Skip to content

ggml : add ggml_gelu_erf() CUDA kernel #13719

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

Open
wants to merge 2 commits into
base: master
Choose a base branch
from
+16 −0
Open

ggml : add ggml_gelu_erf() CUDA kernel #13719

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
9d89543
ggml : add ggml_gelu_erf() CUDA kernel
ngxson May 23, 2025
242068c
missing semicolon
ngxson May 23, 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
4 changes: 4 additions & 0 deletions ggml/src/ggml-cuda/ggml-cuda.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2192,6 +2192,9 @@ static bool ggml_cuda_compute_forward(ggml_backend_cuda_context & ctx, struct gg
case GGML_UNARY_OP_SILU:
ggml_cuda_op_silu(ctx, dst);
break;
case GGML_UNARY_OP_GELU_ERF:
ggml_cuda_op_gelu_erf(ctx, dst);
break;
case GGML_UNARY_OP_GELU_QUICK:
ggml_cuda_op_gelu_quick(ctx, dst);
break;
Expand Down Expand Up @@ -2977,6 +2980,7 @@ static bool ggml_backend_cuda_device_supports_op(ggml_backend_dev_t dev, const g
case GGML_UNARY_OP_SIGMOID:
case GGML_UNARY_OP_HARDSIGMOID:
case GGML_UNARY_OP_HARDSWISH:
case GGML_UNARY_OP_GELU_ERF:
case GGML_UNARY_OP_GELU_QUICK:
case GGML_UNARY_OP_TANH:
case GGML_UNARY_OP_EXP:
Expand Down
10 changes: 10 additions & 0 deletions ggml/src/ggml-cuda/unary.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -23,6 +23,12 @@ static __device__ __forceinline__ float op_gelu(float x) {
return 0.5f*x*(1.0f + tanhf(SQRT_2_OVER_PI*x*(1.0f + GELU_COEF_A*x*x)));
}

static __device__ __forceinline__ float op_gelu_erf(float x) {
const float SQRT_2_INV = 0.70710678118654752440084436210484f;

return 0.5f*x*(1.0f + erff(x*SQRT_2_INV));
}

static __device__ __forceinline__ float op_gelu_quick(float x) {
const float GELU_QUICK_COEF = -1.702f;

Expand Down Expand Up @@ -134,6 +140,10 @@ void ggml_cuda_op_gelu(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
ggml_cuda_op_unary<op_gelu>(ctx, dst);
}

void ggml_cuda_op_gelu_erf(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
ggml_cuda_op_unary<op_gelu_erf>(ctx, dst);
}

void ggml_cuda_op_gelu_quick(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
ggml_cuda_op_unary<op_gelu_quick>(ctx, dst);
}
Expand Down
2 changes: 2 additions & 0 deletions ggml/src/ggml-cuda/unary.cuh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -30,6 +30,8 @@ void ggml_cuda_op_silu(ggml_backend_cuda_context & ctx, ggml_tensor * dst);

void ggml_cuda_op_silu_back(ggml_backend_cuda_context & ctx, ggml_tensor * dst);

void ggml_cuda_op_gelu_erf(ggml_backend_cuda_context & ctx, ggml_tensor * dst);

void ggml_cuda_op_gelu_quick(ggml_backend_cuda_context & ctx, ggml_tensor * dst);

void ggml_cuda_op_tanh(ggml_backend_cuda_context & ctx, ggml_tensor * dst);
Expand Down
Loading