ver4a/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

ggml : Depthwise 2D convolution (ggml/1152)

Browse source 
* ggml-cpu : kernels for faster depthwise 2D convolution

* fix compile: remove static after moving to ops.cpp

* add dilation for depthwise_conv_2d

* review: rename to ggml_conv_2d_dw_direct, remove redundant struct keywords, pass by ref, whitespace

* review: rename depthwise_conv_2d -> conv_2d_dw everywhere
This commit is contained in:
Acly 2025-04-17 14:16:45 +02:00 • committed by Georgi Gerganov
parent b10d8bfdb1
commit c6e8cc28c1
5 changed files with 250 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

5
ggml/src/ggml-cpu/ggml-cpu.c
View file

@ -1932,6 +1932,10 @@ static void ggml_compute_forward(struct ggml_compute_params * params, struct ggm
{
ggml_compute_forward_im2col_back_f32(params, tensor);
} break;
case GGML_OP_CONV_2D_DW:
{
ggml_compute_forward_conv_2d_dw(params, tensor);
} break;
case GGML_OP_CONV_TRANSPOSE_2D:
{
ggml_compute_forward_conv_transpose_2d(params, tensor);
@ -2268,6 +2272,7 @@ static int ggml_get_n_tasks(struct ggml_tensor * node, int n_threads) {
} break;
case GGML_OP_IM2COL:
case GGML_OP_IM2COL_BACK:
case GGML_OP_CONV_2D_DW:
case GGML_OP_CONV_TRANSPOSE_1D:
case GGML_OP_CONV_TRANSPOSE_2D:
{