Files
composable_kernel/library/include
Bartłomiej Kocot d63ce9c186 Grouped convolution backward weight special vector size loads (#1772)
* Grouped convolution backward weight special vector size loads

* Instnaces and tests

* Fixes

* Add 7 and 13 special cases

* fix comments

* Fix

* Fix2

* fixes

* fix atomic add bf16

[ROCm/composable_kernel commit: fd46a01d8b]
2025-01-10 22:02:30 +08:00
..