Ville Pietilä
|
3aec8b5493
|
Improve profiler output.
|
2025-10-20 08:04:59 +00:00 |
|
Ville Pietilä
|
dc65dc98e1
|
Optimize calculation of the CPU reference.
|
2025-10-17 14:48:12 +00:00 |
|
Ville Pietilä
|
28055fdd9a
|
Improve profiler output.
|
2025-10-17 13:17:29 +00:00 |
|
Ville Pietilä
|
7722f901df
|
Fix validation.
|
2025-10-17 13:07:06 +00:00 |
|
Ville Pietilä
|
ef3e871e6e
|
Add grouped conv fwd direction profiling into CK Tile profiler.
|
2025-10-17 10:47:23 +00:00 |
|
Ville Pietilä
|
0e0fb54b9f
|
Rename conv factory.
|
2025-10-17 06:26:41 +00:00 |
|
Ville Pietilä
|
6c5531a4ae
|
Disqualify benchmarking results from kernels that do not pass validation.
|
2025-10-16 12:22:51 +00:00 |
|
Ville Pietilä
|
96a7c26a0b
|
Better split-K handling in the template instantiation.
|
2025-10-15 13:47:04 +00:00 |
|
Ville Pietilä
|
3d0db2ca63
|
Fix transferring data back to host for validation.
|
2025-10-14 15:02:51 +00:00 |
|
Ville Pietilä
|
bbed3a62dc
|
Fully functional CK Tile profiler.
|
2025-10-14 13:35:37 +00:00 |
|
Ville Pietilä
|
eaf9ba4e45
|
Rename CK Tile grouped conv factory.
|
2025-10-14 06:31:34 +00:00 |
|
Ville Pietilä
|
fc6a9e3931
|
Create invoker for the kernel and a factory for creating invokers.
|
2025-10-13 15:22:50 +00:00 |
|
Ville Pietilä
|
a60dab521e
|
Added a placeholder conv bwd instance factory for CK Tile profiler.
|
2025-10-13 14:32:20 +00:00 |
|
Ville Pietilä
|
6dcee56fee
|
WIP: CK Tile conv bwd profiler.
|
2025-10-13 13:03:21 +00:00 |
|
Ville Pietilä
|
d62f34348a
|
Skeleton for the ckTileProfiler.
|
2025-10-13 11:40:31 +00:00 |
|