wgblas
    Preparing search index...

    Module benchmarks/nvidia-geforce-gtx-1650/dcopy

    Benchmark results for dcopy on Nvidia Geforce Gtx 1650.

    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    32 0.0069 0.0739 0.0034 0.1502 49.2%
    64 0.0070 0.1468 0.0033 0.3092 47.5%
    128 0.0072 0.2838 0.0033 0.6275 45.2%
    512 0.0073 1.1155 0.0034 2.3925 46.6%
    1024 0.0072 2.2806 0.0033 5.0196 45.4%
    4096 0.0076 8.6780 0.0034 19.3208 44.9%
    16384 0.0081 32.1886 0.0040 65.5360 49.1%
    65536 0.0119 87.9678 0.0075 140.6352 62.6%
    262144 0.0304 138.1159 0.0276 152.0557 90.8%
    1048576 0.1039 161.4187 0.0996 168.3918 95.9%
    4194304 0.3965 169.2480 0.3872 173.3327 97.6%
    16777216 1.5647 171.5585 1.5631 171.7323 99.9%

    Efficiency = wgblas GB/s ÷ cuBLAS GB/s × 100. 100% means parity with cuBLAS; values above 100% mean wgblas achieved greater throughput.

    dcopy-default GB/s chart

    dcopy-default ms chart

    • dcopy.js — WebGPU benchmark script
    • dcopy.c — CUDA / cuBLAS reference script

    Unless noted otherwise, every result above uses unit stride (incx = incy = 1) — the normal case, and the coalesced, best-case GPU access pattern. Real usage sometimes passes a non-unit stride (e.g. operating on a row or column of a larger matrix, where incx = lda), which breaks memory coalescing and costs measurably more. This section sweeps a few representative strides to characterize that cost separately, collapsed below by default — expand a stride to see its table and chart.

    Nvidia Geforce Gtx 1650 — stride = 4
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    32 0.0060 0.0860 0.0034 0.1509 57.0%
    64 0.0061 0.1684 0.0035 0.2963 56.8%
    128 0.0061 0.3377 0.0036 0.5766 58.6%
    512 0.0060 1.3690 0.0037 2.1880 62.6%
    1024 0.0061 2.6806 0.0040 4.1290 64.9%
    4096 0.0063 10.4224 0.0041 16.0000 65.1%
    16384 0.0095 27.6757 0.0064 40.6551 68.1%
    65536 0.0396 26.5006 0.0410 25.6000 103.5%
    262144 0.1551 27.0503 0.1548 27.1034 99.8%
    1048576 0.6157 27.2506 0.6083 27.5825 98.8%
    4194304 2.4394 27.5103 2.4390 27.5146 100.0%

    dcopy-stride4 GB/s chart

    dcopy-stride4 ms chart

    Nvidia Geforce Gtx 1650 — stride = 5
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    32 0.0060 0.0849 0.0028 0.1860 45.6%
    64 0.0060 0.1711 0.0028 0.3699 46.3%
    128 0.0060 0.3432 0.0028 0.7191 47.7%
    512 0.0060 1.3690 0.0032 2.5347 54.0%
    1024 0.0061 2.6667 0.0034 4.8302 55.2%
    4096 0.0065 10.0392 0.0035 18.5339 54.2%
    16384 0.0117 22.3215 0.0092 28.4939 78.3%
    65536 0.0492 21.3333 0.0439 23.9008 89.3%
    262144 0.1965 21.3420 0.1685 24.8950 85.7%
    1048576 0.7803 21.5013 0.6676 25.1288 85.6%

    dcopy-stride5 GB/s chart

    dcopy-stride5 ms chart

    Nvidia Geforce Gtx 1650 — stride = 32
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    32 0.0060 0.0851 0.0028 0.1839 46.3%
    64 0.0061 0.1675 0.0028 0.3596 46.6%
    128 0.0061 0.3333 0.0030 0.6809 49.0%
    512 0.0063 1.3028 0.0037 2.2358 58.3%
    1024 0.0065 2.5347 0.0036 4.5714 55.4%
    4096 0.0105 6.2534 0.0046 14.3217 43.7%
    16384 0.0368 7.1173 0.0217 12.0826 58.9%
    65536 0.1398 7.5010 0.0817 12.8301 58.5%
    262144 0.5712 7.3424 0.3276 12.8038 57.3%

    dcopy-stride32 GB/s chart

    dcopy-stride32 ms chart

    Nvidia Geforce Gtx 1650 — stride = 33
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    64 0.0092 0.1113 0.0029 0.3497 31.8%
    128 0.0094 0.2188 0.0029 0.7111 30.8%
    512 0.0098 0.8380 0.0035 2.3704 35.4%
    1024 0.0102 1.6000 0.0035 4.7407 33.8%
    4096 0.0164 4.0000 0.0041 16.0000 25.0%
    16384 0.0445 5.8851 0.0240 10.9009 54.0%
    65536 0.1777 5.9020 0.0910 11.5238 51.2%
    262144 0.8093 5.1825 0.3737 11.2243 46.2%

    dcopy-stride33 GB/s chart

    dcopy-stride33 ms chart

    Nvidia Geforce Gtx 1650 — stride = 255
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    512 0.0104 0.7901 0.0036 2.3063 34.3%
    1024 0.0111 1.4776 0.0036 4.5511 32.5%
    4096 0.0205 3.2000 0.0042 15.6935 20.4%
    16384 0.0796 3.2919 0.0355 7.3868 44.6%
    65536 0.3502 2.9942 0.1405 7.4608 40.1%

    dcopy-stride255 GB/s chart

    dcopy-stride255 ms chart

    Nvidia Geforce Gtx 1650 — stride = 256
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    512 0.0103 0.7975 0.0035 2.3273 34.3%
    1024 0.0115 1.4242 0.0036 4.5919 31.0%
    4096 0.0184 3.5556 0.0043 15.1704 23.4%
    16384 0.0737 3.5556 0.0341 7.6812 46.3%
    65536 0.3745 2.8002 0.1372 7.6418 36.6%

    dcopy-stride256 GB/s chart

    dcopy-stride256 ms chart

    See also:

    Settings
    Member Visibility

    On This Page
    Nvidia Geforce Gtx 1650 — wgblas vs cuBLASSee alsoStride sweep