diff options
| author | erifan01 <eric.fang@arm.com> | 2018-05-22 06:58:32 +0000 |
|---|---|---|
| committer | Cherry Zhang <cherryyz@google.com> | 2018-09-13 14:52:51 +0000 |
| commit | 8149db4f64aa72407d8be2184d0b414b535cd124 (patch) | |
| tree | 0c0699f22bd284fe467d05df24c489bce827e745 /test/codegen | |
| parent | a2a3dd00c934fa15ad880ee5fe1f64308cbc73a7 (diff) | |
| download | go-8149db4f64aa72407d8be2184d0b414b535cd124.tar.xz | |
cmd/compile: intrinsify math.RoundToEven and math.Abs on arm64
math.RoundToEven can be done by one arm64 instruction FRINTND, intrinsify it to improve performance.
The current pure Go implementation of the function Abs is translated into five instructions on arm64:
str, ldr, and, str, ldr. The intrinsic implementation requires only one instruction, so in terms of
performance, intrinsify it is worthwhile.
Benchmarks:
name old time/op new time/op delta
Abs-8 3.50ns ± 0% 1.50ns ± 0% -57.14% (p=0.000 n=10+10)
RoundToEven-8 9.26ns ± 0% 1.50ns ± 0% -83.80% (p=0.000 n=10+10)
Change-Id: I9456b26ab282b544dfac0154fc86f17aed96ac3d
Reviewed-on: https://go-review.googlesource.com/116535
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Diffstat (limited to 'test/codegen')
| -rw-r--r-- | test/codegen/math.go | 2 |
1 files changed, 2 insertions, 0 deletions
diff --git a/test/codegen/math.go b/test/codegen/math.go index 3d5f0917ef..6afe183345 100644 --- a/test/codegen/math.go +++ b/test/codegen/math.go @@ -32,6 +32,7 @@ func approx(x float64) { sink64[3] = math.Trunc(x) // s390x:"FIDBR\t[$]4" + // arm64:"FRINTND" sink64[4] = math.RoundToEven(x) } @@ -48,6 +49,7 @@ func sqrt(x float64) float64 { // Check that it's using integer registers func abs(x, y float64) { // amd64:"BTRQ\t[$]63" + // arm64:"FABSD\t" // s390x:"LPDFR\t",-"MOVD\t" (no integer load/store) // ppc64le:"FABS\t" sink64[0] = math.Abs(x) |