ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-04-20 06:49:15 +00:00
Code Issues Packages Projects Releases Wiki Activity

fixed fp8 issues (#894)

Browse Source 
* fixed fp8 init; and reference gemm

* Update host_tensor_generator.hpp

* fixed convert

* fixed reference gemm

* fixed comments

* fixed comments

* fixed ci

* fixed computeType

---------

Co-authored-by: Jing Zhang <jizha@amd.com>
This commit is contained in:
zjing14
2023-09-12 22:17:56 -05:00
committed by GitHub
parent 74d32f0719
commit a66d14edf2
6 changed files with 45 additions and 25 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
include/ck/tensor_operation/gpu/thread/threadwise_tensor_slice_transfer.hpp
View File

@@ -137,13 +137,12 @@ struct ThreadwiseTensorSliceTransfer_v1r3
constexpr index_t src_offset = src_desc.CalculateOffset(
src_slice_origin_idx + idx_md + i * dst_scalar_step_in_vector);
SrcData v;
DstData v;
// apply element-wise operation
element_op_(v, src_buf[Number<src_offset>{}]);
// apply type convert
dst_vector.template AsType<DstData>()(i) = type_convert<DstData>(v);
dst_vector.template AsType<DstData>()(i) = v;
});
const bool is_dst_valid =
@@ -1289,13 +1288,13 @@ struct ThreadwiseTensorSliceTransfer_StaticToStatic
constexpr index_t dst_offset = dst_desc.CalculateOffset(
dst_slice_origin_idx + idx_md + i * dst_scalar_step_in_vector);
SrcData v;
DstData v;
// apply element-wise operation
element_op_(v, src_buf[Number<src_offset>{}]);
// apply type convert
dst_buf(Number<dst_offset>{}) = type_convert<DstData>(v);
dst_buf(Number<dst_offset>{}) = v;
});
});
}