mirror of
https://github.com/ikawrakow/ik_llama.cpp.git
synced 2026-03-14 07:48:16 +00:00
7f61b3068e
* iq1_m_r4: basics (quantize/dequantize) * iq1_m_r4: Zen4 gemm * iq1_m_r4: neon gemm * iq1_m_r4: switch to q8_0_x4 also on AVX2/Zen4 With the deltas being per group of 8, we cannot make use of the q8 sums stored in q8_1, so we get a tiny gain by using q8_0_x4. * iq1_m_r4: rename mul_mat_iq1_m_r4_q8_1 to mul_mat_iq1_m_r4_q8_0 --------- Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
167 lines
14 KiB
C
167 lines
14 KiB
C
//
|
|
// Copyright (C) 2023-2024 The ggml authors
|
|
// Copyright (C) 2024 Iwan Kawrakow
|
|
// MIT license
|
|
// SPDX-License-Identifier: MIT
|
|
//
|
|
|
|
#pragma once
|
|
|
|
#define GGML_COMMON_DECL_C
|
|
#include "ggml-common.h"
|
|
|
|
#include "ggml.h"
|
|
|
|
// GGML internal header
|
|
|
|
#ifdef __cplusplus
|
|
extern "C" {
|
|
#endif
|
|
|
|
// Quantization
|
|
void quantize_row_q4_0_ref(const float * GGML_RESTRICT x, block_q4_0 * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_q4_1_ref(const float * GGML_RESTRICT x, block_q4_1 * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_q5_0_ref(const float * GGML_RESTRICT x, block_q5_0 * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_q5_1_ref(const float * GGML_RESTRICT x, block_q5_1 * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_q8_0_ref(const float * GGML_RESTRICT x, block_q8_0 * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_q8_1_ref(const float * GGML_RESTRICT x, block_q8_1 * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_q6_0_ref(const float * GGML_RESTRICT x, block_q6_0 * GGML_RESTRICT y, int64_t k);
|
|
|
|
void quantize_row_q2_K_ref(const float * GGML_RESTRICT x, block_q2_K * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_q3_K_ref(const float * GGML_RESTRICT x, block_q3_K * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_q4_K_ref(const float * GGML_RESTRICT x, block_q4_K * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_q5_K_ref(const float * GGML_RESTRICT x, block_q5_K * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_q6_K_ref(const float * GGML_RESTRICT x, block_q6_K * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_q8_K_ref(const float * GGML_RESTRICT x, block_q8_K * GGML_RESTRICT y, int64_t k);
|
|
|
|
void quantize_row_iq2_xxs_ref(const float * GGML_RESTRICT x, block_iq2_xxs * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_iq2_xs_ref (const float * GGML_RESTRICT x, block_iq2_xs * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_iq3_xxs_ref(const float * GGML_RESTRICT x, block_iq3_xxs * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_iq4_nl_ref (const float * GGML_RESTRICT x, block_iq4_nl * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_iq4_xs_ref (const float * GGML_RESTRICT x, block_iq4_xs * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_iq3_s_ref (const float * GGML_RESTRICT x, block_iq3_s * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_iq2_s_ref (const float * GGML_RESTRICT x, block_iq2_s * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_iq1_bn_ref (const float * GGML_RESTRICT x, block_iq1_bn * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_iq1_s_ref (const float * GGML_RESTRICT x, block_iq1_s * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_iq1_m_ref (const float * GGML_RESTRICT x, block_iq1_m * GGML_RESTRICT y, int64_t k);
|
|
|
|
void quantize_row_q4_0(const float * GGML_RESTRICT x, void * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_q4_1(const float * GGML_RESTRICT x, void * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_q5_0(const float * GGML_RESTRICT x, void * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_q5_1(const float * GGML_RESTRICT x, void * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_q8_0(const float * GGML_RESTRICT x, void * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_q8_1(const float * GGML_RESTRICT x, void * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_q6_0(const float * GGML_RESTRICT x, void * GGML_RESTRICT y, int64_t k);
