Emily Martins 02ab76c2cb Fix CK Tile DP + 2 Tile Stream-K Validation Errors (#3269) ...

When there are multiple workgroups contributing to a tile, when using
atomics, there may be round off error in cases where the accumulator
type is not the same as the C type. To compute an error tolerance for
test validation, the Stream-K Tile Partitioner has a function called
estimate_num_wgs_per_tile to estimate the number of workgroups per tile.
That said, this function only provides an estimate. In some cases for
DP+2TSK, the function returns 1 rather than the more accurate value of
2.

Thus, this change updates the estimate_num_wgs_per_tile function to
explicitely return the value of 2 in cases for DP+2TSK to ensure that we
have a better error tolerance to avoid test failures due to round-off
error.

2025-11-21 20:29:47 -07:00

..

block

Remove "basic" and universal GEMM tests, and incorporate their test cases into the GEMM pipeline tests (#3094)

2025-11-13 11:01:27 -08:00

kernel

Fix CK Tile DP + 2 Tile Stream-K Validation Errors (#3269)

2025-11-21 20:29:47 -07:00

pipeline

[CK_TILE] Non-K Major from old CK to CK-Tile - fix reverted PR (#3199)

2025-11-18 10:17:02 +02:00

warp

[CK_TILE] Improve F8F6F4 Scaled WarpGemm (#3197)

2025-11-13 20:22:05 +08:00