zjing14
cab6416fa5
Grouped Gemm device with multiD grid (#319)
* replace gridwise_v2r3 with multiD
* adjust parameters
* add instances
* fixed test_grouped_gemm
* fix standalone softmax race condition around blockwise reduction
* fixed ci
* fixed comment: remove redundant workspace
* use instanceFactory
* add test layout
* add empty Ds
* add bias example
* use array
* sperate examples
Co-authored-by: Anthony Chang <ac.chang@outlook.com>
[ROCm/composable_kernel commit: 7959dad566]
2022-07-21 10:07:01 -05:00
..
2022-07-08 15:55:14 -05:00
2022-07-07 14:31:11 -05:00
2022-07-07 14:31:11 -05:00
2022-07-07 14:31:11 -05:00
2022-06-24 23:32:43 -05:00
2022-06-24 23:32:43 -05:00
2022-06-24 23:32:43 -05:00
2022-06-24 23:32:43 -05:00
2022-06-24 23:32:43 -05:00
2022-06-24 23:32:43 -05:00
2022-06-24 23:32:43 -05:00
2022-06-24 23:32:43 -05:00
2022-07-21 10:07:01 -05:00
2022-06-27 14:25:10 -05:00
2022-06-24 23:32:43 -05:00
2022-06-27 14:25:10 -05:00
2022-06-27 14:25:10 -05:00
2022-06-24 23:32:43 -05:00
2022-07-13 11:16:14 -05:00
2022-06-24 23:32:43 -05:00
2022-07-13 11:16:14 -05:00
2022-07-06 10:38:29 -05:00
2022-06-30 19:55:09 -05:00
2022-07-08 15:55:14 -05:00
2022-07-13 11:16:14 -05:00
2022-07-21 10:07:01 -05:00
2022-07-21 10:07:01 -05:00