mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-11-25 04:02:41 +01:00
85f191fc73
Check for single use of fmul node in fused multiply patterns to allow generation of fused multiply add/sub instructions. Otherwise fmul operation ends up being repeated more than once which does not help peformance on targets with only one MAC unit, as for example cortex-a53. llvm-svn: 197929
133 lines
5.3 KiB
LLVM
133 lines
5.3 KiB
LLVM
; RUN: llc < %s -verify-machineinstrs -mtriple=aarch64-none-linux-gnu -mattr=+neon -fp-contract=fast | FileCheck %s
|
|
|
|
define <2 x float> @fmla2xfloat(<2 x float> %A, <2 x float> %B, <2 x float> %C) {
|
|
;CHECK: fmla {{v[0-9]+}}.2s, {{v[0-9]+}}.2s, {{v[0-9]+}}.2s
|
|
%tmp1 = fmul <2 x float> %A, %B;
|
|
%tmp2 = fadd <2 x float> %C, %tmp1;
|
|
ret <2 x float> %tmp2
|
|
}
|
|
|
|
define <4 x float> @fmla4xfloat(<4 x float> %A, <4 x float> %B, <4 x float> %C) {
|
|
;CHECK: fmla {{v[0-9]+}}.4s, {{v[0-9]+}}.4s, {{v[0-9]+}}.4s
|
|
%tmp1 = fmul <4 x float> %A, %B;
|
|
%tmp2 = fadd <4 x float> %C, %tmp1;
|
|
ret <4 x float> %tmp2
|
|
}
|
|
|
|
define <2 x double> @fmla2xdouble(<2 x double> %A, <2 x double> %B, <2 x double> %C) {
|
|
;CHECK: fmla {{v[0-9]+}}.2d, {{v[0-9]+}}.2d, {{v[0-9]+}}.2d
|
|
%tmp1 = fmul <2 x double> %A, %B;
|
|
%tmp2 = fadd <2 x double> %C, %tmp1;
|
|
ret <2 x double> %tmp2
|
|
}
|
|
|
|
|
|
define <2 x float> @fmls2xfloat(<2 x float> %A, <2 x float> %B, <2 x float> %C) {
|
|
;CHECK: fmls {{v[0-9]+}}.2s, {{v[0-9]+}}.2s, {{v[0-9]+}}.2s
|
|
%tmp1 = fmul <2 x float> %A, %B;
|
|
%tmp2 = fsub <2 x float> %C, %tmp1;
|
|
ret <2 x float> %tmp2
|
|
}
|
|
|
|
define <4 x float> @fmls4xfloat(<4 x float> %A, <4 x float> %B, <4 x float> %C) {
|
|
;CHECK: fmls {{v[0-9]+}}.4s, {{v[0-9]+}}.4s, {{v[0-9]+}}.4s
|
|
%tmp1 = fmul <4 x float> %A, %B;
|
|
%tmp2 = fsub <4 x float> %C, %tmp1;
|
|
ret <4 x float> %tmp2
|
|
}
|
|
|
|
define <2 x double> @fmls2xdouble(<2 x double> %A, <2 x double> %B, <2 x double> %C) {
|
|
;CHECK: fmls {{v[0-9]+}}.2d, {{v[0-9]+}}.2d, {{v[0-9]+}}.2d
|
|
%tmp1 = fmul <2 x double> %A, %B;
|
|
%tmp2 = fsub <2 x double> %C, %tmp1;
|
|
ret <2 x double> %tmp2
|
|
}
|
|
|
|
|
|
; Another set of tests for when the intrinsic is used.
|
|
|
|
declare <2 x float> @llvm.fma.v2f32(<2 x float>, <2 x float>, <2 x float>)
|
|
declare <4 x float> @llvm.fma.v4f32(<4 x float>, <4 x float>, <4 x float>)
|
|
declare <2 x double> @llvm.fma.v2f64(<2 x double>, <2 x double>, <2 x double>)
|
|
|
|
define <2 x float> @fmla2xfloat_fused(<2 x float> %A, <2 x float> %B, <2 x float> %C) {
|
|
;CHECK: fmla {{v[0-9]+}}.2s, {{v[0-9]+}}.2s, {{v[0-9]+}}.2s
|
|
%val = call <2 x float> @llvm.fma.v2f32(<2 x float> %A, <2 x float> %B, <2 x float> %C)
|
|
ret <2 x float> %val
|
|
}
|
|
|
|
define <4 x float> @fmla4xfloat_fused(<4 x float> %A, <4 x float> %B, <4 x float> %C) {
|
|
;CHECK: fmla {{v[0-9]+}}.4s, {{v[0-9]+}}.4s, {{v[0-9]+}}.4s
|
|
%val = call <4 x float> @llvm.fma.v4f32(<4 x float> %A, <4 x float> %B, <4 x float> %C)
|
|
ret <4 x float> %val
|
|
}
|
|
|
|
define <2 x double> @fmla2xdouble_fused(<2 x double> %A, <2 x double> %B, <2 x double> %C) {
