Advertisement
Not a member of Pastebin yet?
Sign Up,
it unlocks many cool features!
- int bnum = 0;
- buf[tid] = in[idx];
- barrier(CLK_LOCAL_MEM_FENCE);
- for (int i = 1; i < dim; i*=2) {
- bnum = 1 - bnum;
- if (tid >= i) {
- buf[bnum*dim+tid] = buf[(1-bnum)*dim+tid] + buf[(1-bnum)*dim+tid-i];
- } else {
- buf[bnum*dim+tid] = buf[(1-bnum)*dim+tid];
- }
- barrier(CLK_LOCAL_MEM_FENCE);
- }
- out[idx] = buf[bnum*dim+tid];
- barrier(CLK_LOCAL_MEM_FENCE);
- if (tid == 0) {
- if (dim < n) {
- bout[gid] = buf[dim-1];
- } else {
- bout[gid] = buf[n-1];
- }
- }
- the problematic output:
- 116: cpu=58 gpu=58
- 117: cpu=59 gpu=59
- 118: cpu=60 gpu=60
- 119: cpu=60 gpu=60
- 120: cpu=60 gpu=60
- 121: cpu=61 gpu=61
- 122: cpu=61 gpu=61
- 123: cpu=62 gpu=62
- 124: cpu=62 gpu=62
- 125: cpu=62 gpu=62
- 126: cpu=62 gpu=62
- 127: cpu=62 gpu=62
- 128: cpu=63 gpu=26
- 129: cpu=64 gpu=27
- 130: cpu=64 gpu=27
- 131: cpu=65 gpu=28
- 132: cpu=66 gpu=29
- 133: cpu=67 gpu=30
- 134: cpu=68 gpu=31
- 135: cpu=69 gpu=32
- 136: cpu=70 gpu=33
- 137: cpu=71 gpu=34
- 138: cpu=71 gpu=34
- 139: cpu=72 gpu=35
- 140: cpu=72 gpu=35
- 141: cpu=73 gpu=36
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement