jinhongyii commented on code in PR #16098:
URL: https://github.com/apache/tvm/pull/16098#discussion_r1391658093
##########
python/tvm/relax/op/distributed/distributed.py:
##########
@@ -59,3 +59,28 @@ def redistribute(input: Expr, device_mesh: DeviceMesh,
placement: Placement) ->
The tensor after redistribution.
"""
return _ffi_api.redistribute(input, device_mesh, placement) # type: ignore
+
+
+def redistribute_replica_to_shard(input: Expr, num_workers: int, axis: int) ->
Expr:
Review Comment:
I don't understand the case you are talking about. Why do we want to use a
symbolic var for a value that is constant after lowering? And why will we have
a variety of static case to specialize, given disco runtime regards
`num_workers` as a global constant?
--
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: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
