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

Fix error in convert f16 to f32 kernel call

Browse source
This commit is contained in:
0cc4m 2023-05-16 13:05:33 +02:00
parent 342d346c13
commit b3ff66d87f
1 changed files with 1 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

2
ggml-opencl.cpp
View file

@ -801,7 +801,7 @@ static void ggml_cl_mul_mat_q_f32(const ggml_tensor * src0, const ggml_tensor *
CL_CHECK(clSetKernelArg(*to_fp32_cl, 0, sizeof(cl_mem), &d_Q), "clSetKernelArg");
CL_CHECK(clSetKernelArg(*to_fp32_cl, 1, sizeof(cl_mem), &d_X), "clSetKernelArg");
CL_CHECK(clFinish(queue), "clFinish");
CL_CHECK(clEnqueueNDRangeKernel(queue, *to_fp32_cl, 1, NULL, &global, &local, 0, NULL, NULL), "clEnqueueNDRangeKernel");
CL_CHECK(clEnqueueNDRangeKernel(queue, *to_fp32_cl, 1, NULL, &global, type == GGML_TYPE_F16 ? NULL : &local, 0, NULL, NULL), "clEnqueueNDRangeKernel");
// copy src1 to device
CL_CHECK(ggml_cl_h2d_tensor_2d(queue, d_Y, 0, src1, i03, i02, NULL), "ggml_cl_h2d_tensor_2d");