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

Add e8m0 scaled convert into CK_TILE (#2617)

Browse Source 
* first commit

* remove redundent code

* modify according to comments.

* fix type_convert error with scaled_type_convert
This commit is contained in:
Gino Lu
2025-08-07 21:37:28 +08:00
committed by GitHub
parent b0a97498b0
commit 5d6d236b25
8 changed files with 419 additions and 86 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
include/ck_tile/host/host_tensor.hpp
View File

@@ -409,7 +409,13 @@ struct HostTensor
}
// void SetZero() { ck_tile::ranges::fill<T>(mData, 0); }
void SetZero() { std::fill(mData.begin(), mData.end(), 0); }
void SetZero()
{
if constexpr(std::is_same_v<T, e8m0_t>)
std::fill(mData.begin(), mData.end(), e8m0_t{1.f});
else
std::fill(mData.begin(), mData.end(), 0);
}
template <typename F>
void ForEach_impl(F&& f, std::vector<size_t>& idx, size_t rank)