aurel333 opened a new pull request #6412:
URL: https://github.com/apache/incubator-tvm/pull/6412


   Thanks for contributing to TVM!   Please refer to guideline 
https://tvm.apache.org/docs/contribute/ for useful information and tips. After 
the pull request is submitted, please request code reviews from 
[Reviewers](https://github.com/apache/incubator-tvm/blob/master/CONTRIBUTORS.md#reviewers)
 by @ them in the pull request thread.
   
   As TVM supports OpenCL and OpenCL supports accelerators in the same way as 
GPU or CPU devices, I think TVM should accept generic accelerators too. 
   
   I made minor changes to the OpenCL c++ part in order to make TVM recognize 
accelerator devices. However it may deserve another pull request to overhaul 
the present system in order to allow the user to explicitly chose between which 
device type to use.
   In addition I noticed that CUDA, and therefore OpenCL, schedules used 
sometimes hardcoded constants for the number of threads. While it is acceptable 
for GPU that have many threads available, it is not for generic accelerators 
that may have a small number of threads (as low as 16 for example). I tried to 
make these constants more flexible by using the max_num_threads of the target 
kind.


----------------------------------------------------------------
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