Age | Commit message (Collapse) | Author |
|
i5-4570
|
|
Cuda API to CLBlast
|
|
re-compilation every time
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
changed transpose kernel code
|
|
|
|
(default ON)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
direct/in-direct GEMM kernels separately
|
|
|
|
|
|
|
|
Cuda API preparation
|
|
it belongs
|
|
|
|
Single temporary GEMM buffer
|
|
|
|
optional temporary buffers
|
|
|
|
|
|
|
|
Preparation for size specific parameters
|
|
|
|
|
|
|
|
|
|
|
|
|
|
of the name
|
|
|
|
|
|
device/platform IDs
|