jinhongyii commented on code in PR #16098:
URL: https://github.com/apache/tvm/pull/16098#discussion_r1393323681
##########
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:
@Lunderberg I just come up with an additional question on the attrs->args
change: Do we need to recompile each time for different num_workers after this
change with specialization? If yes, then what's the difference between
define-symbolic->specialize flow and assume-constant flow?
--
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]
