Icemist commented on code in PR #11465:
URL: https://github.com/apache/tvm/pull/11465#discussion_r900014528
##########
python/tvm/contrib/debugger/debug_result.py:
##########
@@ -157,7 +155,12 @@ def s_to_us(t):
return t * 10**6
starting_times = np.zeros(len(self._time_list) + 1)
- starting_times[1:] = np.cumsum([times[0] for times in self._time_list])
+ starting_times[1:] = np.cumsum(
+ [
+ np.mean([np.mean(repeat_data) for repeat_data in node_data])
+ for node_data in self._time_list
+ ]
+ )
Review Comment:
Before my changes, it was an average of 1 repeat. Here. I returned not only
the first repeat of the node, but all of them for each node, and also in the
form of arrays as raw measurements.
Now I just return all repeats for each node, not just the first. So I take
the average here. Below where the duration is calculated, I also calculate the
average.
--
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]
