[CK_tile] Add rotating buffer feature for universal gemm (#2200)

* Add rotating buffer feature for universal gemm

* adding changes in tile_engine

* Updated code to merge kernel_launch

* removing comments

* Enable rotating buffer changes to flatmm

* Created diff launch_kernel function for rotating buffer

* Simplfied calculation using macros

* merge code with new changes in tile_engine

* clang formatted

* Redefine macros
This commit is contained in:
Khushbu Agarwal
2025-05-27 23:00:58 -07:00
committed by GitHub
parent c52649ad57
commit 99857e10e6
17 changed files with 409 additions and 74 deletions

View File

@@ -178,7 +178,7 @@ float invoke_gemm(ck_tile::DeviceMem& a_m_k_dev_buf,
float ave_time =
gemm_calc<ADataType, BDataType, AccDataType, CDataType, ALayout, BLayout, CLayout>(
args, ck_tile::stream_config{nullptr, true, 1, n_warmup, n_repeat});
args, ck_tile::stream_config{nullptr, true, 1, n_warmup, n_repeat, true, true, 50});
std::size_t flop = std::size_t(2) * M * N * K;
std::size_t num_byte =