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

ggml : fix Q4_3 cuBLAS

Browse source
This commit is contained in:
Georgi Gerganov 2023-04-22 16:31:56 +03:00
parent 857308d1e8
commit 0e018fe008
No known key found for this signature in database
GPG key ID: 449E073F9DC10735
2 changed files with 5 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
CMakeLists.txt
View file

@ -312,6 +312,7 @@ add_library(ggml OBJECT
target_include_directories(ggml PUBLIC .)
target_compile_features(ggml PUBLIC c_std_11) # don't bump
target_link_libraries(ggml PUBLIC Threads::Threads ${LLAMA_EXTRA_LIBS})
if (BUILD_SHARED_LIBS)
set_target_properties(ggml PROPERTIES POSITION_INDEPENDENT_CODE ON)
endif()
@ -324,6 +325,7 @@ add_library(llama
target_include_directories(llama PUBLIC .)
target_compile_features(llama PUBLIC cxx_std_11) # don't bump
target_link_libraries(llama PRIVATE ggml ${LLAMA_EXTRA_LIBS})
if (BUILD_SHARED_LIBS)
set_target_properties(llama PROPERTIES POSITION_INDEPENDENT_CODE ON)
target_compile_definitions(llama PRIVATE LLAMA_SHARED LLAMA_BUILD)

3
ggml.c
View file

@ -7992,6 +7992,9 @@ static void ggml_compute_forward_mul_mat_q_f32(
else if (type == GGML_TYPE_Q4_2) {
dequantize_row_q_cuda = dequantize_row_q4_2_cuda;
}
else if (type == GGML_TYPE_Q4_3) {
dequantize_row_q_cuda = dequantize_row_q4_3_cuda;
}
else {
GGML_ASSERT(false);
}