nicolas
/
hybrid-llama
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Silence IQK merge build warnings 

- Guard GGML_RESTRICT redefine in iqk_quantize.h

- Declare iq1_bn quantize helpers in iqk_quantize.h

- Handle IQK repacked types in clamp abort switch (ops.cpp)

Assisted-by: opencode (Muse Spark)
This commit is contained in:
Marvin 2026-09-05 18:19:49 -03:00
parent eaaa668420
commit 067f41bdf0
2 changed files with 67 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

57
turboquant/ggml/src/ggml-cpu/ops.cpp
View File

@ -5838,6 +5838,63 @@ void ggml_compute_forward_clamp(
case GGML_TYPE_I32: case GGML_TYPE_I32:
case GGML_TYPE_I64: case GGML_TYPE_I64:
case GGML_TYPE_F64: case GGML_TYPE_F64:
case GGML_TYPE_Q6_0:
case GGML_TYPE_Q8_0_X4:
case GGML_TYPE_Q8_1_X4:
case GGML_TYPE_Q8_2_X4:
case GGML_TYPE_Q4_0_R8:
case GGML_TYPE_Q5_0_R4:
case GGML_TYPE_Q8_0_R8:
case GGML_TYPE_Q2_K_R4:
case GGML_TYPE_Q3_K_R4:
case GGML_TYPE_Q4_K_R4:
case GGML_TYPE_Q5_K_R4:
case GGML_TYPE_Q6_K_R4:
case GGML_TYPE_Q6_0_R4:
case GGML_TYPE_IQ2_XXS_R4:
case GGML_TYPE_IQ2_XS_R4:
case GGML_TYPE_IQ3_XXS_R4:
case GGML_TYPE_IQ1_S_R4:
case GGML_TYPE_IQ4_NL_R4:
case GGML_TYPE_IQ3_S_R4:
case GGML_TYPE_IQ2_S_R4:
case GGML_TYPE_IQ4_XS_R8:
case GGML_TYPE_IQ1_M_R4:
case GGML_TYPE_BF16_R16:
case GGML_TYPE_IQ2_BN_R4:
case GGML_TYPE_IQ2_K_R4:
case GGML_TYPE_IQ3_K_R4:
case GGML_TYPE_IQ4_K_R4:
case GGML_TYPE_IQ5_K_R4:
case GGML_TYPE_IQ4_KS_R4:
case GGML_TYPE_IQ5_KS_R4:
case GGML_TYPE_MXFP4_R8:
case GGML_TYPE_Q8_K_R16:
case GGML_TYPE_Q8_KV_R8:
case GGML_TYPE_Q8_K_R8:
case GGML_TYPE_Q8_K64:
case GGML_TYPE_Q8_K128:
case GGML_TYPE_Q8_K32:
case GGML_TYPE_Q8_K16:
case GGML_TYPE_Q8_KV:
case GGML_TYPE_IQ1_BN:
case GGML_TYPE_IQ2_BN:
case GGML_TYPE_IQ2_K:
case GGML_TYPE_IQ3_K:
case GGML_TYPE_IQ4_K:
case GGML_TYPE_IQ5_K:
case GGML_TYPE_IQ6_K:
case GGML_TYPE_IQ4_KS:
case GGML_TYPE_IQ2_KS:
case GGML_TYPE_IQ4_KSS:
case GGML_TYPE_IQ5_KS:
case GGML_TYPE_IQ2_KT:
case GGML_TYPE_IQ3_KT:
case GGML_TYPE_IQ4_KT:
case GGML_TYPE_IQ3_KS:
case GGML_TYPE_IQ2_KL:
case GGML_TYPE_IQ1_KT:
case GGML_TYPE_Q8_KR8:
case GGML_TYPE_COUNT: case GGML_TYPE_COUNT:
{ {
GGML_ABORT("fatal error"); GGML_ABORT("fatal error");

11
turboquant/ggml/src/iqk/iqk_quantize.h
View File

@ -13,11 +13,15 @@
#include "ggml-common.h" #include "ggml-common.h"
#ifdef __cplusplus #ifdef __cplusplus
#ifndef GGML_RESTRICT
#define GGML_RESTRICT #define GGML_RESTRICT
#endif
extern "C" { extern "C" {
#else #else
#ifndef GGML_RESTRICT
#define GGML_RESTRICT restrict #define GGML_RESTRICT restrict
#endif #endif
#endif
struct quantize_user_data; struct quantize_user_data;
@ -154,7 +158,12 @@ void dequantize_row_iq4_xs_r8(const block_iq4_xs_r8 * GGML_RESTRICT x, float *
void vec_dot_iq4_xs_r8_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 vec_dot_iq4_xs_r8_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 quantize_row_iq2_bn_ref (const float * GGML_RESTRICT x, block_iq2_bn * GGML_RESTRICT y, int64_t k); void quantize_row_iq2_bn_ref (const float * GGML_RESTRICT x, block_iq2_bn * GGML_RESTRICT y, int64_t k);
void quantize_row_iq2_bn (const float * GGML_RESTRICT x, void * 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_bn(const float * GGML_RESTRICT x, void * 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);
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, const struct quantize_user_data * user_data);
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);void quantize_row_iq2_bn (const float * GGML_RESTRICT x, void * GGML_RESTRICT y, int64_t k);
void dequantize_row_iq2_bn (const block_iq2_bn * GGML_RESTRICT x, float * GGML_RESTRICT y, int64_t k); void dequantize_row_iq2_bn (const block_iq2_bn * GGML_RESTRICT x, float * GGML_RESTRICT y, int64_t k);
size_t quantize_iq2_bn (const float * GGML_RESTRICT src, void * GGML_RESTRICT dst, int64_t nrows, int64_t n_per_row, const float * imatrix, const struct quantize_user_data * use_data); size_t quantize_iq2_bn (const float * GGML_RESTRICT src, void * GGML_RESTRICT dst, int64_t nrows, int64_t n_per_row, const float * imatrix, const struct quantize_user_data * use_data);
void vec_dot_iq2_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); void vec_dot_iq2_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);