]>
Cypherpunks repositories - gostls13.git/commit
runtime: optimize the function memmove on loong64
benchmarck on 3A6000:
goos: linux
goarch: loong64
pkg: runtime
cpu: Loongson-3A6000 @ 2500.00MHz
│ old │ new │
│ sec/op │ sec/op vs base │
Memmove/0 0.6003n ± 0% 0.6003n ± 0% ~ (p=0.487 n=20)
Memmove/1 4.402n ± 0% 2.815n ± 0% -36.05% (p=0.000 n=20)
Memmove/2 5.202n ± 0% 3.202n ± 0% -38.45% (p=0.000 n=20)
Memmove/3 6.003n ± 0% 2.820n ± 0% -53.02% (p=0.000 n=20)
Memmove/4 6.803n ± 0% 3.202n ± 0% -52.93% (p=0.000 n=20)
Memmove/5 7.604n ± 0% 3.202n ± 0% -57.89% (p=0.000 n=20)
Memmove/6 8.404n ± 0% 3.202n ± 0% -61.90% (p=0.000 n=20)
Memmove/7 9.204n ± 0% 3.202n ± 0% -65.21% (p=0.000 n=20)
Memmove/8 4.802n ± 0% 3.602n ± 0% -24.99% (p=0.000 n=20)
Memmove/9 6.003n ± 0% 3.202n ± 0% -46.66% (p=0.000 n=20)
Memmove/10 6.803n ± 0% 3.202n ± 0% -52.93% (p=0.000 n=20)
Memmove/11 7.604n ± 0% 3.202n ± 0% -57.89% (p=0.000 n=20)
Memmove/12 8.404n ± 0% 3.202n ± 0% -61.90% (p=0.000 n=20)
Memmove/13 9.204n ± 0% 3.202n ± 0% -65.21% (p=0.000 n=20)
Memmove/14 10.000n ± 0% 3.202n ± 0% -67.98% (p=0.000 n=20)
Memmove/15 10.810n ± 0% 3.202n ± 0% -70.38% (p=0.000 n=20)
Memmove/16 6.003n ± 0% 3.202n ± 0% -46.66% (p=0.000 n=20)
Memmove/32 7.604n ± 0% 3.602n ± 0% -52.63% (p=0.000 n=20)
Memmove/64 10.810n ± 0% 4.402n ± 0% -59.28% (p=0.000 n=20)
Memmove/128 17.210n ± 0% 8.004n ± 0% -53.49% (p=0.000 n=20)
Memmove/256 30.41n ± 0% 10.81n ± 0% -64.45% (p=0.000 n=20)
Memmove/512 56.03n ± 0% 17.81n ± 0% -68.21% (p=0.000 n=20)
Memmove/1024 107.30n ± 0% 30.62n ± 0% -71.46% (p=0.000 n=20)
Memmove/2048 209.70n ± 0% 56.23n ± 0% -73.19% (p=0.000 n=20)
Memmove/4096 414.6n ± 0% 107.5n ± 0% -74.07% (p=0.000 n=20)
MemmoveOverlap/32 8.404n ± 0% 4.402n ± 0% -47.62% (p=0.000 n=20)
MemmoveOverlap/64 11.610n ± 0% 5.003n ± 0% -56.91% (p=0.000 n=20)
MemmoveOverlap/128 18.010n ± 0% 9.005n ± 0% -50.00% (p=0.000 n=20)
MemmoveOverlap/256 31.22n ± 0% 12.41n ± 0% -60.25% (p=0.000 n=20)
MemmoveOverlap/512 56.83n ± 0% 19.08n ± 0% -66.43% (p=0.000 n=20)
MemmoveOverlap/1024 108.10n ± 0% 32.00n ± 0% -70.40% (p=0.000 n=20)
MemmoveOverlap/2048 210.50n ± 0% 57.94n ± 0% -72.48% (p=0.000 n=20)
MemmoveOverlap/4096 415.4n ± 0% 108.9n ± 0% -73.78% (p=0.000 n=20)
MemmoveUnalignedDst/0 2.448n ± 0% 2.942n ± 0% +20.16% (p=0.000 n=20)
MemmoveUnalignedDst/1 4.802n ± 0% 3.202n ± 0% -33.32% (p=0.000 n=20)
MemmoveUnalignedDst/2 5.603n ± 0% 3.602n ± 0% -35.71% (p=0.000 n=20)
MemmoveUnalignedDst/3 6.403n ± 0% 3.202n ± 0% -49.99% (p=0.000 n=20)
MemmoveUnalignedDst/4 7.203n ± 0% 3.602n ± 0% -49.99% (p=0.000 n=20)
MemmoveUnalignedDst/5 8.004n ± 0% 3.602n ± 0% -55.00% (p=0.000 n=20)
