CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-11-07 09:57:00 +00:00
Code Issues Packages Projects Releases Wiki Activity

cpu: im2col_3d support non continuous src

Browse Source 
Co-authored-by: Jeff Bolz <jbolz@nvidia.com>
This commit is contained in:
leejet
2025-08-31 11:58:32 +08:00
committed by GitHub
parent 3f901e316b
commit e66bf6e503
1 changed files with 6 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
ggml/src/ggml-cpu/ops.cpp
View File

@@ -7095,7 +7095,7 @@ static void ggml_compute_forward_im2col_3d_f16(
// micro kernel // micro kernel
ggml_fp16_t * dst_data = wdata + (in*OD*OH_OW + iod*OH_OW + ioh*OW + iow)*IC_KD_KH_KW; // [IC, KD, KH, KW] ggml_fp16_t * dst_data = wdata + (in*OD*OH_OW + iod*OH_OW + ioh*OW + iow)*IC_KD_KH_KW; // [IC, KD, KH, KW]
const float * const src_data = (float *) src1->data + (in*IC + iic)*ID_IH_IW; // [ID, IH, IW] const float * const src_data = (const float *) ((const char *)src1->data + (in*IC + iic)*nb13); // [ID, IH, IW]
for (int64_t ikd = 0; ikd < KD; ikd++) { for (int64_t ikd = 0; ikd < KD; ikd++) {
for (int64_t ikh = 0; ikh < KH; ikh++) { for (int64_t ikh = 0; ikh < KH; ikh++) {
@@ -7107,7 +7107,8 @@ static void ggml_compute_forward_im2col_3d_f16(
if (iid < 0 || iid >= ID || iih < 0 || iih >= IH || iiw < 0 || iiw >= IW || iid < 0 || iid >= ID) { if (iid < 0 || iid >= ID || iih < 0 || iih >= IH || iiw < 0 || iiw >= IW || iid < 0 || iid >= ID) {
dst_data[iic*KD_KH_KW + ikd * KH_KW + ikh*KW + ikw] = 0; dst_data[iic*KD_KH_KW + ikd * KH_KW + ikh*KW + ikw] = 0;
} else { } else {
dst_data[iic*KD_KH_KW + ikd * KH_KW + ikh*KW + ikw] = GGML_CPU_FP32_TO_FP16(src_data[iid*IH_IW + iih*IW + iiw]); const float * const s = (const float *) ((const char *)src_data + iid*nb12 + iih*nb11 + iiw*nb10); // [ID, IH, IW]
dst_data[iic*KD_KH_KW + ikd * KH_KW + ikh*KW + ikw] = GGML_CPU_FP32_TO_FP16(*s);
} }
} }
} }
@@ -7187,7 +7188,7 @@ static void ggml_compute_forward_im2col_3d_f32(
// micro kernel // micro kernel
float * dst_data = wdata + (in*OD*OH_OW + iod*OH_OW + ioh*OW + iow)*IC_KD_KH_KW; // [IC, KD, KH, KW] float * dst_data = wdata + (in*OD*OH_OW + iod*OH_OW + ioh*OW + iow)*IC_KD_KH_KW; // [IC, KD, KH, KW]
const float * const src_data = (float *) src1->data + (in*IC + iic)*ID_IH_IW; // [ID, IH, IW] const float * const src_data = (const float *) ((const char *)src1->data + (in*IC + iic)*nb13); // [ID, IH, IW]
for (int64_t ikd = 0; ikd < KD; ikd++) { for (int64_t ikd = 0; ikd < KD; ikd++) {
for (int64_t ikh = 0; ikh < KH; ikh++) { for (int64_t ikh = 0; ikh < KH; ikh++) {
@@ -7199,7 +7200,8 @@ static void ggml_compute_forward_im2col_3d_f32(
if (iid < 0 || iid >= ID || iih < 0 || iih >= IH || iiw < 0 || iiw >= IW || iid < 0 || iid >= ID) { if (iid < 0 || iid >= ID || iih < 0 || iih >= IH || iiw < 0 || iiw >= IW || iid < 0 || iid >= ID) {
dst_data[iic*KD_KH_KW + ikd * KH_KW + ikh*KW + ikw] = 0; dst_data[iic*KD_KH_KW + ikd * KH_KW + ikh*KW + ikw] = 0;
} else { } else {
dst_data[iic*KD_KH_KW + ikd * KH_KW + ikh*KW + ikw] = src_data[iid*IH_IW + iih*IW + iiw]; const float * const s = (const float *) ((const char *)src_data + iid*nb12 + iih*nb11 + iiw*nb10); // [ID, IH, IW]
dst_data[iic*KD_KH_KW + ikd * KH_KW + ikh*KW + ikw] = *s;
} }
} }
} }