KBaba7/llama.cpp
0
1#version 4502 3#include "types.comp"4#include "generic_binary_head.comp"5 6const uint num_threads = 256;7 8layout(local_size_x = num_threads, local_size_y = 1, local_size_z = 1) in;9 10void main() {11 uint idx = get_idx();12 13 // num_threads * num_iter must equal 512, to match the wg_denoms and get_idx calculation14 const uint num_iter = 2;15 16 [[unroll]] for (uint i = 0; i < num_iter; ++i) {17 if (idx >= p.ne) {18 continue;19 }20 uint i00, i01, i02, i03;21 get_indices(idx, i00, i01, i02, i03);22 23 data_d[get_doffset() + dst_idx(i00, i01, i02, i03)] = D_TYPE(FLOAT_TYPE(data_a[get_aoffset() + src0_idx(i00, i01, i02, i03)]) / FLOAT_TYPE(data_b[get_boffset() + src1_idx(i00, i01, i02, i03)]));24 25 idx += num_threads;26 }27}28 