|
|
;CHECK: fmla {{v[0-9]+}}.2d, {{v[0-9]+}}.2d, {{v[0-9]+}}.2d
|
|
%val = call <2 x double> @llvm.fma.v2f64(<2 x double> %A, <2 x double> %B, <2 x double> %C)
|
|
ret <2 x double> %val
|
|
}
|
|
|
|
define <2 x float> @fmls2xfloat_fused(<2 x float> %A, <2 x float> %B, <2 x float> %C) {
|
|
;CHECK: fmls {{v[0-9]+}}.2s, {{v[0-9]+}}.2s, {{v[0-9]+}}.2s
|
|
%negA = fsub <2 x float> <float -0.0, float -0.0>, %A
|
|
%val = call <2 x float> @llvm.fma.v2f32(<2 x float> %negA, <2 x float> %B, <2 x float> %C)
|
|
ret <2 x float> %val
|
|
}
|
|
|
|
define <4 x float> @fmls4xfloat_fused(<4 x float> %A, <4 x float> %B, <4 x float> %C) {
|
|
;CHECK: fmls {{v[0-9]+}}.4s, {{v[0-9]+}}.4s, {{v[0-9]+}}.4s
|
|
%negA = fsub <4 x float> <float -0.0, float -0.0, float -0.0, float -0.0>, %A
|
|
%val = call <4 x float> @llvm.fma.v4f32(<4 x float> %negA, <4 x float> %B, <4 x float> %C)
|
|
ret <4 x float> %val
|
|
}
|
|
|
|
define <2 x double> @fmls2xdouble_fused(<2 x double> %A, <2 x double> %B, <2 x double> %C) {
|
|
;CHECK: fmls {{v[0-9]+}}.2d, {{v[0-9]+}}.2d, {{v[0-9]+}}.2d
|
|
%negA = fsub <2 x double> <double -0.0, double -0.0>, %A
|
|
%val = call <2 x double> @llvm.fma.v2f64(<2 x double> %negA, <2 x double> %B, <2 x double> %C)
|
|
ret <2 x double> %val
|
|
}
|
|
|
|
declare <2 x float> @llvm.fmuladd.v2f32(<2 x float>, <2 x float>, <2 x float>)
|
|
declare <4 x float> @llvm.fmuladd.v4f32(<4 x float>, <4 x float>, <4 x float>)
|
|
declare <2 x double> @llvm.fmuladd.v2f64(<2 x double>, <2 x double>, <2 x double>)
|
|
|
|
define <2 x float> @fmuladd2xfloat(<2 x float> %A, <2 x float> %B, <2 x float> %C) {
|
|
;CHECK: fmla {{v[0-9]+}}.2s, {{v[0-9]+}}.2s, {{v[0-9]+}}.2s
|
|
%val = call <2 x float> @llvm.fmuladd.v2f32(<2 x float> %A, <2 x float> %B, <2 x float> %C)
|
|
ret <2 x float> %val
|
|
}
|
|
|
|
define <4 x float> @fmuladd4xfloat_fused(<4 x float> %A, <4 x float> %B, <4 x float> %C) {
|
|
;CHECK: fmla {{v[0-9]+}}.4s, {{v[0-9]+}}.4s, {{v[0-9]+}}.4s
|
|
%val = call <4 x float> @llvm.fmuladd.v4f32(<4 x float> %A, <4 x float> %B, <4 x float> %C)
|
|
ret <4 x float> %val
|
|
}
|
|
|
|
define <2 x double> @fmuladd2xdouble_fused(<2 x double> %A, <2 x double> %B, <2 x double> %C) {
|
|
;CHECK: fmla {{v[0-9]+}}.2d, {{v[0-9]+}}.2d, {{v[0-9]+}}.2d
|
|
%val = call <2 x double> @llvm.fmuladd.v2f64(<2 x double> %A, <2 x double> %B, <2 x double> %C)
|
|
ret <2 x double> %val
|
|
}
|
|
|
|
|
|
; Another set of tests that check for multiply single use
|
|
|
|
define <2 x float> @fmla2xfloati_su(<2 x float> %A, <2 x float> %B, <2 x float> %C) {
|
|
;CHECK-NOT: fmla {{v[0-9]+}}.2s, {{v[0-9]+}}.2s, {{v[0-9]+}}.2s
|
|
%tmp1 = fmul <2 x float> %A, %B;
|
|
%tmp2 = fadd <2 x float> %C, %tmp1;
|
|
%tmp3 = fadd <2 x float> %tmp2, %tmp1;
|
|
ret <2 x float> %tmp3
|
|
}
|
|
|
|
define <2 x double> @fmls2xdouble_su(<2 x double> %A, <2 x double> %B, <2 x double> %C) {
|
|
;CHECK-NOT: fmls {{v[0-9]+}}.2d, {{v[0-9]+}}.2d, {{v[0-9]+}}.2d
|
|
%tmp1 = fmul <2 x double> %A, %B;
|
|
%tmp2 = fsub <2 x double> %C, %tmp1;
|
|
%tmp3 = fsub <2 x double> %tmp2, %tmp1;
|
|
ret <2 x double> %tmp3
|
|
}
|
|
|