8000 Refactor/online repacking by Djip007 · Pull Request #10446 · ggml-org/llama.cpp · GitHub
[go: up one dir, main page]
Skip to content

Refactor/online repacking #10446

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 9 commits into from
Dec 7, 2024
Merged

Refactor/online repacking #10446

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
9 commits
Select commit Hold shift + click to select a range
9ac05c1
rename ggml-cpu-aarch64.c to .cpp
Djip007 Dec 1, 2024
98ea414
reformat extra cpu backend.
Djip007 Nov 19, 2024
95322e9
clang-format
Djip007 Dec 2, 2024
3a042b4
Clean Q4_0_N_M ref
Djip007 Dec 4, 2024
0a2be72
add op GGML_OP_MUL_MAT_ID for Q4_0_N_M with runtime repack
Djip007 Dec 5, 2024
b14b471
added/corrected control on tensor size for Q4 repacking.
Djip007 Dec 6, 2024
7dc8a3e
Update ggml/src/ggml-cpu/ggml-cpu-aarch64.cpp
Djip007 Dec 7, 2024
e115f6f
Update ggml/src/ggml-cpu/ggml-cpu-aarch64.cpp
Djip007 Dec 7, 2024
1221d13
add debug logs on repacks.
Djip007 Dec 7, 2024
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
Prev Previous commit
add debug logs on repacks.
  • Loading branch information
@Djip007
Djip007 committed Dec 7, 2024
commit 1221d13df81ec7e31c0e02b3a8b87255fce4fc50
1 change: 1 addition & 0 deletions ggml/src/ggml-cpu/amx/amx.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -66,6 +66,7 @@ static void ggml_backend_amx_buffer_memset_tensor(ggml_backend_buffer_t buffer,
static void ggml_backend_amx_buffer_set_tensor(ggml_backend_buffer_t buffer, struct ggml_tensor * tensor,
const void * data, size_t offset, size_t size) {
if (qtype_has_amx_kernels(tensor->type)) {
GGML_LOG_DEBUG("%s: amx repack tensor %s of type %s\n", __func__, tensor->name, ggml_type_name(tensor->type));
ggml_backend_amx_convert_weight(tensor, data, offset, size);
} else {
memcpy((char *) tensor->data + offset, data, size);
Expand Down
2 changes: 2 additions & 0 deletions ggml/src/ggml-cpu/ggml-cpu-aarch64.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4104,6 +4104,8 @@ template <typename BLOC_TYPE, int64_t INTER_SIZE, int64_t NB_COLS> class tensor_
}

int repack(struct ggml_tensor * t, const void * data, size_t data_size) override {
GGML_LOG_DEBUG("%s: repack tensor %s with %s_%dx%d\n", __func__, t->name, ggml_type_name(t->type),
(int) NB_COLS, (int) INTER_SIZE);
return ggml::cpu::aarch64::repack<BLOC_TYPE, INTER_SIZE, NB_COLS>(t, data, data_size);
}
};
Expand Down
Loading
< 2917 /div>
0