Fix kernel crash on nvidia, caused by float4 alignement error

clMathLibraries / clBLAS

a software library containing BLAS functions written in OpenCL

Apache License 2.0

839 stars 240 forks source link

Fix kernel crash on nvidia, caused by float4 alignement error #109

Closed hughperkins closed 8 years ago

hughperkins commented 9 years ago

Fix kernel crash on nvidia, caused by float4 alignement error, see https://github.com/clMathLibraries/clBLAS/issues/108 for more details.

Excuse my forking off 'master', but 'develop' didnt build for me.

I reckon this alignment issue potentially might be behind a bunch of exisitng issue reports for both nvidia and intel potentially. I suspect it might generalize to some other kernels too plausibly.

TimmyLiu commented 8 years ago

I will close this one since it seems to be a master branch version of pr #115

hughperkins commented 8 years ago

Yes, sounds good.