|
|
|
|
void quantize_row_q2_K(const float * GGML_RESTRICT x, void * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_q3_K(const float * GGML_RESTRICT x, void * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_q4_K(const float * GGML_RESTRICT x, void * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_q5_K(const float * GGML_RESTRICT x, void * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_q6_K(const float * GGML_RESTRICT x, void * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_q8_K(const float * GGML_RESTRICT x, void * GGML_RESTRICT y, int64_t k);
|
|
|
|
void quantize_row_iq2_xxs(const float * GGML_RESTRICT x, void * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_iq2_xs (const float * GGML_RESTRICT x, void * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_iq3_xxs(const float * GGML_RESTRICT x, void * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_iq4_nl (const float * GGML_RESTRICT x, void * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_iq4_xs (const float * GGML_RESTRICT x, void * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_iq3_s (const float * GGML_RESTRICT x, void * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_iq2_s (const float * GGML_RESTRICT x, void * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_iq1_bn (const float * GGML_RESTRICT x, void * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_iq1_s (const float * GGML_RESTRICT x, void * GGML_RESTRICT y, int64_t k);
|
|
void quantize_row_iq1_m (const float * GGML_RESTRICT x, void * GGML_RESTRICT y, int64_t k);
|
|
|
|
// Dequantization
|
|
void dequantize_row_q4_0(const block_q4_0 * GGML_RESTRICT x, float * GGML_RESTRICT y, int64_t k);
|
|
void dequantize_row_q4_1(const block_q4_1 * GGML_RESTRICT x, float * GGML_RESTRICT y, int64_t k);
|
|
void dequantize_row_q5_0(const block_q5_0 * GGML_RESTRICT x, float * GGML_RESTRICT y, int64_t k);
|
|
void dequantize_row_q5_1(const block_q5_1 * GGML_RESTRICT x, float * GGML_RESTRICT y, int64_t k);
|
|
void dequantize_row_q8_0(const block_q8_0 * GGML_RESTRICT x, float * GGML_RESTRICT y, int64_t k);
|
|
//void dequantize_row_q8_1(const block_q8_1 * GGML_RESTRICT x, float * GGML_RESTRICT y, int64_t k);
|
|
void dequantize_row_q6_0(const block_q6_0 * GGML_RESTRICT x, float * GGML_RESTRICT y, int64_t k);
|
|
|
|
void dequantize_row_q2_K(const block_q2_K * GGML_RESTRICT x, float * GGML_RESTRICT y, int64_t k);
|
|
void dequantize_row_q3_K(const block_q3_K * GGML_RESTRICT x, float * GGML_RESTRICT y, int64_t k);
|
|
void dequantize_row_q4_K(const block_q4_K * GGML_RESTRICT x, float * GGML_RESTRICT y, int64_t k);
|
|
void dequantize_row_q5_K(const block_q5_K * GGML_RESTRICT x, float * GGML_RESTRICT y, int64_t k);
|
|
void dequantize_row_q6_K(const block_q6_K * GGML_RESTRICT x, float * GGML_RESTRICT y, int64_t k);
|
|
void dequantize_row_q8_K(const block_q8_K * GGML_RESTRICT x, float * GGML_RESTRICT y, int64_t k);
|
|
|
|
void dequantize_row_iq2_xxs(const block_iq2_xxs * GGML_RESTRICT x, float * GGML_RESTRICT y, int64_t k);
|
|
void dequantize_row_iq2_xs (const block_iq2_xs * GGML_RESTRICT x, float * GGML_RESTRICT y, int64_t k);
|
|
void dequantize_row_iq2_s (const block_iq2_s * GGML_RESTRICT x, float * GGML_RESTRICT y, int64_t k);
|
|
void dequantize_row_iq3_xxs(const block_iq3_xxs * GGML_RESTRICT x, float * GGML_RESTRICT y, int64_t k);
|
|
void dequantize_row_iq1_s (const block_iq1_s * GGML_RESTRICT x, float * GGML_RESTRICT y, int64_t k);
|
|
