From 154e0d75fccf1784fe9ff6fd76a630b66563da3d Mon Sep 17 00:00:00 2001 From: Kawrakow <48489457+ikawrakow@users.noreply.github.com> Date: Sat, 27 Jul 2024 07:55:01 +0200 Subject: Merge mainline llama.cpp (#3) * 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 --- tests/test-quantize-perf.cpp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'tests/test-quantize-perf.cpp') diff --git a/tests/test-quantize-perf.cpp b/tests/test-quantize-perf.cpp index 48d9fae3..24e06605 100644 --- a/tests/test-quantize-perf.cpp +++ b/tests/test-quantize-perf.cpp @@ -285,7 +285,7 @@ int main(int argc, char * argv[]) { for (size_t size : params.test_sizes) { printf(" %zu values (%.2f MB)\n", size, 4*size/(float)(1024*1024)); auto quantize_fn = [&](void) -> float { - qfns.from_float_reference(test_data1, test_q1, size); + qfns.from_float_ref(test_data1, test_q1, size); return test_q1[0]; }; size_t quantized_size = ggml_row_size(type, size); -- cgit v1.2.3