ret
}

define double @fmsubd(double %a, double %b, double %c) nounwind {
; NO-UA2007-LABEL: fmsubd:
; NO-UA2007: ! %bb.0:
; NO-UA2007-NEXT: fmuld %f0, %f2, %f0
; NO-UA2007-NEXT: retl
; NO-UA2007-NEXT: fsubd %f0, %f4, %f0
;
; UA2007-LABEL: fmsubd:
; UA2007: ! %bb.0:
; UA2007-NEXT: retl
; UA2007-NEXT: fmsubd %f0, %f2, %f4, %f0
%neg = fneg double %c
%ret = call double @llvm.fmuladd.f64(double %a, double %b, double %neg)
ret double %ret
}

define float @fnmadds(float %a, float %b, float %c) nounwind {
; NO-UA2007-LABEL: fnmadds:
; NO-UA2007: ! %bb.0:
; NO-UA2007-NEXT: fmuls %f1, %f3, %f0
; NO-UA2007-NEXT: fadds %f0, %f5, %f0
; NO-UA2007-NEXT: retl
; NO-UA2007-NEXT: fnegs %f0, %f0
;
; UA2007-LABEL: fnmadds:
; UA2007: ! %bb.0:
; UA2007-NEXT: retl
; UA2007-NEXT: fnmadds %f1, %f3, %f5, %f0
%fma = call float @llvm.fmuladd.f32(float %a, float %b, float %c)
%ret = fneg float %fma
ret float %ret
}

define double @fnmaddd(double %a, double %b, double %c) nounwind {
; NO-UA2007-LABEL: fnmaddd:
; NO-UA2007: ! %bb.0:
; NO-UA2007-NEXT: fmuld %f0, %f2, %f0
; NO-UA2007-NEXT: faddd %f0, %f4, %f0
; NO-UA2007-NEXT: retl
; NO-UA2007-NEXT: fnegd %f0, %f0
;
; UA2007-LABEL: fnmaddd:
; UA2007: ! %bb.0:
; UA2007-NEXT: retl
; UA2007-NEXT: fnmaddd %f0, %f2, %f4, %f0
%fma = call double @llvm.fmuladd.f64(double %a, double %b, double %c)
%ret = fneg double %fma
ret double %ret
}

define float @fnmsubs(float %a, float %b, float %c) nounwind {
; NO-UA2007-LABEL: fnmsubs:
; NO-UA2007: ! %bb.0:
; NO-UA2007-NEXT: fmuls %f1, %f3, %f0
; NO-UA2007-NEXT: fsubs %f0, %f5, %f0
; NO-UA2007-NEXT: retl
; NO-UA2007-NEXT: fnegs %f0, %f0
;
; UA2007-LABEL: fnmsubs:
; UA2007: ! %bb.0:
; UA2007-NEXT: retl
; UA2007-NEXT: fnmsubs %f1, %f3, %f5, %f0
%neg = fneg float %c
%fma = call float @llvm.fmuladd.f32(float %a, float %b, float %neg)
%ret = fneg float %fma
ret float %ret
}

define double @fnmsubd(double %a, double %b, double %c) nounwind {
; NO-UA2007-LABEL: fnmsubd:
; NO-UA2007: ! %bb.0:
; NO-UA2007-NEXT: fmuld %f0, %f2, %f0
; NO-UA2007-NEXT: fsubd %f0, %f4, %f0
; NO-UA2007-NEXT: retl
; NO-UA2007-NEXT: fnegd %f0, %f0
;
; UA2007-LABEL: fnmsubd:
; UA2007: ! %bb.0:
; UA2007-NEXT: retl
; UA2007-NEXT: fnmsubd %f0, %f2, %f4, %f0
%neg = fneg double %c
%fma = call double @llvm.fmuladd.f64(double %a, double %b, double %neg)
%ret = fneg double %fma
ret double %ret
}

declare float @llvm.fmuladd.f32(float, float, float)
declare double @llvm.fmuladd.f64(double, double, double)