junrushao1994 commented on a change in pull request #10366:
URL: https://github.com/apache/tvm/pull/10366#discussion_r841031450
##########
File path: tests/python/unittest/test_meta_schedule_tune_tir.py
##########
@@ -98,14 +97,14 @@ def test_tune_matmul_cuda_tensor_core():
target = Target("nvidia/geforce-rtx-3070")
config = ReplayTraceConfig(
num_trials_per_iter=32,
- num_trials_total=320,
+ max_trials_per_task=320,
Review comment:
> I understand the risk of being an early user, but I'd like to see at
least reasonable documentation of the API at all times, regardless of being
stable / unstable. Especially when you introduce a breaking change. In this
particular case, the new attribute `max_trials_global` is not documented and
there is no example of how to use this value, as far as I can see.
That's a reasonable ask. I'm drafting a bugfix PR which will in the mean
time make sure the document is clear. I'm sorry that it could potentially miss
some corner cases during our death march last week on performance alignment in
BERT.
A further step, I would love to have a user-facing API audit with you and
the rest of the community, so that we could build broader consensus of what
things should look like.
--
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]
