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 2022/02/01 00:50:10 UTC

[GitHub] [tvm] jwfromm opened a new pull request #10123: [AutoScheduler] Allow device specification for AutoScheduler Runners.

jwfromm opened a new pull request #10123:
URL: https://github.com/apache/tvm/pull/10123


   Currently, all measurements in the autoscheduler are run on device 0. Although this is fine for most cases, on some platforms with multiple gpus, it may be important to run on a specific device. This small PR adds an optional `device` argument to the various `Runner` classes in `auto_scheduler/measure` to address this limitation.
   


-- 
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.

To unsubscribe, e-mail: commits-unsubscribe@tvm.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org



[GitHub] [tvm] jwfromm commented on pull request #10123: [AutoScheduler] Allow device specification for AutoScheduler Runners.

Posted by GitBox <gi...@apache.org>.
jwfromm commented on pull request #10123:
URL: https://github.com/apache/tvm/pull/10123#issuecomment-1026360930


   @tkonolige can you take a look at this PR?


-- 
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.

To unsubscribe, e-mail: commits-unsubscribe@tvm.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org



[GitHub] [tvm] masahi merged pull request #10123: [AutoScheduler] Allow device specification for AutoScheduler Runners.

Posted by GitBox <gi...@apache.org>.
masahi merged pull request #10123:
URL: https://github.com/apache/tvm/pull/10123


   


-- 
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.

To unsubscribe, e-mail: commits-unsubscribe@tvm.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org



[GitHub] [tvm] tkonolige commented on a change in pull request #10123: [AutoScheduler] Allow device specification for AutoScheduler Runners.

Posted by GitBox <gi...@apache.org>.
tkonolige commented on a change in pull request #10123:
URL: https://github.com/apache/tvm/pull/10123#discussion_r796187587



##########
File path: python/tvm/auto_scheduler/measure.py
##########
@@ -1257,6 +1276,8 @@ def rpc_runner_run(
         This is only has effect on CPU task.
     verbose: int = 1
         Verbosity level. 0 for silent, 1 to output information during program measuring.
+    device: int = 0
+        Which device to run on if multiople are available.

Review comment:
       ```suggestion
           Which device to run on if multiple are available.
   ```

##########
File path: tests/python/relay/test_auto_scheduler_tuning.py
##########
@@ -36,7 +36,7 @@ def tune_network(network, target):
         log_file = fp.name
 
         # Tuning
-        measure_ctx = auto_scheduler.LocalRPCMeasureContext(timeout=60)
+        measure_ctx = auto_scheduler.LocalRPCMeasureContext(timeout=60, device=0)

Review comment:
       It would be nice if there was some way to test this with device != 0, but I can't think of a good way.




-- 
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.

To unsubscribe, e-mail: commits-unsubscribe@tvm.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org



[GitHub] [tvm] jwfromm commented on a change in pull request #10123: [AutoScheduler] Allow device specification for AutoScheduler Runners.

Posted by GitBox <gi...@apache.org>.
jwfromm commented on a change in pull request #10123:
URL: https://github.com/apache/tvm/pull/10123#discussion_r796277944



##########
File path: tests/python/relay/test_auto_scheduler_tuning.py
##########
@@ -36,7 +36,7 @@ def tune_network(network, target):
         log_file = fp.name
 
         # Tuning
-        measure_ctx = auto_scheduler.LocalRPCMeasureContext(timeout=60)
+        measure_ctx = auto_scheduler.LocalRPCMeasureContext(timeout=60, device=0)

Review comment:
       Yeah I dont think it would work on any of the CI targets unfortunately.




-- 
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.

To unsubscribe, e-mail: commits-unsubscribe@tvm.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org



[GitHub] [tvm] jwfromm commented on pull request #10123: [AutoScheduler] Allow device specification for AutoScheduler Runners.

Posted by GitBox <gi...@apache.org>.
jwfromm commented on pull request #10123:
URL: https://github.com/apache/tvm/pull/10123#issuecomment-1027370076


   @masahi can you also check to make sure this all looks ok?


-- 
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.

To unsubscribe, e-mail: commits-unsubscribe@tvm.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org