Alex Brown

Reduces data copying and comparisons when reading arrays, maps, and enums.

Script to remove kernels from a logic file that are not used

6

Script to remove kernels from a logic file that are not used by any of the tuned sizes

Two-tile algorithm with SK after DP

1

Alternative implementation of the 2-tile algorithm that does DP tiles first and SK tiles after. This method should have a small boost in performance.

Fix short circuit logic for alpha=0 with stream-k

2

This update fixes the case when alpha=0 by ensuring that A/B matrices are not read and main loop does not run. Also added a new small test case with stream-k...

Disable GSU code when GSO=0

4

GSU=0 should disable all GSU code. This change updates some sections of code that were still generating GSU-related code when GSU was disabled.

Allow wavegroup to be less than 4 in stream-k kernels. This change updates the partials and fixup code to tkae number of waves into account. Added new test cases to...