Matrix multiplication SLM benchmarks

calculating
N=

WebGL2-compute matrix multiplication C = A x B (SGEMM) demo. Based on Intel highly optimised OpenCL kernel SLM_8x8_4x16. All A, B elements are random (0 - 1). N ≤ 1024 is multiple of 128.

On Intel i3-810 (Windows 10 64-bit with D3D11 backend) I get ~208 GFLOPS.


SGEMM in WebGL2-compute     updated 28 June 2019