wgblas
    Preparing search index...

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

    Benchmark results for sger on Nvidia Geforce Gtx 1650.

    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    32 0.0075 1.1210 0.0041 2.0625 54.4%
    64 0.0075 4.4255 0.0041 8.1250 54.5%
    128 0.0082 16.1250 0.0041 31.8764 50.6%
    256 0.0101 52.0506 0.0059 89.1490 58.4%
    512 0.0184 114.0000 0.0152 138.6779 82.2%
    1024 0.0607 138.3601 0.0621 135.1532 102.4%
    1280 0.0935 140.3595 0.0962 136.3900 102.9%
    2048 0.2232 150.3853 0.2356 142.4810 105.5%
    4096 0.8588 156.3146 0.9321 144.0258 108.5%

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

    sger-default GB/s chart

    sger-default ms chart

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

    Unless noted otherwise, every result above uses a tight lda (no padding). Padding the row stride changes throughput here — the exact mechanism and shape of that effect is routine-specific — collapsed below by default, expand a pad value to see its table and chart.

    Nvidia Geforce Gtx 1650 — pad = 0
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    32 0.0071 1.1839 0.0034 2.4789 47.8%
    64 0.0072 4.6222 0.0036 9.2035 50.2%
    128 0.0077 17.2000 0.0036 36.2105 47.5%
    256 0.0095 55.4739 0.0045 117.4857 47.2%
    512 0.0176 119.6066 0.0128 164.1600 72.9%
    1024 0.0589 142.5700 0.0557 150.7180 94.6%
    1280 0.0922 142.3333 0.0876 149.6878 95.1%
    2048 0.2211 151.8107 0.2131 157.5089 96.4%
    4096 0.8489 158.1531 0.8476 158.3800 99.9%

    sger-pad0 GB/s chart

    sger-pad0 ms chart

    Nvidia Geforce Gtx 1650 — pad = 1
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    32 0.0063 1.3401 0.0035 2.4332 55.1%
    64 0.0064 5.2000 0.0033 9.9522 52.2%
    128 0.0068 19.2897 0.0036 36.3700 53.0%
    256 0.0083 63.5058 0.0050 104.7643 60.6%
    512 0.0188 111.6735 0.0147 142.9032 78.1%
    1024 0.1063 78.9648 0.0741 113.3233 69.7%
    1280 0.1676 78.2663 0.1167 112.3684 69.7%
    2048 0.4064 82.6021 0.2945 114.0065 72.5%
    4096 1.5194 88.3599 1.1817 113.6083 77.8%

    sger-pad1 GB/s chart

    sger-pad1 ms chart

    Nvidia Geforce Gtx 1650 — pad = 8
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    32 0.0061 1.3750 0.0035 2.4332 56.5%
    64 0.0061 5.4167 0.0035 9.6296 56.3%
    128 0.0067 19.8462 0.0036 36.8571 53.8%
    256 0.0082 64.2500 0.0045 115.8310 55.5%
    512 0.0201 104.5605 0.0166 126.7645 82.5%
    1024 0.1044 80.3922 0.0735 114.2360 70.4%
    1280 0.1707 76.8648 0.1154 113.6772 67.6%
    2048 0.3840 87.4204 0.2973 112.9084 77.4%
    4096 1.4808 90.6588 1.1733 114.4186 79.2%

    sger-pad8 GB/s chart

    sger-pad8 ms chart

    Nvidia Geforce Gtx 1650 — pad = 16
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    32 0.0061 1.3750 0.0035 2.4332 56.5%
    64 0.0061 5.4167 0.0035 9.5853 56.5%
    128 0.0066 19.9420 0.0036 36.6933 54.3%
    256 0.0082 64.2500 0.0046 114.6202 56.1%
    512 0.0207 101.4900 0.0182 115.6056 87.8%
    1024 0.1024 82.0385 0.0735 114.2360 71.8%
    1280 0.1622 80.8920 0.1165 112.5999 71.8%
    2048 0.3942 85.1567 0.2934 114.4293 74.4%
    4096 1.4631 91.7573 1.1854 113.2541 81.0%

    sger-pad16 GB/s chart

    sger-pad16 ms chart

    Nvidia Geforce Gtx 1650 — pad = 32
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    32 0.0061 1.3750 0.0034 2.4906 55.2%
    64 0.0061 5.4167 0.0035 9.6296 56.3%
    128 0.0067 19.8462 0.0037 35.2821 56.2%
    256 0.0082 64.3757 0.0045 115.8310 55.6%
    512 0.0165 127.2558 0.0150 139.8594 91.0%
    1024 0.0574 146.1838 0.0561 149.7290 97.6%
    1280 0.0881 148.9535 0.0880 149.1160 99.9%
    2048 0.2171 154.6301 0.2191 153.1963 100.9%
    4096 0.8783 152.8603 0.8622 155.7055 98.2%

    sger-pad32 GB/s chart

    sger-pad32 ms chart

    Nvidia Geforce Gtx 1650 — pad = 48
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    32 0.0061 1.3750 0.0035 2.4110 57.0%
    64 0.0061 5.4167 0.0034 9.8113 55.2%
    128 0.0066 19.9420 0.0036 36.8571 54.1%
    256 0.0082 64.2500 0.0046 115.0210 55.9%
    512 0.0225 93.2727 0.0187 112.1503 83.2%
    1024 0.1063 78.9767 0.0746 112.5938 70.1%
    1280 0.1659 79.0741 0.1176 111.4973 70.9%
    2048 0.3963 84.7166 0.2949 113.8333 74.4%
    4096 1.4950 89.7973 1.1869 113.1121 79.4%

    sger-pad48 GB/s chart

    sger-pad48 ms chart

    Nvidia Geforce Gtx 1650 — pad = 64
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    32 0.0061 1.3750 0.0035 2.4220 56.8%
    64 0.0061 5.4167 0.0035 9.6296 56.3%
    128 0.0066 20.0388 0.0035 37.3575 53.6%
    256 0.0081 64.7559 0.0046 113.4345 57.1%
    512 0.0164 128.2500 0.0150 140.0085 91.6%
    1024 0.0591 142.0682 0.0567 147.9977 96.0%
    1280 0.0889 147.5328 0.0886 148.0123 99.7%
    2048 0.2252 149.0605 0.2229 150.5796 99.0%
    4096 0.8747 153.4867 0.8801 152.5435 100.6%

    sger-pad64 GB/s chart

    sger-pad64 ms chart

    Nvidia Geforce Gtx 1650 — pad = 128
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    32 0.0061 1.3750 0.0035 2.4332 56.5%
    64 0.0062 5.3333 0.0038 8.8511 60.3%
    128 0.0067 19.8462 0.0037 35.5862 55.8%
    256 0.0082 64.2500 0.0046 113.8270 56.4%
    512 0.0166 126.5202 0.0151 138.8245 91.1%
    1024 0.0580 144.6527 0.0570 147.2090 98.3%
    1280 0.0891 147.2943 0.0879 149.1703 98.7%
    2048 0.2281 147.1888 0.2275 147.5926 99.7%
    4096 0.9418 142.5503 0.9053 148.2972 96.1%

    sger-pad128 GB/s chart

    sger-pad128 ms chart

    See also:

    • lda.sger.js — WebGPU lda-sweep benchmark script
    • lda.sger.c — CUDA / cuBLAS lda-sweep reference script

    Settings
    Member Visibility

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