enhancement of tune tool for better sgemm performance

clMathLibraries / clBLAS

a software library containing BLAS functions written in OpenCL

Apache License 2.0

839 stars 240 forks source link

enhancement of tune tool for better sgemm performance #47

Closed TimmyLiu closed 10 years ago

TimmyLiu commented 10 years ago

1, enable the tune tool to test all workgroup sizes that are multiple of wavefront size instead of a hard coded 64; 2, increase the tuning coverage for bigger workgroup sizes.