ikawrakow/ik_llama.cpp
1
0
Fork 0
mirror of https://github.com/ikawrakow/ik_llama.cpp.git synced 2026-02-24 07:04:11 +00:00
Code Issues Packages Projects Releases Wiki Activity

Repack q4_0 and q8_0 to q8_0_R8

Browse Source 
q8_0 is fine, but I observe a very significant PPL increase
for q4_0. Best guess: precision loss with the 32 bit <-> 16 bit
scale conversions.
This commit is contained in:
Iwan Kawrakow
2025-06-18 08:46:47 +03:00
parent dc96820ddb
commit a16f961033
3 changed files with 83 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

76
ggml/src/iqk/iqk_gemm_legacy_quants.cpp
View File

@@ -1690,6 +1690,73 @@ static void mul_mat_q8_1_r8_q8_2(int n, const void * vx, size_t bx, const DataIn
}
}
void iqk_convert_q80_q80_r8(int n, const void * vx, size_t bx, void * vy, int nrc_x) {
static_assert(QK4_0 == QK8_0);
GGML_ASSERT(n%QK4_0 == 0);
GGML_ASSERT(nrc_x%8 == 0);
const int nb = n/QK4_0;
block_q8_0_r8 * y = (block_q8_0_r8 *)vy;
const block_q8_0 * x8[8];
uint32_t block[8];
for (int ix = 0; ix < nrc_x; ix += 8) {
for (int k = 0; k < 8; ++k) x8[k] = (const block_q8_0 *)((const char *)vx + (ix + k)*bx);
for (int i = 0; i < nb; ++i) {
for (int k = 0; k < 8; ++k) {
y[i].d[k] = x8[k][i].d;
_mm256_storeu_si256((__m256i *)block, _mm256_loadu_si256((const __m256i *)x8[k][i].qs));
auto qs = (uint32_t *)y[i].qs;
for (int l = 0; l < 4; ++l) {
qs[8*l + k + 0] = block[l + 0];
qs[8*l + k + 32] = block[l + 4];
}
}
}
y += nb;
}
}
void iqk_convert_q40_q80_r8(int n, const void * vx, size_t bx, void * vy, int nrc_x) {
static_assert(QK4_0 == QK8_0);
GGML_ASSERT(n%QK4_0 == 0);
GGML_ASSERT(nrc_x%8 == 0);
const int nb = n/QK4_0;
block_q8_0_r8 * y = (block_q8_0_r8 *)vy;
const block_q4_0 * x8[8];
uint32_t block[8];
for (int ix = 0; ix < nrc_x; ix += 8) {
for (int k = 0; k < 8; ++k) x8[k] = (const block_q4_0 *)((const char *)vx + (ix + k)*bx);
for (int i = 0; i < nb; ++i) {
for (int k = 0; k < 8; ++k) {
y[i].d[k] = x8[k][i].d;
auto bits = _mm_loadu_si128((const __m128i *)x8[k][i].qs);
auto val = _mm256_and_si256(MM256_SET_M128I(_mm_srli_epi16(bits, 4), bits), _mm256_set1_epi8(0xf));
val = _mm256_add_epi8(val, _mm256_set1_epi8(-8));
_mm256_storeu_si256((__m256i *)block, val);
auto qs = (uint32_t *)y[i].qs;
for (int l = 0; l < 4; ++l) {
qs[8*l + k + 0] = block[l + 0];
qs[8*l + k + 32] = block[l + 4];
}
}
}
y += nb;
}
}
template <typename Dequantizer> void set_functions(std::array<mul_mat_t, IQK_MAX_NY>& funcs) {
if constexpr (std::is_same_v<Dequantizer, Q4_0_Unpacker> || std::is_same_v<Dequantizer, Q5_0_Unpacker> ||
std::is_same_v<Dequantizer, Q8_0_Unpacker>) {
@@ -1713,6 +1780,15 @@ template <typename Dequantizer> void set_functions(std::array<mul_mat_t, IQK_MAX
} // namespace
bool iqk_convert_legacy_quants_q8_r8(int type, int n, const void * vx, size_t bx, void * vy, int nrc_x) {
switch (type) {
case GGML_TYPE_Q4_0: iqk_convert_q40_q80_r8(n, vx, bx, vy, nrc_x); break;
case GGML_TYPE_Q8_0: iqk_convert_q80_q80_r8(n, vx, bx, vy, nrc_x); break;
default: return false;
}
return true;
}
bool iqk_set_kernels_legacy_quants(int ne00, int typeA, int typeB, std::array<mul_mat_t, IQK_MAX_NY>& kernels, mul_mat_t& func16) {
if (ne00%QK8_0 != 0) return false;

2
ggml/src/iqk/iqk_gemm_legacy_quants.h
View File

@@ -11,4 +11,6 @@ bool iqk_set_kernels_legacy_quants(int ne00, int typeA, int typeB, std::array<mu
void iqk_gemm_legacy_fa(int D, int nq, int type_k, const char * k, size_t stride_k, DataInfo& info, int k_step);
bool iqk_convert_legacy_quants_q8_r8(int type, int n, const void * vx, size_t bx, void * vy, int nrc_x);
#endif

8
ggml/src/iqk/iqk_mul_mat.cpp
View File

@@ -258,6 +258,8 @@ struct MulMat {
case GGML_TYPE_IQ5_KS : return nrc_y >= 32 ? GGML_TYPE_Q8_K_R8 : type;
case GGML_TYPE_IQ5_K : return nrc_y >= 32 ? GGML_TYPE_Q8_K_R8 : type;
case GGML_TYPE_IQ6_K : return nrc_y >= 32 ? GGML_TYPE_Q8_K_R8 : type;
case GGML_TYPE_Q4_0 : return nrc_y >= 32 ? GGML_TYPE_Q8_0_R8 : type;
case GGML_TYPE_Q8_0 : return nrc_y >= 32 ? GGML_TYPE_Q8_0_R8 : type;
default: break;
}
#else
@@ -403,19 +405,19 @@ bool iqk_convert_repack(int typeA, int n, const void * vx, size_t bx, void * vy,
case GGML_TYPE_IQ3_KT:
case GGML_TYPE_IQ4_KT:
return iqk_dequantize_ktquants(typeA, n, vx, bx, vy, stride_y, nrc_x);
//case GGML_TYPE_Q4_0:
case GGML_TYPE_Q4_0:
//case GGML_TYPE_Q4_1:
//case GGML_TYPE_Q5_0:
//case GGML_TYPE_Q5_1:
//case GGML_TYPE_Q6_0:
//case GGML_TYPE_Q8_0:
case GGML_TYPE_Q8_0:
//case GGML_TYPE_IQ4_NL:
//case GGML_TYPE_Q4_0_R8:
//case GGML_TYPE_Q5_0_R4:
//case GGML_TYPE_Q6_0_R4:
//case GGML_TYPE_Q8_0_R8:
//case GGML_TYPE_IQ4_NL_R4:
// return iqk_set_kernels_legacy_quants(ne00, typeA, typeB, mm.funcs, mm.func16);
return iqk_convert_legacy_quants_q8_r8(typeA, n, vx, bx, vy, nrc_x);
case GGML_TYPE_IQ1_S:
case GGML_TYPE_IQ1_M:
//case GGML_TYPE_IQ1_S_R4: