vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

metal : fix binary ops for ne10 % 4 != 0

Browse source
This commit is contained in:
Georgi Gerganov 2023-12-12 15:55:42 +02:00
parent 08eb99179a
commit a51bc0c1c0
No known key found for this signature in database
GPG key ID: 449E073F9DC10735
2 changed files with 2 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

2
ggml-metal.m
View file

@ -1080,7 +1080,7 @@ void ggml_metal_graph_compute(
int64_t nb = ne00;
if (ggml_nelements(src1) == ne10 && ggml_is_contiguous(src1) && ne00 % 4 == 0) {
if (ggml_nelements(src1) == ne10 && ggml_is_contiguous(src1) && ne00 % 4 == 0 && ne10 % 4 == 0) {
GGML_ASSERT(ggml_is_contiguous(src0));
// src1 is a row

1
tests/test-backend-ops.cpp
View file

@ -1272,6 +1272,7 @@ static bool test_backend(ggml_backend_t backend, test_mode mode, const char * op
};
add_test_bin_bcast(GGML_TYPE_F32, {1, 1, 8, 1}, {1, 1, 1, 1});
add_test_bin_bcast(GGML_TYPE_F32, {1, 1, 1, 1}, {32, 1, 1, 1});
add_test_bin_bcast(GGML_TYPE_F32, {1, 1, 320, 320}, {1, 1, 1, 1});
add_test_bin_bcast(GGML_TYPE_F32, {16, 10, 1, 1}, {1, 1, 1, 1});
add_test_bin_bcast(GGML_TYPE_F32, {16, 10, 10, 1}, {1, 1, 1, 1});