mirror of
https://github.com/RPCSX/llvm.git
synced 2025-01-09 13:41:35 +00:00
e023141322
boolean flag to an enum: { Fast, Standard, Strict } (default = Standard).
This option controls the creation by optimizations of fused FP ops that store
intermediate results in higher precision than IEEE allows (E.g. FMAs). The
behavior of this option is intended to match the behaviour specified by a
soon-to-be-introduced frontend flag: '-ffuse-fp-ops'.
Fast mode - allows formation of fused FP ops whenever they're profitable.
Standard mode - allow fusion only for 'blessed' FP ops. At present the only
blessed op is the fmuladd intrinsic. In the future more blessed ops may be
added.
Strict mode - allow fusion only if/when it can be proven that the excess
precision won't effect the result.
Note: This option only controls formation of fused ops by the optimizers. Fused
operations that are explicitly requested (e.g. FMA via the llvm.fma.* intrinsic)
will always be honored, regardless of the value of this option.
Internally TargetOptions::AllowExcessFPPrecision has been replaced by
TargetOptions::AllowFPOpFusion.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@158956 91177308-0d34-0410-b5e6-96231b3b80d8
34 lines
1.3 KiB
LLVM
34 lines
1.3 KiB
LLVM
; RUN: llc < %s -march=ppc32 -mcpu=440 -fuse-fp-ops=fast | FileCheck %s
|
|
|
|
%0 = type { double, double }
|
|
|
|
define void @maybe_an_fma(%0* sret %agg.result, %0* byval %a, %0* byval %b, %0* byval %c) nounwind {
|
|
entry:
|
|
%a.realp = getelementptr inbounds %0* %a, i32 0, i32 0
|
|
%a.real = load double* %a.realp
|
|
%a.imagp = getelementptr inbounds %0* %a, i32 0, i32 1
|
|
%a.imag = load double* %a.imagp
|
|
%b.realp = getelementptr inbounds %0* %b, i32 0, i32 0
|
|
%b.real = load double* %b.realp
|
|
%b.imagp = getelementptr inbounds %0* %b, i32 0, i32 1
|
|
%b.imag = load double* %b.imagp
|
|
%mul.rl = fmul double %a.real, %b.real
|
|
%mul.rr = fmul double %a.imag, %b.imag
|
|
%mul.r = fsub double %mul.rl, %mul.rr
|
|
%mul.il = fmul double %a.imag, %b.real
|
|
%mul.ir = fmul double %a.real, %b.imag
|
|
%mul.i = fadd double %mul.il, %mul.ir
|
|
%c.realp = getelementptr inbounds %0* %c, i32 0, i32 0
|
|
%c.real = load double* %c.realp
|
|
%c.imagp = getelementptr inbounds %0* %c, i32 0, i32 1
|
|
%c.imag = load double* %c.imagp
|
|
%add.r = fadd double %mul.r, %c.real
|
|
%add.i = fadd double %mul.i, %c.imag
|
|
%real = getelementptr inbounds %0* %agg.result, i32 0, i32 0
|
|
%imag = getelementptr inbounds %0* %agg.result, i32 0, i32 1
|
|
store double %add.r, double* %real
|
|
store double %add.i, double* %imag
|
|
ret void
|
|
; CHECK: fmadd
|
|
}
|