wgblas
    Preparing search index...

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

    Benchmark results for ddot on Nvidia Geforce Gtx 1650.

    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    32 0.0224 0.0229 0.0182 0.0281 81.5%
    64 0.0226 0.0454 0.0181 0.0564 80.5%
    128 0.0232 0.0882 0.0182 0.1127 78.2%
    512 0.0236 0.3476 0.0179 0.4580 75.9%
    1024 0.0233 0.7019 0.0181 0.9062 77.4%
    4096 0.0225 2.9091 0.0181 3.6120 80.5%
    16384 0.0241 10.8647 0.0181 14.4991 74.9%
    65536 0.0318 32.9327 0.0183 57.3870 57.4%
    262144 0.0512 81.8944 0.0432 97.0904 84.3%
    1048576 0.1200 139.8101 0.1113 150.7441 92.7%
    4194304 0.3970 169.0434 0.3792 176.9524 95.5%
    16777216 1.4836 180.9391 1.4542 184.5881 98.0%

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

    ddot-default GB/s chart

    ddot-default ms chart

    • ddot.js — WebGPU benchmark script
    • ddot.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 0.0201 0.0255 0.0188 0.0273 93.4%
    64 0.0199 0.0515 0.0188 0.0544 94.7%
    128 0.0201 0.1017 0.0188 0.1090 93.3%
    512 0.0199 0.4116 0.0188 0.4369 94.2%
    1024 0.0199 0.8232 0.0188 0.8730 94.3%
    4096 0.0207 3.1727 0.0187 3.5038 90.6%
    16384 0.0271 9.6775 0.0206 12.7403 76.0%
    65536 0.0518 20.2397 0.0423 24.7680 81.7%
    262144 0.1188 35.3103 0.1080 38.8189 91.0%
    1048576 0.3997 41.9766 0.3885 43.1797 97.2%
    4194304 1.5181 44.2069 1.4773 45.4273 97.3%

    ddot-stride4 GB/s chart

    ddot-stride4 ms chart

    Nvidia Geforce Gtx 1650 — stride = 5
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    32 0.0183 0.0280 0.0149 0.0344 81.5%
    64 0.0182 0.0563 0.0150 0.0681 82.7%
    128 0.0182 0.1127 0.0157 0.1303 86.5%
    512 0.0183 0.4487 0.0150 0.5476 81.9%
    1024 0.0183 0.8935 0.0146 1.1228 79.6%
    4096 0.0191 3.4391 0.0154 4.2534 80.9%
    16384 0.0283 9.2775 0.0257 10.1827 91.1%
    65536 0.0505 20.7458 0.0459 22.8508 90.8%
    262144 0.1358 30.8841 0.1270 33.0198 93.5%
    1048576 0.4937 33.9829 0.4640 36.1590 94.0%
    4194304 1.9176 34.9960 1.8006 37.2694 93.9%

    ddot-stride5 GB/s chart

    ddot-stride5 ms chart

    Nvidia Geforce Gtx 1650 — stride = 32
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    32 0.0181 0.0283 0.0180 0.0284 99.7%
    64 0.0181 0.0565 0.0176 0.0582 97.1%
    128 0.0183 0.1119 0.0180 0.1141 98.1%
    512 0.0184 0.4456 0.0177 0.4638 96.1%
    1024 0.0184 0.8889 0.0174 0.9412 94.4%
    4096 0.0266 2.4675 0.0176 3.7338 66.1%
    16384 0.0432 6.0637 0.0288 9.0871 66.7%
    65536 0.1117 9.3905 0.0593 17.6838 53.1%
    262144 0.3953 10.6114 0.1964 21.3559 49.7%
    1048576 1.8346 9.1449 — — —

    ddot-stride32 GB/s chart

    ddot-stride32 ms chart

    Nvidia Geforce Gtx 1650 — stride = 33
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    64 0.0305 0.0336 0.0182 0.0563 59.6%
    128 0.0309 0.0663 0.0179 0.1143 58.0%
    512 0.0306 0.2681 0.0181 0.4515 59.4%
    1024 0.0312 0.5257 0.0147 1.1167 47.1%
    4096 0.0379 1.7283 0.0194 3.3851 51.1%
    16384 0.0556 4.7189 0.0295 8.8947 53.1%
    65536 0.1242 8.4443 0.0596 17.5936 48.0%
    262144 0.4298 9.7578 0.1952 21.4855 45.4%

    ddot-stride33 GB/s chart

    ddot-stride33 ms chart

    Nvidia Geforce Gtx 1650 — stride = 255
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    512 0.0345 0.2374 0.0180 0.4551 52.2%
    1024 0.0336 0.4876 0.0181 0.9030 54.0%
    4096 0.0471 1.3918 0.0191 3.4362 40.5%
    16384 0.0786 3.3341 0.0399 6.5746 50.7%
    65536 0.2560 4.0960 0.1089 9.6263 42.6%

    ddot-stride255 GB/s chart

    ddot-stride255 ms chart

    Nvidia Geforce Gtx 1650 — stride = 256
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    512 0.0315 0.2600 0.0182 0.4507 57.7%
    1024 0.0321 0.5107 0.0181 0.9030 56.6%
    4096 0.0432 1.5170 0.0187 3.5068 43.3%
    16384 0.0703 3.7279 0.0394 6.6494 56.1%
    65536 0.1878 5.5832 0.0902 11.6199 48.0%

    ddot-stride256 GB/s chart

    ddot-stride256 ms chart

    See also:

    Settings
    Member Visibility

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