Matrix multiplication. Wider loads with register blocking v.7 benchmark

WebGL2-compute based NxN matrix multiplication C = A x B Shader v.7 benchmark. Fixed random A, B 4096x4096 buffers are used for calculations, therefore You can set new N value (up to N=4096, multiple of 128). FLOPS = 2 N3 / time. Random CGPU and CGPU - CCPU are shown.


GeForce GT 710, D3D11     ~104 GFLOPS (N=4096).

I re-run all programs many times (press "Run" button) and try to choose "representative" values. Results differ very much on different GPUs.

SGEMM in WebGL2-compute     updated 7 Apr 2019