happyz
/
llama.cpp
mirror of https://github.com/ggerganov/llama.cpp.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

ggml-webgpu: Fix GGML_MEM_ALIGN to 8 for emscripten. (#18628) 

* Fix GGML_MEM_ALIGN to 8 for emscripten.

* Add a comment explaining the need for GGML_MEM_ALIGN == 8 in 64-bit wasm with emscripten
This commit is contained in:
Masashi Yoshimura 2026-01-09 01:36:42 +09:00 committed by GitHub
parent 15bff84bf5
commit 480160d472
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 5 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
ggml/include/ggml.h
View File

@ -234,6 +234,11 @@
#if UINTPTR_MAX == 0xFFFFFFFF
#define GGML_MEM_ALIGN 4
#elif defined(__EMSCRIPTEN__)
// emscripten uses max_align_t == 8, so we need GGML_MEM_ALIGN == 8 for 64-bit wasm.
// (for 32-bit wasm, the first conditional is true and GGML_MEM_ALIGN stays 4.)
// ref: https://github.com/ggml-org/llama.cpp/pull/18628
#define GGML_MEM_ALIGN 8
#else
#define GGML_MEM_ALIGN 16
#endif