diff --git a/sgemm.cpp b/sgemm.cpp index bbe263ddd..54390b5e2 100644 --- a/sgemm.cpp +++ b/sgemm.cpp @@ -235,6 +235,11 @@ template <> inline __m512 load(const ggml_fp16_t *p) { } #endif // __AVX512F__ +//////////////////////////////////////////////////////////////////////////////////////////////////// +// CONSTANTS +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); + //////////////////////////////////////////////////////////////////////////////////////////////////// // FLOATING POINT MATRIX MULTIPLICATION @@ -787,6 +792,20 @@ class tinyBLAS_Q0_AVX { return _mm_sub_epi8(_mm_and_si128(_mm_set1_epi8(15), _mm_srli_epi16(x, 4)), _mm_set1_epi8(8)); } + inline __m256i load(const block_iq4_nl *b) { + return = MM256_SET_M128I(load1(b), load0(b)); + } + + inline __m128i load0(const block_iq4_nl *b) { + const __m128i x = _mm_loadu_si128((const __m128i *)(b->qs)); + return _mm_shuffle_epi8(iq4nlt, _mm_and_si128(_mm_set1_epi8(15), x)); + } + + inline __m128i load1(const block_iq4_nl *b) { + const __m128i x = _mm_loadu_si128((const __m128i *)(b->qs)); + return _mm_shuffle_epi8(iq4nlt, _mm_and_si128(_mm_set1_epi8(15), _mm_srli_epi16(x, 4))); + } + inline __m256 updot(__m256i u, __m256i s) { __m256i res; #if defined(__AVXVNNI__) || (defined(__AVX512VNNI__) && defined(__AVX512VL__)) @@ -1010,6 +1029,22 @@ bool llamafile_sgemm(int64_t m, int64_t n, int64_t k, const void *A, int64_t lda #endif } + case GGML_TYPE_IQ4_NL: { + if (Btype != GGML_TYPE_Q8_0) + return false; +#if defined(__AVX2__) || defined(__AVX512F__) || defined(__AVX__) + tinyBLAS_Q0_AVX tb{ + k, (const block_iq4_nl *)A, lda, + (const block_q8_0 *)B, ldb, + (float *)C, ldc, + ith, nth}; + tb.matmul(m, n, task); + return true; +#else + return false; +#endif + } + default: return false; }