MemmoveUnalignedDst/6 8.804n ± 0% 3.602n ± 0% -59.09% (p=0.000 n=20)
MemmoveUnalignedDst/7 9.605n ± 0% 3.602n ± 0% -62.50% (p=0.000 n=20)
MemmoveUnalignedDst/8 10.400n ± 0% 4.002n ± 0% -61.52% (p=0.000 n=20)
MemmoveUnalignedDst/9 11.210n ± 0% 3.802n ± 0% -66.08% (p=0.000 n=20)
MemmoveUnalignedDst/10 12.010n ± 0% 3.802n ± 0% -68.34% (p=0.000 n=20)
MemmoveUnalignedDst/11 12.810n ± 0% 3.802n ± 0% -70.32% (p=0.000 n=20)
MemmoveUnalignedDst/12 13.610n ± 0% 3.802n ± 0% -72.06% (p=0.000 n=20)
MemmoveUnalignedDst/13 14.410n ± 0% 3.802n ± 0% -73.62% (p=0.000 n=20)
MemmoveUnalignedDst/14 15.210n ± 0% 3.802n ± 0% -75.00% (p=0.000 n=20)
MemmoveUnalignedDst/15 16.010n ± 0% 3.802n ± 0% -76.25% (p=0.000 n=20)
MemmoveUnalignedDst/16 17.210n ± 0% 3.802n ± 0% -77.91% (p=0.000 n=20)
MemmoveUnalignedDst/32 30.020n ± 0% 4.202n ± 0% -86.00% (p=0.000 n=20)
MemmoveUnalignedDst/64 56.030n ± 0% 6.804n ± 0% -87.86% (p=0.000 n=20)
MemmoveUnalignedDst/128 106.90n ± 0% 13.61n ± 0% -87.27% (p=0.000 n=20)
MemmoveUnalignedDst/256 209.50n ± 0% 17.07n ± 1% -91.85% (p=0.000 n=20)
MemmoveUnalignedDst/512 414.60n ± 0% 24.95n ± 0% -93.98% (p=0.000 n=20)
MemmoveUnalignedDst/1024 828.40n ± 0% 42.82n ± 0% -94.83% (p=0.000 n=20)
MemmoveUnalignedDst/2048 1648.00n ± 0% 78.04n ± 0% -95.26% (p=0.000 n=20)
MemmoveUnalignedDst/4096 3287.0n ± 0% 148.4n ± 0% -95.49% (p=0.000 n=20)
MemmoveUnalignedDstOverlap/32 30.810n ± 0% 5.603n ± 0% -81.81% (p=0.000 n=20)
MemmoveUnalignedDstOverlap/64 56.430n ± 0% 7.604n ± 0% -86.52% (p=0.000 n=20)
MemmoveUnalignedDstOverlap/128 107.700n ± 0% 9.812n ± 0% -90.89% (p=0.000 n=20)
MemmoveUnalignedDstOverlap/256 210.10n ± 0% 13.50n ± 0% -93.57% (p=0.000 n=20)
MemmoveUnalignedDstOverlap/512 415.00n ± 0% 21.21n ± 0% -94.89% (p=0.000 n=20)
MemmoveUnalignedDstOverlap/1024 828.80n ± 0% 41.02n ± 0% -95.05% (p=0.000 n=20)
MemmoveUnalignedDstOverlap/2048 1648.00n ± 0% 80.23n ± 0% -95.13% (p=0.000 n=20)
MemmoveUnalignedDstOverlap/4096 3288.0n ± 0% 162.4n ± 0% -95.06% (p=0.000 n=20)
MemmoveUnalignedSrc/0 2.468n ± 1% 2.913n ± 0% +18.01% (p=0.000 n=20)
MemmoveUnalignedSrc/1 4.802n ± 0% 3.202n ± 0% -33.32% (p=0.000 n=20)
MemmoveUnalignedSrc/2 5.603n ± 0% 3.603n ± 0% -35.70% (p=0.000 n=20)
MemmoveUnalignedSrc/3 6.403n ± 0% 3.207n ± 0% -49.91% (p=0.000 n=20)
MemmoveUnalignedSrc/4 7.203n ± 0% 3.603n ± 0% -49.98% (p=0.000 n=20)
MemmoveUnalignedSrc/5 8.004n ± 0% 3.602n ± 0% -55.00% (p=0.000 n=20)
MemmoveUnalignedSrc/6 8.804n ± 0% 3.602n ± 0% -59.09% (p=0.000 n=20)
MemmoveUnalignedSrc/7 9.605n ± 0% 3.602n ± 0% -62.50% (p=0.000 n=20)
MemmoveUnalignedSrc/8 10.410n ± 0% 4.002n ± 0% -61.56% (p=0.000 n=20)
MemmoveUnalignedSrc/9 11.210n ± 0% 3.802n ± 0% -66.08% (p=0.000 n=20)
