mirror of
https://github.com/ROCm/composable_kernel.git
synced 2026-05-03 05:01:25 +00:00
[CK TILE] GEMM and Batched GEMM SplitK support (#1724)
* [CK TILE] Add split K support in GEMM * Updates * Fixes * rebase * fix * Fix * fixes * support for batched gemm
This commit is contained in:
@@ -54,8 +54,7 @@ using CDataType = Types::CDataType;
|
||||
auto create_args(int argc, char* argv[])
|
||||
{
|
||||
ck_tile::ArgParser arg_parser;
|
||||
arg_parser.insert("b", "1", "batch size")
|
||||
.insert("m", "3840", "m dimension")
|
||||
arg_parser.insert("m", "3840", "m dimension")
|
||||
.insert("n", "4096", "n dimension")
|
||||
.insert("k", "2048", "k dimension")
|
||||
.insert("a_layout", "R", "A tensor data layout - Row by default")
|
||||
@@ -68,7 +67,8 @@ auto create_args(int argc, char* argv[])
|
||||
.insert("prec", "fp16", "data type. fp16/bf16/fp8/bf8")
|
||||
.insert("warmup", "50", "number of iterations before benchmark the kernel")
|
||||
.insert("repeat", "100", "number of iterations to benchmark the kernel")
|
||||
.insert("timer", "gpu", "gpu:gpu timer, cpu:cpu timer");
|
||||
.insert("timer", "gpu", "gpu:gpu timer, cpu:cpu timer")
|
||||
.insert("split_k", "1", "splitK value");
|
||||
|
||||
bool result = arg_parser.parse(argc, argv);
|
||||
return std::make_tuple(result, arg_parser);
|
||||
|
||||
Reference in New Issue
Block a user