Skip to content

fix: mv #include <musa_fp16.h> ahead of polyfills.cuh #32

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
zhangyue207 wants to merge 1 commit into
base: feat/dev-infra
Choose a base branch
from
+3 −22
Open

fix: mv #include <musa_fp16.h> ahead of polyfills.cuh #32

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions src/common/cuda/kernel_commons.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -21,6 +21,7 @@ using cuda_bfloat162 = maca_bfloat162;
#include <musa_bf16.h>
#include <musa_fp16.h>
#include <musa_runtime.h>
#include "moore/polyfills.cuh"
using cuda_bfloat16 = __mt_bfloat16;
using cuda_bfloat162 = __mt_bfloat162;
#endif
Expand Down
8 changes: 0 additions & 8 deletions src/moore/add/kernel.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3,14 +3,6 @@

#include <utility>

// clang-format off
#include <musa_runtime.h>
// clang-format on

// clang-format off
#include "moore/polyfills.cuh"
// clang-format on

#include "cuda/add/kernel.h"

namespace infini::ops {
Expand Down
8 changes: 2 additions & 6 deletions src/moore/polyfills.cuh
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,12 +1,6 @@
#ifndef INFINI_OPS_MOORE_POLYFILLS_CUH_
#define INFINI_OPS_MOORE_POLYFILLS_CUH_

#include <type_traits>

// clang-format off
#include <musa_bf16.h>
// clang-format on

namespace infini::ops {

template <typename T>
Expand Down Expand Up @@ -36,6 +30,8 @@ __device__ __forceinline__ T hrcp(const T& a) {

} // namespace infini::ops

// Force unqualified `hrcp` calls to resolve to our polyfill, bypassing ADL
// which would otherwise pick the MUSA SDK's declared-but-undefined hrcp(__half).
#define hrcp infini::ops::hrcp

#endif
8 changes: 0 additions & 8 deletions src/moore/swiglu/kernel.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3,14 +3,6 @@

#include <utility>

// clang-format off
#include <musa_runtime.h>
// clang-format on

// clang-format off
#include "moore/polyfills.cuh"
// clang-format on

#include "cuda/swiglu/kernel.h"

namespace infini::ops {
Expand Down