]>
Cypherpunks repositories - gostls13.git/commit
cmd/compile/internal: intrinsify publicationBarrier on loong64
The publication barrier is a StoreStore barrier, which is implemented
by "DBAR 0x1A" [1] on loong64.
goos: linux
goarch: loong64
pkg: runtime
cpu: Loongson-3A6000 @ 2500.00MHz
| bench.old | bench.new |
| sec/op | sec/op vs base |
Malloc8 31.76n ± 0% 22.79n ± 0% -28.24% (p=0.000 n=20)
Malloc8-2 25.46n ± 0% 18.33n ± 0% -28.00% (p=0.000 n=20)
Malloc8-4 25.75n ± 0% 18.43n ± 0% -28.41% (p=0.000 n=20)
Malloc16 62.97n ± 0% 42.41n ± 0% -32.65% (p=0.000 n=20)
Malloc16-2 49.11n ± 0% 31.68n ± 0% -35.50% (p=0.000 n=20)
Malloc16-4 49.64n ± 1% 31.95n ± 0% -35.62% (p=0.000 n=20)
MallocTypeInfo8 58.57n ± 0% 46.51n ± 0% -20.61% (p=0.000 n=20)
MallocTypeInfo8-2 51.43n ± 0% 38.01n ± 0% -26.09% (p=0.000 n=20)
MallocTypeInfo8-4 51.65n ± 0% 38.15n ± 0% -26.13% (p=0.000 n=20)
MallocTypeInfo16 68.07n ± 0% 51.62n ± 0% -24.17% (p=0.000 n=20)
MallocTypeInfo16-2 54.73n ± 0% 41.13n ± 0% -24.85% (p=0.000 n=20)
MallocTypeInfo16-4 55.05n ± 0% 41.28n ± 0% -25.02% (p=0.000 n=20)
MallocLargeStruct 491.5n ± 0% 454.8n ± 0% -7.47% (p=0.000 n=20)
MallocLargeStruct-2 351.8n ± 1% 323.8n ± 0% -7.94% (p=0.000 n=20)
MallocLargeStruct-4 333.6n ± 0% 316.7n ± 0% -5.10% (p=0.000 n=20)
geomean 71.01n 53.78n -24.26%
[1]: https://loongson.github.io/LoongArch-Documentation/LoongArch-Vol1-EN.html
Change-Id: Ica0c89db6f2bebd55d9b3207a1c462a9454e9268
Reviewed-on: https://go-review.googlesource.com/c/go/+/577515
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: sophie zhao <zhaoxiaolin@loongson.cn>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Qiqi Huang <huangqiqi@loongson.cn>
Reviewed-by: Meidan Li <limeidan@loongson.cn>
Reviewed-by: Carlos Amedee <carlos@golang.org>