CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-10-30 08:42:00 +00:00
Code Issues Packages Projects Releases Wiki Activity

llamafile: use member variable instead of constant for iq4nlt (#11780)

Browse Source
This commit is contained in:
Jeffrey Morgan
2025-02-13 09:05:04 -08:00
committed by GitHub
parent c1f958c038
commit 8a8c4ceb60
1 changed files with 9 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files

17
ggml/src/ggml-cpu/llamafile/sgemm.cpp
View File

@@ -280,14 +280,6 @@ template <> inline __m256bh load(const float *p) {
}
#endif
////////////////////////////////////////////////////////////////////////////////////////////////////
// CONSTANTS
#if defined(__AVX__) || defined(__AVX2__) || defined(__AVX512F__)
static const int8_t kvalues_iq4nl[16] = {-127, -104, -83, -65, -49, -35, -22, -10, 1, 13, 25, 38, 53, 69, 89, 113};
static const __m128i iq4nlt = _mm_loadu_si128((const __m128i *) kvalues_iq4nl);
#endif
////////////////////////////////////////////////////////////////////////////////////////////////////
// FLOATING POINT MATRIX MULTIPLICATION
@@ -614,6 +606,14 @@ class tinyBLAS_Q0_AVX {
TC *C, int64_t ldc,
int ith, int nth)
: A(A), B(B), C(C), k(k), lda(lda), ldb(ldb), ldc(ldc), ith(ith), nth(nth) {
const int8_t kvalues_iq4nl[16] = {
-127, -104, -83, -65,
-49, -35, -22, -10,
1, 13, 25, 38,
53, 69, 89, 113
};
iq4nlt = _mm_loadu_si128((const __m128i *)kvalues_iq4nl);
}
void matmul(int64_t m, int64_t n) {
@@ -1038,6 +1038,7 @@ class tinyBLAS_Q0_AVX {
const int64_t ldc;
const int ith;
const int nth;
__m128i iq4nlt;
};
#endif // __AVX__