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

disable CL f16 for now until it's sorted out

Browse source
This commit is contained in:
Concedo 2023-05-18 14:23:56 +08:00
parent b8fe4bd37d
commit dae9a14b50
1 changed files with 3 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

4
ggml-opencl.cpp
View file

@ -311,7 +311,7 @@ static size_t cl_size_a = 0, cl_size_qb = 0, cl_size_b = 0, cl_size_c = 0;
static cl_kernel convert_fp16_to_fp32_cl; static cl_kernel convert_fp16_to_fp32_cl;
static cl_kernel dequantize_row_q4_0_cl, dequantize_row_q4_1_cl, dequantize_row_q5_0_cl, dequantize_row_q5_1_cl, dequantize_row_q8_0_cl; static cl_kernel dequantize_row_q4_0_cl, dequantize_row_q4_1_cl, dequantize_row_q5_0_cl, dequantize_row_q5_1_cl, dequantize_row_q8_0_cl;
static cl_kernel dequantize_mul_mat_vec_q4_0_cl, dequantize_mul_mat_vec_q4_1_cl, dequantize_mul_mat_vec_q5_0_cl, dequantize_mul_mat_vec_q5_1_cl, dequantize_mul_mat_vec_q8_0_cl, convert_mul_mat_vec_f16_cl; static cl_kernel dequantize_mul_mat_vec_q4_0_cl, dequantize_mul_mat_vec_q4_1_cl, dequantize_mul_mat_vec_q5_0_cl, dequantize_mul_mat_vec_q5_1_cl, dequantize_mul_mat_vec_q8_0_cl, convert_mul_mat_vec_f16_cl;
static bool fp16_support; static bool fp16_support = false;
static cl_program build_program_from_source(cl_context ctx, cl_device_id dev, const char* program_buffer) { static cl_program build_program_from_source(cl_context ctx, cl_device_id dev, const char* program_buffer) {
cl_program p; cl_program p;
@ -377,6 +377,8 @@ void ggml_cl_init(void) {
} }
free(ext_buffer); free(ext_buffer);
printf("Using Platform: %s Device: %s FP16: %d\n", platform_buffer, device_buffer, fp16_support); printf("Using Platform: %s Device: %s FP16: %d\n", platform_buffer, device_buffer, fp16_support);
fp16_support = false;
printf("CL FP16 temporarily disabled pending further optimization.\n");
context = clCreateContext(NULL, 1, &device, NULL, NULL, &err); context = clCreateContext(NULL, 1, &device, NULL, NULL, &err);
CL_CHECK(err, "clCreateContext"); CL_CHECK(err, "clCreateContext");
queue = clCreateCommandQueue(context, device, CL_QUEUE_OUT_OF_ORDER_EXEC_MODE_ENABLE, &err); queue = clCreateCommandQueue(context, device, CL_QUEUE_OUT_OF_ORDER_EXEC_MODE_ENABLE, &err);