void dequantize_row_iq1_m (const block_iq1_m * GGML_RESTRICT x, float * GGML_RESTRICT y, int64_t k);
|
|
void dequantize_row_iq4_nl (const block_iq4_nl * GGML_RESTRICT x, float * GGML_RESTRICT y, int64_t k);
|
|
void dequantize_row_iq4_xs (const block_iq4_xs * GGML_RESTRICT x, float * GGML_RESTRICT y, int64_t k);
|
|
void dequantize_row_iq3_s (const block_iq3_s * GGML_RESTRICT x, float * GGML_RESTRICT y, int64_t k);
|
|
void dequantize_row_iq1_bn (const block_iq1_bn * GGML_RESTRICT x, float * GGML_RESTRICT y, int64_t k);
|
|
|
|
// Dot product
|
|
void ggml_vec_dot_q4_0_q8_0(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, size_t bx, const void * GGML_RESTRICT vy, size_t by, int nrc);
|
|
void ggml_vec_dot_q4_1_q8_1(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, size_t bx, const void * GGML_RESTRICT vy, size_t by, int nrc);
|
|
void ggml_vec_dot_q5_0_q8_0(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, size_t bx, const void * GGML_RESTRICT vy, size_t by, int nrc);
|
|
void ggml_vec_dot_q5_1_q8_1(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, size_t bx, const void * GGML_RESTRICT vy, size_t by, int nrc);
|
|
void ggml_vec_dot_q8_0_q8_0(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, size_t bx, const void * GGML_RESTRICT vy, size_t by, int nrc);
|
|
void ggml_vec_dot_q6_0_q8_0(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, size_t bx, const void * GGML_RESTRICT vy, size_t by, int nrc);
|
|
|
|
void ggml_vec_dot_q2_K_q8_K(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, size_t bx, const void * GGML_RESTRICT vy, size_t by, int nrc);
|
|
void ggml_vec_dot_q3_K_q8_K(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, size_t bx, const void * GGML_RESTRICT vy, size_t by, int nrc);
|
|
void ggml_vec_dot_q4_K_q8_K(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, size_t bx, const void * GGML_RESTRICT vy, size_t by, int nrc);
|
|
void ggml_vec_dot_q5_K_q8_K(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, size_t bx, const void * GGML_RESTRICT vy, size_t by, int nrc);
|
|
void ggml_vec_dot_q6_K_q8_K(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, size_t bx, const void * GGML_RESTRICT vy, size_t by, int nrc);
|
|
|
|
void ggml_vec_dot_iq2_xxs_q8_K(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, size_t bx, const void * GGML_RESTRICT vy, size_t by, int nrc);
|
|
void ggml_vec_dot_iq2_xs_q8_K (int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, size_t bx, const void * GGML_RESTRICT vy, size_t by, int nrc);
|
|
void ggml_vec_dot_iq2_s_q8_K (int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, size_t bx, const void * GGML_RESTRICT vy, size_t by, int nrc);
|
|
void ggml_vec_dot_iq3_xxs_q8_K(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, size_t bx, const void * GGML_RESTRICT vy, size_t by, int nrc);
|
|
void ggml_vec_dot_iq1_s_q8_K (int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, size_t bx, const void * GGML_RESTRICT vy, size_t by, int nrc);
|
|
void ggml_vec_dot_iq1_m_q8_K (int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, size_t bx, const void * GGML_RESTRICT vy, size_t by, int nrc);
|
|
void ggml_vec_dot_iq4_nl_q8_0 (int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, size_t bx, const void * GGML_RESTRICT vy, size_t by, int nrc);
|
|
void ggml_vec_dot_iq4_xs_q8_K (int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, size_t bx, const void * GGML_RESTRICT vy, size_t by, int nrc);
|
|
void ggml_vec_dot_iq3_s_q8_K (int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, size_t bx, const void * GGML_RESTRICT vy, size_t by, int nrc);
|
|
void ggml_vec_dot_iq1_bn_q8_K64(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, size_t bx, const void * GGML_RESTRICT vy, size_t by, int nrc);
|
|
|
|
