char-fork/llama.cpp

git clone https://git.t4t.associates/char-fork/llama.cpp

lhezopencl: split `ggml-opencl.cl` into multiple files and cleanup (#12886)80f19b418

char/aliceai-t5-moe
529 B16 linesraw
1#pragma OPENCL EXTENSION cl_khr_fp16 : enable
2
3//------------------------------------------------------------------------------
4// relu
5//------------------------------------------------------------------------------
6kernel void kernel_relu(
7        global float * src0,
8        ulong offset0,
9        global float * dst,
10        ulong offsetd
11) {
12    src0 = (global float*)((global char*)src0 + offset0);
13    dst = (global float*)((global char*)dst + offsetd);
14
15    dst[get_global_id(0)] = fmax(0.0f, src0[get_global_id(0)]);
16}