ikawrakow/ik_llama.cpp
1
0
Fork 0
mirror of https://github.com/ikawrakow/ik_llama.cpp.git synced 2026-02-24 15:14:10 +00:00
Code Issues Packages Projects Releases Wiki Activity

Fix iq4_k_r4 on NEON

Browse Source
This commit is contained in:
Iwan Kawrakow
2025-05-19 19:46:44 +03:00
parent 06efa17fa9
commit 7090f171e1
2 changed files with 5 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
ggml/src/iqk/iqk_gemm_iqk_quants.cpp
View File

@@ -3225,6 +3225,8 @@ bool iqk_set_kernels_iqk_quants(int ne00, int typeA, int typeB, std::array<mul_m
return false;
}
func16 = nullptr;
switch (typeA) {
case GGML_TYPE_IQ2_KS:
IQK_SET_MUL_MAT_FUNCTIONS_T(mul_mat_qX_K_q8_K_T, DequantizerIQ2KS, kernels);
@@ -3267,8 +3269,10 @@ bool iqk_set_kernels_iqk_quants(int ne00, int typeA, int typeB, std::array<mul_m
break;
case GGML_TYPE_IQ5_KS_R4:
IQK_SET_MUL_MAT_FUNCTIONS(mul_mat_iq5_ks_r4_q8_k, kernels);
break;
case GGML_TYPE_IQ5_K_R4:
IQK_SET_MUL_MAT_FUNCTIONS(mul_mat_iq5_k_r4_q8_k, kernels);
break;
default:
return false;
}

1
ggml/src/iqk/iqk_gemm_kquants.cpp
View File

@@ -2836,6 +2836,7 @@ bool iqk_set_kernels_kquants(int ne00, int typeA, int typeB, std::array<mul_mat_
IQK_SET_MUL_MAT_FUNCTIONS(mul_mat_q8_KV_q8_KV, kernels)
kernels[0] = mul_mat_q8_KV_q8_KV_1;
func16 = mul_mat_q8_KV_q8_KV<16>;
break;
case GGML_TYPE_Q8_KV_R8:
IQK_SET_MUL_MAT_FUNCTIONS(mul_mat_q8_KV_r8_q8_KV, kernels);
break;