You are viewing a plain text version of this content. The canonical link for it is here.

Posted to commits@tvm.apache.org by GitBox <gi...@apache.org> on 2020/11/24 04:41:16 UTC

[GitHub] [incubator-tvm] merrymercy opened a new pull request #6965: [AutoScheduler] Fix task extraction

merrymercy opened a new pull request #6965:
URL: https://github.com/apache/incubator-tvm/pull/6965


   Followup of #6956 .
   The previous PR does not work for the cases where `select_implementation` is during alter op layout in TOPI.
   https://github.com/apache/incubator-tvm/blob/b2e418cb109df4cd1f17a2cf2894a1b396a6b838/python/tvm/topi/cuda/conv2d_alter_op.py#L47-L49 
   


----------------------------------------------------------------
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:
users@infra.apache.org

[GitHub] [incubator-tvm] merrymercy merged pull request #6965: [AutoScheduler] Fix task extraction

Posted by GitBox <gi...@apache.org>.

merrymercy merged pull request #6965:
URL: https://github.com/apache/incubator-tvm/pull/6965


   


----------------------------------------------------------------
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:
users@infra.apache.org

[GitHub] [incubator-tvm] merrymercy commented on a change in pull request #6965: [AutoScheduler] Fix task extraction

Posted by GitBox <gi...@apache.org>.

merrymercy commented on a change in pull request #6965:
URL: https://github.com/apache/incubator-tvm/pull/6965#discussion_r529263752



##########
File path: python/tvm/relay/backend/compile_engine.py
##########
@@ -186,6 +186,11 @@ def select_implementation(op, attrs, inputs, out_type, target, use_autotvm=True)
     all_impls = get_valid_implementations(op, attrs, inputs, out_type, target)
     best_plevel_impl = max(all_impls, key=lambda x: x.plevel)
 
+    # Disable autotvm if auto_scheduler is enabled.
+    # (i.e. always return the implementation with the highest priority for auto-scheduler).

Review comment:
       ```suggestion
       # (i.e., always return the implementation with the highest priority for auto-scheduler).
   ```




----------------------------------------------------------------
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:
users@infra.apache.org