ikawrakow/ik_llama.cpp
1
0
Fork 0
mirror of https://github.com/ikawrakow/ik_llama.cpp.git synced 2026-03-13 23:40:09 +00:00
Code Issues Packages Projects Releases Wiki Activity

Merge mainline llama.cpp (#3)

Browse Source 
* Merging mainline - WIP

* Merging mainline - WIP

AVX2 and CUDA appear to work.
CUDA performance seems slightly (~1-2%) lower as it is so often
the case with llama.cpp/ggml after some "improvements" have been made.

* Merging mainline - fix Metal

* Remove check

---------

Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
This commit is contained in:
Kawrakow
2024-07-27 07:55:01 +02:00
committed by GitHub
parent 0684c3e9c7
commit 154e0d75fc
612 changed files with 50817 additions and 165936 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
tests/test-quantize-fns.cpp
View File

@@ -69,7 +69,7 @@ static float reference_quantization_error(ggml_type_traits_t & qfns, size_t test
qfns.from_float(test_data, tmp_q.data(), test_size);
qfns.to_float(tmp_q.data(), tmp_out.data(), test_size);
qfns.from_float_reference(test_data, tmp_q.data(), test_size);
qfns.from_float_ref(test_data, tmp_q.data(), test_size);
qfns.to_float(tmp_q.data(), tmp_out_ref.data(), test_size);
return array_rmse(tmp_out.data(), tmp_out_ref.data(), test_size);
@@ -92,7 +92,7 @@ static float dot_product_error(
auto vdot = ggml_internal_get_type_traits(qfns.vec_dot_type);
qfns.from_float_reference(test_data1, tmp_q1.data(), test_size);
qfns.from_float_ref(test_data1, tmp_q1.data(), test_size);
vdot.from_float(test_data2, tmp_q2.data(), test_size);
float result = INFINITY;