xutianming opened a new issue #6356:
URL: https://github.com/apache/incubator-tvm/issues/6356


   Dear developers,
   
   I was optimizing a TextCNN model with TVM on intel x86.
   I wrote my own Conv1D NCWc schedule based on Conv2D
   
   
![image](https://user-images.githubusercontent.com/4970790/91565965-8fba0600-e975-11ea-85a4-83d66b2cbdf3.png)
   
   TVM stack only has operator-level performance tools. 
   **How could I further locate the hot-spot of the operator**  ?
   
   I tried gdb, and the ZMM registers was utilized.
   
![image](https://user-images.githubusercontent.com/4970790/91566257-f808e780-e975-11ea-9c36-abc16099fb04.png)
   
   I also tried linux-perf, but didn't get much clue.
   
   
   I asked the same question on [TVM 
discuss](https://discuss.tvm.ai/t/how-to-further-improve-the-performance-of-given-schedule/7711),
 but got no reply.
   
   Sincerely hope for your reply.
   


----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

For queries about this service, please contact Infrastructure at:
[email protected]


Reply via email to