ikawrakow/ik_llama.cpp
1
0
Fork 0
mirror of https://github.com/ikawrakow/ik_llama.cpp.git synced 2026-02-28 09:04:10 +00:00
Code Issues Packages Projects Releases Wiki Activity

CUDA: faster iq2_k_r4 GEMV

Browse Source
This commit is contained in:
Iwan Kawrakow
2025-05-26 16:29:36 +03:00
parent fa011c9017
commit 1a8145e48e
1 changed files with 41 additions and 39 deletions
Download Patch File Download Diff File Expand all files Collapse all files

80
ggml/src/ggml-cuda/iqk_mmvq.cu
View File

@@ -527,45 +527,6 @@ __device__ __forceinline__ void vec_dot_iq3_k_r4_q8_1(
}
}
__device__ __forceinline__ void vec_dot_iq2_k_r4_q8_1(
const void * __restrict__ vbq, const block_q8_1 * __restrict__ bq8_1, const int & kbx, const int & iqs, float * result) {
const block_iq2_k_r4 * bq2 = (const block_iq2_k_r4 *)vbq + kbx;
// iqs is 0...30 in steps of 2
const int ib16 = iqs/2;
const float d8 = __low2float(bq8_1[ib16/2].ds);
const int32_t * q8 = (const int *)bq8_1[ib16/2].qs + 4*(ib16%2);
int ib32 = ib16/2;
int is = ib16%2;
const int * scales_l = (const int *)bq2->scales;
int scales = __vsub4(((scales_l[2*(ib32%4)+is] >> 4*(ib32/4)) & 0x0f0f0f0f), 0x08080808);
const int8_t * s8 = (const int8_t *)&scales;
int2 val1;
const int * q2 = (const int *)bq2->qs + 8*ib32 + 4*is;
int aux32[2];
const uint8_t * aux8 = (const uint8_t *)aux32;
for (int i = 0; i < 4; ++i) {
auto values1 = iq2nl_values + (((bq2->extra[i+4*is] >> ib32) & 1) << 2);
int sumi1 = 0;
aux32[0] = ((q2[i] >> 0) & 0x03030303);
aux32[1] = ((q2[i] >> 2) & 0x03030303);
// TODO: int_from_table_4
val1.x = int_from_table(aux8+0, (const uint8_t *)values1);
val1.y = int_from_table(aux8+4, (const uint8_t *)values1);
sumi1 = ggml_cuda_dp4a(val1.x, q8[0], ggml_cuda_dp4a(val1.y, q8[1], sumi1));
aux32[0] = ((q2[i] >> 4) & 0x03030303);
aux32[1] = ((q2[i] >> 6) & 0x03030303);
val1.x = int_from_table(aux8+0, (const uint8_t *)values1);
val1.y = int_from_table(aux8+4, (const uint8_t *)values1);
sumi1 = ggml_cuda_dp4a(val1.x, q8[2], ggml_cuda_dp4a(val1.y, q8[3], sumi1));
const float d = __half2float(bq2->d[i]) * d8;
result[i] += d * sumi1 * s8[i];
}
}
#define VDR_IQ6_K_Q8_1_MMVQ 4
#define VDR_IQ6_K_Q8_1_MMQ 4
@@ -793,6 +754,47 @@ __device__ __forceinline__ void vec_dot_iq2_ks_q8_1(
+ __low2float(bq8_1[4*(i4/4)+3].ds) * sumi4);
}
__device__ __forceinline__ void vec_dot_iq2_k_r4_q8_1(
const void * __restrict__ vbq, const block_q8_1 * __restrict__ bq8_1, const int & kbx, const int & iqs, float * result) {
const block_iq2_k_r4 * bq2 = (const block_iq2_k_r4 *)vbq + kbx;
// iqs is 0...30 in steps of 2
const int ib16 = iqs/2;
const float d8 = __low2float(bq8_1[ib16/2].ds);
const int32_t * q8 = (const int *)bq8_1[ib16/2].qs + 4*(ib16%2);
int ib32 = ib16/2;
int is = ib16%2;
const int * scales_l = (const int *)bq2->scales;
const int * all_values = (const int *)iq2k_table;
int scales = __vsub4(((scales_l[2*(ib32%4)+is] >> 4*(ib32/4)) & 0x0f0f0f0f), 0x08080808);
const int8_t * s8 = (const int8_t *)&scales;
int2 val1;
const int * q2 = (const int *)bq2->qs + 8*ib32 + 4*is;
int aux32[2];
const uint8_t * aux8 = (const uint8_t *)aux32;
#pragma unroll
for (int i = 0; i < 4; ++i) {
auto values1 = all_values + (((bq2->extra[i+4*is] >> ib32) & 1) << 8);
int sumi1 = 0;
aux32[0] = ((q2[i] >> 0) & 0x03030303);
aux32[1] = ((q2[i] >> 2) & 0x03030303);
val1.x = int_from_table_4(aux8+0, values1);
val1.y = int_from_table_4(aux8+4, values1);
sumi1 = ggml_cuda_dp4a(val1.x, q8[0], ggml_cuda_dp4a(val1.y, q8[1], sumi1));
aux32[0] = ((q2[i] >> 4) & 0x03030303);
aux32[1] = ((q2[i] >> 6) & 0x03030303);
val1.x = int_from_table_4(aux8+0, values1);
val1.y = int_from_table_4(aux8+4, values1);
sumi1 = ggml_cuda_dp4a(val1.x, q8[2], ggml_cuda_dp4a(val1.y, q8[3], sumi1));
const float d = __half2float(bq2->d[i]) * d8;
result[i] += d * sumi1 * s8[i];
}
}
#define VDR_IQ3_K_Q8_1_MMVQ 4
#define VDR_IQ3_K_Q8_1_MMQ 4