[GitHub] rahul003 commented on a change in pull request #10183: [MXNET-120] Float16 support for distributed training

[GitBox <gi...@apache.org>]

rahul003 commented on a change in pull request #10183: [MXNET-120] Float16 support for distributed training
URL: https://github.com/apache/incubator-mxnet/pull/10183#discussion_r176194754
 
 

 ##########
 File path: src/kvstore/kvstore_dist.h
 ##########
 @@ -619,15 +629,17 @@ class KVStoreDist : public KVStoreLocal {
           push_pskv.keys.push_back(ps_key);
           pull_pskv.keys.push_back(ps_key);
           // push_pskv stores lengths of compressed blocks
-          push_pskv.lens.push_back(part_compr);
+          push_pskv.lens.push_back(part_compr * num_bytes);
           // pull_pskv stores lengths of original data
-          pull_pskv.lens.push_back(part_orig);
+          pull_pskv.lens.push_back(part_orig * num_bytes);
           push_pskv.size += part_compr;
           pull_pskv.size += part_orig;
         }
-        CHECK_EQ(static_cast<size_t>(push_pskv.size), compr_size);
-        CHECK_EQ(static_cast<size_t>(pull_pskv.size), original_size);
-        CHECK_EQ(push_pskv.lens.size(), num_servers*2);
+        push_pskv.size *= num_bytes;
+        pull_pskv.size *= num_bytes;
+        CHECK_EQ(static_cast<size_t>(push_pskv.size), compr_size * num_bytes);
+        CHECK_EQ(static_cast<size_t>(pull_pskv.size), original_size * num_bytes);
+        CHECK_EQ(push_pskv.lens.size(), num_servers * 2);
 
 Review comment:
   Good idea, will do
   

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
users@infra.apache.org


With regards,
Apache Git Services