MemmoveUnalignedSrc/10 12.010n ± 0% 3.802n ± 0% -68.34% (p=0.000 n=20)
MemmoveUnalignedSrc/11 12.810n ± 0% 3.802n ± 0% -70.32% (p=0.000 n=20)
MemmoveUnalignedSrc/12 13.610n ± 0% 3.802n ± 0% -72.06% (p=0.000 n=20)
MemmoveUnalignedSrc/13 14.410n ± 0% 3.802n ± 0% -73.62% (p=0.000 n=20)
MemmoveUnalignedSrc/14 15.210n ± 0% 3.802n ± 0% -75.00% (p=0.000 n=20)
MemmoveUnalignedSrc/15 16.010n ± 0% 3.802n ± 0% -76.25% (p=0.000 n=20)
MemmoveUnalignedSrc/16 16.810n ± 0% 3.802n ± 0% -77.38% (p=0.000 n=20)
MemmoveUnalignedSrc/32 30.410n ± 0% 4.301n ± 0% -85.86% (p=0.000 n=20)
MemmoveUnalignedSrc/64 55.630n ± 0% 5.203n ± 0% -90.65% (p=0.000 n=20)
MemmoveUnalignedSrc/128 107.300n ± 0% 8.805n ± 0% -91.79% (p=0.000 n=20)
MemmoveUnalignedSrc/256 209.50n ± 0% 12.41n ± 6% -94.08% (p=0.000 n=20)
MemmoveUnalignedSrc/512 414.20n ± 0% 20.41n ± 0% -95.07% (p=0.000 n=20)
MemmoveUnalignedSrc/1024 828.00n ± 0% 36.92n ± 0% -95.54% (p=0.000 n=20)
MemmoveUnalignedSrc/2048 1648.00n ± 0% 71.41n ± 0% -95.67% (p=0.000 n=20)
MemmoveUnalignedSrc/4096 3287.0n ± 0% 132.2n ± 0% -95.98% (p=0.000 n=20)
MemmoveUnalignedSrcDst/f_16_0 7.203n ± 0% 5.002n ± 0% -30.56% (p=0.000 n=20)
MemmoveUnalignedSrcDst/b_16_0 7.604n ± 0% 5.002n ± 0% -34.22% (p=0.000 n=20)
MemmoveUnalignedSrcDst/f_16_1 13.210n ± 0% 5.002n ± 0% -62.13% (p=0.000 n=20)
MemmoveUnalignedSrcDst/b_16_1 13.210n ± 0% 5.002n ± 0% -62.13% (p=0.000 n=20)
MemmoveUnalignedSrcDst/f_16_4 13.210n ± 0% 5.002n ± 0% -62.13% (p=0.000 n=20)
MemmoveUnalignedSrcDst/b_16_4 13.610n ± 0% 5.002n ± 0% -63.24% (p=0.000 n=20)
MemmoveUnalignedSrcDst/f_16_7 12.810n ± 0% 5.002n ± 0% -60.95% (p=0.000 n=20)
MemmoveUnalignedSrcDst/b_16_7 13.610n ± 0% 5.002n ± 0% -63.24% (p=0.000 n=20)
MemmoveUnalignedSrcDst/f_64_0 12.010n ± 0% 7.191n ± 0% -40.12% (p=0.000 n=20)
MemmoveUnalignedSrcDst/b_64_0 12.410n ± 0% 7.194n ± 0% -42.03% (p=0.000 n=20)
MemmoveUnalignedSrcDst/f_64_1 18.410n ± 0% 7.604n ± 0% -58.70% (p=0.000 n=20)
MemmoveUnalignedSrcDst/b_64_1 18.410n ± 0% 7.604n ± 0% -58.70% (p=0.000 n=20)
MemmoveUnalignedSrcDst/f_64_4 18.410n ± 0% 7.604n ± 0% -58.70% (p=0.000 n=20)
MemmoveUnalignedSrcDst/b_64_4 18.810n ± 0% 7.604n ± 0% -59.57% (p=0.000 n=20)
MemmoveUnalignedSrcDst/f_64_7 18.010n ± 0% 7.604n ± 0% -57.78% (p=0.000 n=20)
MemmoveUnalignedSrcDst/b_64_7 18.810n ± 0% 7.604n ± 0% -59.57% (p=0.000 n=20)
MemmoveUnalignedSrcDst/f_256_0 31.62n ± 0% 14.19n ± 0% -55.12% (p=0.000 n=20)
MemmoveUnalignedSrcDst/b_256_0 32.02n ± 0% 13.61n ± 0% -57.50% (p=0.000 n=20)
MemmoveUnalignedSrcDst/f_256_1 38.02n ± 0% 18.20n ± 0% -52.13% (p=0.000 n=20)
MemmoveUnalignedSrcDst/b_256_1 38.02n ± 0% 18.41n ± 0% -51.58% (p=0.000 n=20)
