CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-11-15 11:17:31 +00:00
Code Issues Packages Projects Releases Wiki Activity

ggml : implement GEGLU_ERF and GEGLU_QUICK ops (#14445)

Browse Source
This commit is contained in:
Sigbjørn Skjæret
2025-07-03 23:07:22 +02:00
committed by GitHub
parent bee28421be
commit 28657a8229
20 changed files with 789 additions and 32 deletions
Download Patch File Download Diff File Expand all files Collapse all files

36
ggml/src/ggml-opencl/ggml-opencl.cpp
View File

@@ -402,8 +402,8 @@ struct ggml_backend_opencl_context {
cl_kernel kernel_relu;
cl_kernel kernel_sigmoid_f32, kernel_sigmoid_f16;
cl_kernel kernel_clamp;
cl_kernel kernel_geglu, kernel_reglu, kernel_swiglu,
kernel_geglu_f16, kernel_reglu_f16, kernel_swiglu_f16;
cl_kernel kernel_geglu, kernel_reglu, kernel_swiglu, kernel_geglu_erf, kernel_geglu_quick,
kernel_geglu_f16, kernel_reglu_f16, kernel_swiglu_f16, kernel_geglu_erf_f16, kernel_geglu_quick_f16;
cl_kernel kernel_norm;
cl_kernel kernel_rms_norm;
cl_kernel kernel_group_norm;
@@ -753,12 +753,16 @@ static void load_cl_kernels(ggml_backend_opencl_context *backend_ctx, ggml_cl_ve
backend_ctx->program_glu =
build_program_from_source(backend_ctx->context, backend_ctx->device, kernel_src.c_str(), compile_opts);
CL_CHECK((backend_ctx->kernel_geglu = clCreateKernel(backend_ctx->program_glu, "kernel_geglu", &err), err));
CL_CHECK((backend_ctx->kernel_reglu = clCreateKernel(backend_ctx->program_glu, "kernel_reglu", &err), err));
CL_CHECK((backend_ctx->kernel_swiglu = clCreateKernel(backend_ctx->program_glu, "kernel_swiglu", &err), err));
CL_CHECK((backend_ctx->kernel_geglu_f16 = clCreateKernel(backend_ctx->program_glu, "kernel_geglu_f16", &err), err));
CL_CHECK((backend_ctx->kernel_reglu_f16 = clCreateKernel(backend_ctx->program_glu, "kernel_reglu_f16", &err), err));
CL_CHECK((backend_ctx->kernel_swiglu_f16 = clCreateKernel(backend_ctx->program_glu, "kernel_swiglu_f16", &err), err));
CL_CHECK((backend_ctx->kernel_geglu = clCreateKernel(backend_ctx->program_glu, "kernel_geglu", &err), err));
CL_CHECK((backend_ctx->kernel_reglu = clCreateKernel(backend_ctx->program_glu, "kernel_reglu", &err), err));
CL_CHECK((backend_ctx->kernel_swiglu = clCreateKernel(backend_ctx->program_glu, "kernel_swiglu", &err), err));
CL_CHECK((backend_ctx->kernel_geglu_erf = clCreateKernel(backend_ctx->program_glu, "kernel_geglu_erf", &err), err));
CL_CHECK((backend_ctx->kernel_geglu_quick = clCreateKernel(backend_ctx->program_glu, "kernel_geglu_quick", &err), err));
CL_CHECK((backend_ctx->kernel_geglu_f16 = clCreateKernel(backend_ctx->program_glu, "kernel_geglu_f16", &err), err));
CL_CHECK((backend_ctx->kernel_reglu_f16 = clCreateKernel(backend_ctx->program_glu, "kernel_reglu_f16", &err), err));
CL_CHECK((backend_ctx->kernel_swiglu_f16 = clCreateKernel(backend_ctx->program_glu, "kernel_swiglu_f16", &err), err));
CL_CHECK((backend_ctx->kernel_geglu_erf_f16 = clCreateKernel(backend_ctx->program_glu, "kernel_geglu_erf_f16", &err), err));
CL_CHECK((backend_ctx->kernel_geglu_quick_f16 = clCreateKernel(backend_ctx->program_glu, "kernel_geglu_quick_f16", &err), err));
GGML_LOG_CONT(".");
}
@@ -2277,6 +2281,8 @@ static bool ggml_opencl_supports_op(ggml_backend_dev_t dev, const struct ggml_te
case GGML_GLU_OP_GEGLU:
case GGML_GLU_OP_REGLU:
case GGML_GLU_OP_SWIGLU:
case GGML_GLU_OP_GEGLU_ERF:
case GGML_GLU_OP_GEGLU_QUICK:
return ggml_is_contiguous_1(op->src[0]) && (op->type == GGML_TYPE_F32 || op->type == GGML_TYPE_F16);
default:
return false;
@@ -6254,6 +6260,20 @@ static void ggml_cl_glu(ggml_backend_t backend, const ggml_tensor * src0, const
kernel = backend_ctx->kernel_swiglu_f16;
}
break;
case GGML_GLU_OP_GEGLU_ERF:
if (dst->type == GGML_TYPE_F32) {
kernel = backend_ctx->kernel_geglu_erf;
} else {
kernel = backend_ctx->kernel_geglu_erf_f16;
}
break;
case GGML_GLU_OP_GEGLU_QUICK:
if (dst->type == GGML_TYPE_F32) {
kernel = backend_ctx->kernel_geglu_quick;
} else {
kernel = backend_ctx->kernel_geglu_quick_f16;
}
break;
default:
GGML_ABORT("Unsupported glu op");
}