[GitHub] cjolivier01 commented on issue #9727: Refactored storage handling

2018-02-08 Thread GitBox
cjolivier01 commented on issue #9727: Refactored storage handling
URL: https://github.com/apache/incubator-mxnet/pull/9727#issuecomment-364188669
 
 
   How many reference count changes occur for a given pointer during alloc? 
during free?  Is it accessed anywhere during its useful lifetime in such a way 
that it causes a reference count change? (ie is there an assignment operation 
where a copy of the shared pointer is taken (thus increasing the reference 
count) anywhere or is the shared pointer just used from its place in a 
class/struct data member)?
   
   What's the code path/situation where it might have a problem that makes it 
need to be reference-counted?  I there a situation where it might be destroyed 
while another thread is using it? If so, where?


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


[GitHub] cjolivier01 commented on issue #9727: Refactored storage handling

2018-02-08 Thread GitBox
cjolivier01 commented on issue #9727: Refactored storage handling
URL: https://github.com/apache/incubator-mxnet/pull/9727#issuecomment-364189590
 
 
   I think for the performance question, piiswrong@ is talking about the fact 
that the cache/memory may be locked/discarded during the interlocked 
increment/decrement (i.e. "lock inc ds:[esi]")


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


[GitHub] cjolivier01 commented on issue #9727: Refactored storage handling

2018-02-08 Thread GitBox
cjolivier01 commented on issue #9727: Refactored storage handling
URL: https://github.com/apache/incubator-mxnet/pull/9727#issuecomment-364189590
 
 
   I think for the performance question, piiswrong@ is talking about the fact 
that the cache/memory may be locked during the interlocked increment/decrement 
(i.e. "lock inc ds:[esi]")


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


[GitHub] cjolivier01 commented on issue #9727: Refactored storage handling

2018-02-08 Thread GitBox
cjolivier01 commented on issue #9727: Refactored storage handling
URL: https://github.com/apache/incubator-mxnet/pull/9727#issuecomment-364188669
 
 
   How many reference count changes occur for a given pointer during alloc? 
during free?  Is it accessed anywhere during its useful lifetime in such a way 
that it causes a reference count change?
   
   What's the code path/situation where it might have a problem that makes it 
need to be reference-counted?  I there a situation where it might be destroyed 
while another thread is using it? If so, where?


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


[GitHub] cjolivier01 commented on issue #9727: Refactored storage handling

2018-02-08 Thread GitBox
cjolivier01 commented on issue #9727: Refactored storage handling
URL: https://github.com/apache/incubator-mxnet/pull/9727#issuecomment-364188669
 
 
   How many reference count changes occur for a given pointer during alloc? 
during free?  Is it accessed anywhere during its useful lifetime in such a way 
that it causes a reference count change?
   
   What's the code path/situation where it might have a problem that makes it 
need to be reference-counted?  I there a situation where it might be destroyed 
why another thread is using it? If so, where?


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