MemmoveUnalignedSrcDst/f_256_4 38.02n ± 0% 17.21n ± 0% -54.73% (p=0.000 n=20)
MemmoveUnalignedSrcDst/b_256_4 38.42n ± 0% 16.81n ± 0% -56.25% (p=0.000 n=20)
MemmoveUnalignedSrcDst/f_256_7 37.62n ± 0% 15.61n ± 0% -58.51% (p=0.000 n=20)
MemmoveUnalignedSrcDst/b_256_7 38.42n ± 0% 15.01n ± 0% -60.93% (p=0.000 n=20)
MemmoveUnalignedSrcDst/f_4096_0 415.8n ± 0% 111.1n ± 0% -73.28% (p=0.000 n=20)
MemmoveUnalignedSrcDst/b_4096_0 416.2n ± 0% 110.5n ± 0% -73.45% (p=0.000 n=20)
MemmoveUnalignedSrcDst/f_4096_1 422.2n ± 0% 114.3n ± 0% -72.93% (p=0.000 n=20)
MemmoveUnalignedSrcDst/b_4096_1 422.2n ± 0% 114.7n ± 0% -72.83% (p=0.000 n=20)
MemmoveUnalignedSrcDst/f_4096_4 422.2n ± 0% 113.3n ± 0% -73.16% (p=0.000 n=20)
MemmoveUnalignedSrcDst/b_4096_4 422.6n ± 0% 113.1n ± 0% -73.24% (p=0.000 n=20)
MemmoveUnalignedSrcDst/f_4096_7 421.8n ± 0% 111.7n ± 0% -73.52% (p=0.000 n=20)
MemmoveUnalignedSrcDst/b_4096_7 422.6n ± 0% 111.7n ± 0% -73.57% (p=0.000 n=20)
MemmoveUnalignedSrcDst/f_65536_0 6.568µ ± 0% 4.869µ ± 0% -25.88% (p=0.000 n=20)
MemmoveUnalignedSrcDst/b_65536_0 6.568µ ± 0% 5.009µ ± 0% -23.74% (p=0.000 n=20)
MemmoveUnalignedSrcDst/f_65536_1 6.574µ ± 0% 4.743µ ± 0% -27.85% (p=0.000 n=20)
MemmoveUnalignedSrcDst/b_65536_1 6.574µ ± 0% 4.770µ ± 0% -27.44% (p=0.000 n=20)
MemmoveUnalignedSrcDst/f_65536_4 6.574µ ± 0% 4.758µ ± 0% -27.63% (p=0.000 n=20)
MemmoveUnalignedSrcDst/b_65536_4 6.574µ ± 0% 4.768µ ± 0% -27.48% (p=0.000 n=20)
MemmoveUnalignedSrcDst/f_65536_7 6.574µ ± 0% 4.757µ ± 0% -27.64% (p=0.000 n=20)
MemmoveUnalignedSrcDst/b_65536_7 6.574µ ± 0% 4.583µ ± 0% -30.29% (p=0.000 n=20)
MemmoveUnalignedSrcOverlap/32 30.410n ± 0% 6.804n ± 0% -77.63% (p=0.000 n=20)
MemmoveUnalignedSrcOverlap/64 56.03n ± 0% 10.01n ± 0% -82.13% (p=0.000 n=20)
MemmoveUnalignedSrcOverlap/128 107.30n ± 0% 14.01n ± 0% -86.94% (p=0.000 n=20)
MemmoveUnalignedSrcOverlap/256 209.70n ± 0% 13.43n ± 1% -93.60% (p=0.000 n=20)
MemmoveUnalignedSrcOverlap/512 414.60n ± 0% 22.23n ± 0% -94.64% (p=0.000 n=20)
MemmoveUnalignedSrcOverlap/1024 828.40n ± 0% 37.62n ± 0% -95.46% (p=0.000 n=20)
MemmoveUnalignedSrcOverlap/2048 1648.00n ± 0% 68.04n ± 0% -95.87% (p=0.000 n=20)
MemmoveUnalignedSrcOverlap/4096 3287.0n ± 0% 128.9n ± 0% -96.08% (p=0.000 n=20)
geomean 48.94n 13.58n -72.26%
The relevant performance improved by 72.26%.
Change-Id: If2d3e09c3d687e733e6ff2c50feb8d6a8eb7e63b
Reviewed-on: https://go-review.googlesource.com/c/go/+/589537
Reviewed-by: Tim King <taking@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Qiqi Huang <huangqiqi@loongson.cn>
Reviewed-by: Meidan Li <limeidan@loongson.cn>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: abner chenc <chenguoqi@loongson.cn>
Auto-Submit: Tim King <taking@google.com>