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

Apply suggestions from code review

Browse source
This commit is contained in:
Diego Devesa 2024-12-31 14:44:19 +01:00 • committed by GitHub
parent 9ad89bc9d3
commit 1c8ba922ec
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
3 changed files with 3 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

2
ggml/src/ggml-cpu/ggml-cpu-aarch64.cpp
View file

@ -194,7 +194,7 @@ static inline __m256i sum_i16_pairs_int32x8(const __m256i x) {
} }
static inline __m256i mul_sum_us8_pairs_int32x8(const __m256i ax, const __m256i sy) { static inline __m256i mul_sum_us8_pairs_int32x8(const __m256i ax, const __m256i sy) {
#if defined(__AVX512VNNI__) && defined(__AVX512VL__) #if defined(__AVX512VNNI__) && defined(__AVX512VL__)
const __m256i zero = _mm256_setzero_si256(); const __m256i zero = _mm256_setzero_si256();
return _mm256_dpbusd_epi32(zero, ax, sy); return _mm256_dpbusd_epi32(zero, ax, sy);
#elif defined(__AVXVNNI__) #elif defined(__AVXVNNI__)

2
ggml/src/ggml-cpu/ggml-cpu-quants.c
View file

@ -103,7 +103,7 @@ static inline __m256 sum_i16_pairs_float(const __m256i x) {
} }
static inline __m256 mul_sum_us8_pairs_float(const __m256i ax, const __m256i sy) { static inline __m256 mul_sum_us8_pairs_float(const __m256i ax, const __m256i sy) {
#if defined(__AVX512VNNI__) && defined(__AVX512VL__) #if defined(__AVX512VNNI__) && defined(__AVX512VL__)
const __m256i zero = _mm256_setzero_si256(); const __m256i zero = _mm256_setzero_si256();
const __m256i summed_pairs = _mm256_dpbusd_epi32(zero, ax, sy); const __m256i summed_pairs = _mm256_dpbusd_epi32(zero, ax, sy);
return _mm256_cvtepi32_ps(summed_pairs); return _mm256_cvtepi32_ps(summed_pairs);

2
ggml/src/ggml-cpu/llamafile/sgemm.cpp
View file

@ -1000,7 +1000,7 @@ class tinyBLAS_Q0_AVX {
inline __m256 updot(__m256i u, __m256i s) { inline __m256 updot(__m256i u, __m256i s) {
__m256i res; __m256i res;
#if defined(__AVX512VNNI__) && defined(__AVX512VL__) #if defined(__AVX512VNNI__) && defined(__AVX512VL__)
res = _mm256_dpbusd_epi32(_mm256_setzero_si256(), u, s); res = _mm256_dpbusd_epi32(_mm256_setzero_si256(), u, s);
#elif defined(__AVXVNNI__) #elif defined(__AVXVNNI__)
res = _mm256_dpbusd_avx_epi32(_mm256_setzero_si256(), u, s); res = _mm256_dpbusd_avx_epi32(_mm256_setzero_si256(), u, s);