7 Commits

Author	SHA1	Message	Date
Anthony Chang	7db48f9008	Tune & add conflict-free LDS gemm kernels (#159) ... * retune & add conflict-free bf16/fp16 c-shuffle gemm instances amend wrong K1 value in some fp16/bf16 kernel instances * make gemm cshuffle's timing behavior consistent with all other functions * clang-format * retune & add conflict-free fp32 c-shuffle gemm instances * retune & add conflict-free int8 c-shuffle gemm instances * update the underlying gridwise gemm of all c-shuffle gemm kernels * typo	2022-03-31 12:58:41 -05:00
Chao Liu	cd167e492a	Compile for gfx908 and gfx90a (#130) ... * adding compilation for multiple targets * fix build * clean * update Jekinsfile * update readme * update Jenkins * use ck::half_t instead of ushort for bf16 * rename enum classes * clean * rename * clean	2022-03-31 12:33:34 -05:00
Jianfeng Yan	34c661e71c	Batched gemm and reduction (#156) ... * adding batched_gemm_and_reduction * batched_gemm_reduce works with bactch_count=1 * fix a bug in grid_size; batched_gemm_reduce works for batch_count > 1 * adding profiler for batched_gemm_fp16 * fixed a bug in declaration of d1 and d0; both example and profiler work * clang-format * cleanup * batched_gemm_reduce: add test * minor change * fixed some typo in function names	2022-03-30 11:21:18 -05:00
rocking5566	98e1e2d0e9	Refine kernel parameter of int8 (ScalarPerVector) (#155) ... * Change int8 ScalarPerVector * Modify vector width of C	2022-03-29 17:36:21 -05:00
Chao Liu	f95267f166	Gemm+Reduce Fusion (#128) ... * add gridwise gemm v4r1 * rename * adding gemm+reduce * adding gemm+reduce * adding gemm+reduce * adding gemm+reduce * use sfc in shuffling * remove hardcode * remove hardcode * refactor * fix build * adding gemm+reduce * adding gemm+reduce * adding gemm+reduce * adding gemm+reduce * adding gemm+reduce * format * clean * adding gemm+reduce * adding profiler for gemm+reduce * adding gemm+reduce profiler * fix build * clean up * gemm+reduce * fix build * update DeviceGemm_Xdl_CShuffle; update enum to enum class * clean up * add test for gemm+reduce * clean up * refactor * fix build * fix build	2022-03-23 22:18:42 -05:00
rocking5566	9a17e7fbfd	Consider gemm requant relu requant as gemm fusuion (#116) ... * [What] Separate fixpoint gemm from gemm example [Why] let example of gemm_int8 be pure gemm. [What] 1. Add gemm_requant_relu_requant, 2. Let CDataType be int32 in pure gemm, because no one use int8 CDataType. It is also part of gemm_requant_relu_requant * Fix path * Revise cmakelist due to merge develop Co-authored-by: rocking <chunylai@amd.com>	2022-03-11 20:41:03 -06:00
Chao Liu	5d37d7bff4	Reorganize files, Part 1 (#119) ... * delete obselete files * move files * build * update cmake * update cmake * fix build * reorg examples * update cmake for example and test	2022-03-08 21:46:36 -06:00