Skip to content

more perfo with llamafile tinyblas on x86_64. #10714

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 5 commits into from
Dec 24, 2024
+287 −278
Merged

more perfo with llamafile tinyblas on x86_64. #10714

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
3f2bc65
more perfo with llamafile tinyblas on x86_64.
Djip007 Dec 7, 2024
d732874
tinyblas dynamic dispaching
Djip007 Dec 14, 2024
ac2b53c
sgemm: add M blocs.
Djip007 Dec 19, 2024
22924d8
- git 2.47 use short id of len 9.
Djip007 Dec 22, 2024
2c6864a
remove not stable test
Djip007 Dec 24, 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
7 changes: 4 additions & 3 deletions examples/server/tests/unit/test_completion.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -95,7 +95,7 @@ def test_consistent_result_same_seed(n_slots: int):
res = server.make_request("POST", "/completion", data={
"prompt": "I believe the meaning of life is",
"seed": 42,
"temperature": 1.0,
"temperature": 0.0,
"cache_prompt": False, # TODO: remove this once test_cache_vs_nocache_prompt is fixed
})
if last_res is not None:
Expand All @@ -120,9 +120,10 @@ def test_different_result_different_seed(n_slots: int):
assert res.body["content"] != last_res.body["content"]
last_res = res


# TODO figure why it don't work with temperature = 1
# @pytest.mark.parametrize("temperature", [0.0, 1.0])
@pytest.mark.parametrize("n_batch", [16, 32])
@pytest.mark.parametrize("temperature", [0.0, 1.0])
@pytest.mark.parametrize("temperature", [0.0])
def test_consistent_result_different_batch_size(n_batch: int, temperature: float):
global server
server.n_batch = n_batch
Expand Down
8 changes: 4 additions & 4 deletions ggml/src/ggml-cpu/ggml-cpu.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7419,14 +7419,14 @@ static void ggml_compute_forward_mul_mat(
if (src1_cont) {
for (int64_t i13 = 0; i13 < ne13; i13++)
for (int64_t i12 = 0; i12 < ne12; i12++)
if (!llamafile_sgemm(ne01, ne11, ne00/ggml_blck_size(src0->type),
if (!llamafile_sgemm(params,
ne01, ne11, ne00/ggml_blck_size(src0->type),
(const char *)src0->data + i12/r2*nb02 + i13/r3*nb03,
nb01/ggml_type_size(src0->type),
(const char *)src1->data + i12*nb12 + i13*nb13,
nb11/ggml_type_size(src1->type),
(char *)dst->data + i12*nb2 + i13*nb3,
nb1/ggml_type_size(dst->type),
ith, nth,
src0->type,
src1->type,
dst->type))
Expand Down Expand Up @@ -7471,14 +7471,14 @@ UseGgmlGemm1:;

for (int64_t i13 = 0; i13 < ne13; i13++)
for (int64_t i12 = 0; i12 < ne12; i12++)
if (!llamafile_sgemm(ne01, ne11, ne00/ggml_blck_size(src0->type),
if (!llamafile_sgemm(params,
ne01, ne11, ne00/ggml_blck_size(src0->type),
(const char *)src0->data + i12/r2*nb02 + i13/r3*nb03,
nb01/ggml_type_size(src0->type),
(const char *)wdata + (i12*ne11 + i13*ne12*ne11)*row_size,
row_size/ggml_type_size(vec_dot_type),
(char *)dst->data + i12*nb2 + i13*nb3,
nb1/ggml_type_size(dst->type),
ith, nth,
src0->type,
vec_dot_type,
dst->type))
Expand Down
Loading
Loading