ZiyueHuang commented on issue #7554: There is a problem in distribute training with dtype=fp16 URL: https://github.com/apache/incubator-mxnet/issues/7554#issuecomment-323993848 https://github.com/apache/incubator-mxnet/blob/master/src/kvstore/kvstore_dist.h#L243. `real_t* data = static_cast<real_t*>(send_buf.data().dptr_)` is an invalid cast if the underlying datatype is fp16. Although this line could compile because `dptr_` is `void*`. Actually `KVStoreDistServer` and `KVStoreDist` use `real_t` internally regardless of the datatype you specified. ---------------------------------------------------------------- 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: us...@infra.apache.org
With regards, Apache Git Services
