Skip to content

Add conv2d Implicit GEMM #15805

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
wants to merge 9 commits into from
Closed

Add conv2d Implicit GEMM #15805

Show file tree
Hide file tree
Changes from 5 commits
Commits
Show all changes
9 commits
Select commit Hold shift + click to select a range
8a58931
Add implicit GEMM convolution operation for 2D tensors in CUDA
bssrdf Sep 3, 2025
4d77287
Add implicit convolution support for 2D tensors in CPU and CUDA imple…
bssrdf Sep 3, 2025
3877608
fix passing param as reference
bssrdf Sep 3, 2025
6d84cbb
Fix parameter order in conv2d_implicit and add comprehensive test cas…
bssrdf Sep 3, 2025
5ffe97b
Fix boundary check in conv2d_implicit_kernel to include channel limits
bssrdf Sep 4, 2025
4b0f9d5
Refactor conv2d_implicit_kernel for improved readability and consiste…
bssrdf Sep 5, 2025
83a3b7d
Refactor conv2d_implicit_kernel for improved bitwise operations; add …
bssrdf Sep 6, 2025
735886b
merged with upstream master
bssrdf Sep 10, 2025
2ec76aa
Merge branch 'master' into conv2d-implicit
bssrdf Sep 11, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 12 additions & 0 deletions ggml/include/ggml.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -512,6 +512,7 @@ extern "C" {
GGML_OP_IM2COL,
GGML_OP_IM2COL_BACK,
GGML_OP_CONV_2D,
GGML_OP_CONV_2D_IMPLICIT,
GGML_OP_CONV_3D,
GGML_OP_CONV_2D_DW,
GGML_OP_CONV_TRANSPOSE_2D,
Expand Down Expand Up @@ -1941,6 +1942,17 @@ extern "C" {
int d0, // dilation dimension 0
int d1); // dilation dimension 1

GGML_API struct ggml_tensor * ggml_conv_2d_implicitgemm(
struct ggml_context * ctx,
struct ggml_tensor * a, // convolution kernel [KW, KH, IC, OC]
struct ggml_tensor * b, // input data [W, H, C, N]
int s0, // stride dimension 0
int s1, // stride dimension 1
int p0, // padding dimension 0
int p1, // padding dimension 1
int d0, // dilation dimension 0
int d1); // dilation dimension 1

GGML_API struct ggml_tensor * ggml_conv_3d(
struct ggml_context * ctx,
struct ggml_tensor * a, // kernel [KW, KH, KD, IC * OC]
Expand Down
6 changes: 6 additions & 0 deletions ggml/src/ggml-cpu/ggml-cpu.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1880,6 +1880,10 @@ static void ggml_compute_forward(struct ggml_compute_params * params, struct ggm
{
ggml_compute_forward_conv_2d(params, tensor);
} break;
case GGML_OP_CONV_2D_IMPLICIT:
{
ggml_compute_forward_conv_2d(params, tensor);
} break;
case GGML_OP_CONV_3D:
{
ggml_compute_forward_conv_3d(params, tensor);
Expand Down Expand Up @@ -2256,6 +2260,7 @@ static int ggml_get_n_tasks(struct ggml_tensor * node, int n_threads) {
case GGML_OP_IM2COL:
case GGML_OP_IM2COL_BACK:
case GGML_OP_CONV_2D:
case GGML_OP_CONV_2D_IMPLICIT:
case GGML_OP_CONV_3D:
case GGML_OP_CONV_2D_DW:
case GGML_OP_CONV_TRANSPOSE_1D:
Expand Down Expand Up @@ -2778,6 +2783,7 @@ struct ggml_cplan ggml_graph_plan(
}
} break;
case GGML_OP_CONV_2D:
case GGML_OP_CONV_2D_IMPLICIT:
case GGML_OP_CONV_3D:
{
cur = GGML_IM2COL_WORK_SIZE;
Expand Down
Loading