Skip to content

[Hardware][PowerPC] Fix fp16 compilation error for Power in cpu attention backend and bump oneDNN version #28535

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

Merged
merged 4 commits into from
Nov 13, 2025
Merged

[Hardware][PowerPC] Fix fp16 compilation error for Power in cpu attention backend and bump oneDNN version #28535

Show file tree
Hide file tree
Changes from 3 commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
fa1e4d0
fix fp16 condition in cpu attn
Nov 12, 2025
fe8224e
chore: update oneDNN version
Nov 12, 2025
c2072bf
Merge branch 'main' into fix/fp16_onednn_Version
Akashcodes732 Nov 12, 2025
4a61398
chore: update ACL Version
Nov 13, 2025
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
2 changes: 1 addition & 1 deletion cmake/cpu_extension.cmake
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -272,7 +272,7 @@ if ((AVX512_FOUND AND NOT AVX512_DISABLED) OR (ASIMD_FOUND AND NOT APPLE_SILICON
FetchContent_Declare(
oneDNN
GIT_REPOSITORY https://github.com/oneapi-src/oneDNN.git
GIT_TAG v3.9
GIT_TAG v3.10
GIT_PROGRESS TRUE
GIT_SHALLOW TRUE
)
Expand Down
2 changes: 2 additions & 0 deletions csrc/cpu/cpu_attn_impl.hpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -821,10 +821,12 @@ struct VecTypeTrait<c10::BFloat16> {
using vec_t = vec_op::BF16Vec16;
};

#if !defined(__powerpc__)
Copy link
Contributor

@gemini-code-assist gemini-code-assist bot Nov 12, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

high

The fix is incomplete. The s390x architecture also lacks fp16 vector support (vec_op::FP16Vec16) and will fail to compile this specialization, similar to PowerPC. The preprocessor guard should be extended to also cover s390x. A similar guard #if defined(__powerpc64__) || defined(__s390x__) is used in csrc/cpu/dnnl_kernels.cpp, which indicates s390x has similar characteristics.

#if !defined(__powerpc__) && !defined(__s390x__)

template <>
struct VecTypeTrait<c10::Half> {
using vec_t = vec_op::FP16Vec16;
};
#endif

template <typename T>
void print_logits(const char* name, T* ptr, int32_t row, int32_t col,
Expand Down