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

metal : trying bs = 512 performance (wip)

Browse source
This commit is contained in:
Georgi Gerganov 2024-02-12 19:21:57 +02:00
parent e8b00e2941
commit 5a668ea000
No known key found for this signature in database
GPG key ID: 449E073F9DC10735
3 changed files with 44 additions and 15 deletions
Download patch file Download diff file Expand all files Collapse all files

15
tests/test-backend-ops.cpp
View file

@ -2075,7 +2075,7 @@ static bool test_backend(ggml_backend_t backend, test_mode mode, const char * op
test_cases.emplace_back(new test_mul_mat(type_a, type_b, 16, 16, 256, {10, 10}, {2, 2}));
}
}
#else
#elif 0
for (int r0 = 0; r0 < 32; ++r0) {
for (int c0 = 0; c0 < 4096; c0 += 512) {
for (ggml_type type_a : {GGML_TYPE_Q4_0, GGML_TYPE_Q8_0, GGML_TYPE_F16}) {
@ -2092,6 +2092,19 @@ static bool test_backend(ggml_backend_t backend, test_mode mode, const char * op
}
}
}
#elif 1
for (ggml_type type_a : {GGML_TYPE_Q4_0, GGML_TYPE_Q8_0, GGML_TYPE_F16}) {
for (ggml_type type_b : {GGML_TYPE_F32}) {
test_cases.emplace_back(new test_mul_mat(type_a, type_b, 4096, 512, 4096, { 1, 1}, {1, 1}));
test_cases.emplace_back(new test_mul_mat(type_a, type_b, 4096, 512, 4096, { 1, 1}, {1, 1}));
test_cases.emplace_back(new test_mul_mat(type_a, type_b, 4096, 512, 4096, { 1, 1}, {1, 1}));
test_cases.emplace_back(new test_mul_mat(type_a, type_b, 4096, 512, 4096, { 1, 1}, {1, 1}));
test_cases.emplace_back(new test_mul_mat(type_a, type_b, 4096, 512, 4096, { 1, 1}, {1, 1}));
test_cases.emplace_back(new test_mul_mat(type_a, type_b, 4096, 512, 4096, { 1, 1}, {1, 1}));
test_cases.emplace_back(new test_mul_mat(type_a, type_b, 4096, 512, 4096, { 1, 1}, {1, 1}));
test_cases.emplace_back(new test_mul_mat(type_a, type_b, 4096, 512, 4096, { 1, 1}, {1, 1}));
}
}
#endif
for (ggml_type type_a : all_types) {