I am currently trying to implement a backend for clBLAS version 2.10. my development system only has a cpu implementation of opencl, unfortunately clblasSgemm fails, even in the example code.
Steps to reproduce:
run example_sgemm.c while using CL_DEVICE_TYPE_CPU
result:
clblasSgemmEx result:
11 12 13
21 440 460
31 640 660
41 840 860
expected:
something where the 4 lower right entries are not C*beta.
Hi,
I am currently trying to implement a backend for clBLAS version 2.10. my development system only has a cpu implementation of opencl, unfortunately clblasSgemm fails, even in the example code.
Steps to reproduce: run example_sgemm.c while using CL_DEVICE_TYPE_CPU result: clblasSgemmEx result: 11 12 13 21 440 460 31 640 660 41 840 860 expected: something where the 4 lower right entries are not C*beta.