OrgLion-ML
/
dlib
mirror of https://github.com/davisking/dlib.git
1
0
Fork 0
Code Issues Releases Wiki Activity

Fixed spelling error in comment

This commit is contained in:
Davis King 2016-09-29 08:57:01 -04:00
parent 2a0b461926
commit 776678a37c
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
dlib/dnn/cuda_utils.h
View File

@ -261,7 +261,7 @@ namespace dlib
CUDA's y thread index (e.g. threadIdx.y) instead of the x index.
Therefore, if you launch a cuda kernel with a statement like:
dim3 blocks(10,1);
dim3 threads(32,32); // You need to have x any not equal to 1 to get parallelism over both loops.
dim3 threads(32,32); // You need to have x and y not equal to 1 to get parallelism over both loops.
add_arrays<<<blocks,threads>>>(a,b,out,nr,nc);
You can perform a nested 2D parallel for loop rather than doing just a
1D for loop.