ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-14 10:09:41 +00:00
Code Issues Packages Projects Releases Wiki Activity

Add example of Gemm + AddAddFastGelu (data type: int4) (#369)

Browse Source 
* Add custom target to bundle examples together

* Add int4 example conditionally (just copy from int8 example)

* Extract common code into common.hpp

* Move ref gemm type alias into data-type-specific sources

* Add #error directive to prevent compile with wrong setting

* Let AddAddFastGelu support int4 parameter type

* Let check_err() support int4 parameter type

* Add wrapper function to hide value conversion while copying memory

* Finish int4 example for GEMM + AddAddFastGelu

* Add new DeviceMem API to copy memory

* Use new DeviceMem API to implement examples

* Fix wrongly use of macro 'CK_EXPERIMENTAL_BIT_INT_EXTENSION_INT4'

* Revert "Add new DeviceMem API to copy memory"

This reverts commit e26e7af71e.

* Add conversion ctor for Tensor<>

* Add 'const' specifier to Tensor<>::CopyAsType()

* Convert Tensor<> values before/after transfer between host & device

[ROCm/composable_kernel commit: 2327f1a640]
This commit is contained in:
Po Yen Chen
2022-08-23 23:38:41 +08:00
committed by GitHub
parent 1fbd80b0a0
commit 4e53b7beea
11 changed files with 267 additions and 192 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
include/ck/tensor_operation/gpu/element/element_wise_operation.hpp
View File

@@ -177,7 +177,11 @@ struct AddAddFastGelu
template <typename T>
static inline constexpr bool is_valid_param_type_v =
std::is_same_v<T, float> || std::is_same_v<T, half_t> || std::is_same_v<T, bhalf_t> ||
std::is_same_v<T, int32_t> || std::is_same_v<T, int8_t>;
std::is_same_v<T, int32_t> || std::is_same_v<T, int8_t>
#ifdef CK_EXPERIMENTAL_BIT_INT_EXTENSION_INT4
|| std::is_same_v<T, ck::int4_t>
#endif
;
template <typename E, typename C, typename D0, typename D1>
__host__ __device__ constexpr void