Welcome to mirror list, hosted at ThFree Co, Russian Federation.

github.com/torch/torch7.git - Unnamed repository; edit this file 'description' to name the repository.
summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorjokeren <robinho364@gmail.com>2017-02-16 05:16:59 +0300
committerSoumith Chintala <soumith@gmail.com>2017-02-23 14:01:13 +0300
commit8940d7b488528a8484bf664ae4b1638af498912e (patch)
treebae0a186a3d80ce94b735de03ceddf60aff7c65c
parent08067cab497e37cc91cb5ca234258898d0916a75 (diff)
comments fix
-rw-r--r--lib/TH/THTensorApply.h8
1 files changed, 4 insertions, 4 deletions
diff --git a/lib/TH/THTensorApply.h b/lib/TH/THTensorApply.h
index bc180e6..490f871 100644
--- a/lib/TH/THTensorApply.h
+++ b/lib/TH/THTensorApply.h
@@ -502,10 +502,10 @@
* can simply loop for sizeof(A) iterations and perform the operation, without having to
* follow the order described by the strides of A.
*
- * 3. We find the contiguous sections in a tensor and compress the counter array.
- * For instance, if A is a 3x3x4x3 tensor, from which we get A' (3x3x3x3). Then, the
- * original counter for A' is the same as A. But actually the first dimension and the
- * second dimension can be collapsed into a single one.
+ * 3. As an optimization, we merge dimensions of A that are contiguous in memory. For
+ * example, if A is a 3x3x3x3 tensor narrowed from a 3x3x4x3 tensor, then the first two
+ * dimensions can be merged for the purposes of APPLY, reducing the number of nested
+ * loops.
*/
#define TH_TENSOR_APPLY(TYPE, TENSOR, CODE) \
{ \