// Quantization utilizing an importance matrix (a.k.a. "Activation aWare Quantization")
|
|
size_t quantize_iq2_xxs(const float * GGML_RESTRICT src, void * GGML_RESTRICT dst, int64_t nrows, int64_t n_per_row, const float * imatrix);
|
|
size_t quantize_iq2_xs (const float * GGML_RESTRICT src, void * GGML_RESTRICT dst, int64_t nrows, int64_t n_per_row, const float * imatrix);
|
|
size_t quantize_iq2_s (const float * GGML_RESTRICT src, void * GGML_RESTRICT dst, int64_t nrows, int64_t n_per_row, const float * imatrix);
|
|
size_t quantize_iq3_xxs(const float * GGML_RESTRICT src, void * GGML_RESTRICT dst, int64_t nrows, int64_t n_per_row, const float * imatrix);
|
|
size_t quantize_iq1_s (const float * GGML_RESTRICT src, void * GGML_RESTRICT dst, int64_t nrows, int64_t n_per_row, const float * imatrix);
|
|
size_t quantize_iq1_m (const float * GGML_RESTRICT src, void * GGML_RESTRICT dst, int64_t nrows, int64_t n_per_row, const float * imatrix);
|
|
size_t quantize_iq4_nl (const float * GGML_RESTRICT src, void * GGML_RESTRICT dst, int64_t nrows, int64_t n_per_row, const float * imatrix);
|
|
size_t quantize_iq4_xs (const float * GGML_RESTRICT src, void * GGML_RESTRICT dst, int64_t nrows, int64_t n_per_row, const float * imatrix);
|
|
size_t quantize_iq3_s (const float * GGML_RESTRICT src, void * GGML_RESTRICT dst, int64_t nrows, int64_t n_per_row, const float * imatrix);
|
|
size_t quantize_iq1_bn (const float * GGML_RESTRICT src, void * GGML_RESTRICT dst, int64_t nrows, int64_t n_per_row, const float * imatrix);
|
|
|
|
size_t quantize_q2_K(const float * GGML_RESTRICT src, void * GGML_RESTRICT dst, int64_t nrows, int64_t n_per_row, const float * imatrix);
|
|
size_t quantize_q3_K(const float * GGML_RESTRICT src, void * GGML_RESTRICT dst, int64_t nrows, int64_t n_per_row, const float * imatrix);
|
|
size_t quantize_q4_K(const float * GGML_RESTRICT src, void * GGML_RESTRICT dst, int64_t nrows, int64_t n_per_row, const float * imatrix);
|
|
size_t quantize_q5_K(const float * GGML_RESTRICT src, void * GGML_RESTRICT dst, int64_t nrows, int64_t n_per_row, const float * imatrix);
|
|
size_t quantize_q6_K(const float * GGML_RESTRICT src, void * GGML_RESTRICT dst, int64_t nrows, int64_t n_per_row, const float * imatrix);
|
|
size_t quantize_q4_0(const float * GGML_RESTRICT src, void * GGML_RESTRICT dst, int64_t nrows, int64_t n_per_row, const float * imatrix);
|
|
size_t quantize_q4_1(const float * GGML_RESTRICT src, void * GGML_RESTRICT dst, int64_t nrows, int64_t n_per_row, const float * imatrix);
|
|
size_t quantize_q5_0(const float * GGML_RESTRICT src, void * GGML_RESTRICT dst, int64_t nrows, int64_t n_per_row, const float * imatrix);
|
|
size_t quantize_q5_1(const float * GGML_RESTRICT src, void * GGML_RESTRICT dst, int64_t nrows, int64_t n_per_row, const float * imatrix);
|
|
size_t quantize_q8_0(const float * GGML_RESTRICT src, void * GGML_RESTRICT dst, int64_t nrows, int64_t n_per_row, const float * imatrix);
|
|
size_t quantize_q6_0(const float * GGML_RESTRICT src, void * GGML_RESTRICT dst, int64_t nrows, int64_t n_per_row, const float * imatrix);
|
|
|
|
void iq2xs_init_impl(enum ggml_type type);
|
|
void iq2xs_free_impl(enum ggml_type type);
|
|
void iq3xs_init_impl(int grid_size);
|
|
void iq3xs_free_impl(int grid_size);
|
|
|
|
void iq1s_process_1block(int block_size, const float * xb, const float * weight, int8_t * L,
|
|
float * the_scale, uint16_t * the_index, int * the_shift, float * pairs, float * sumx, float * sumw);
|
|
void iq1m_process_1block(const float * xb, const float * weight, int8_t * L,
|
|
float * the_scale, uint16_t * the_index, int * the_shift, float * pairs);
|
|
|
|
#if defined(__ARM_FEATURE_SVE)
|
|
extern int ggml_sve_cnt_b;
|
|
#endif
|
|
|
|
#ifdef __cplusplus
|
|
}
|
|
#endif
|