New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
GH-103906: Remove immortal refcounting in the interpreter #103909
Conversation
brandtbucher
commented
Apr 26, 2023
•
edited by bedevere-bot
edited by bedevere-bot
- Issue: Avoid reference counting operations on None, True, False in the interpreter #103906
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Hooray! Looks great
} | ||
else if (Py_IsFalse(cond)) { | ||
_Py_DECREF_NO_DEALLOC(cond); | ||
if (Py_IsFalse(cond)) { |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Will this cause a performance change, since formerly IsTrue
then IsFalse
, now IsFalse
then IsTrue
Though this comparison is lightweight.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Honestly, I doubt it.
If you want to schedule another build, you need to add the |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
What about other places, like Modules/
? Should it be done later?
@sobolevn possibly, but it's best left to a different PR. In many places it's unlikely to make a significant difference, so it might not be worth the churn to make the change everywhere. |
Annoyingly, perf impact is neutral. However, it appears that the results are being pulled way down by a couple of known unstable benchmarks that should be unaffected by this change. I might try running them again tonight, just to see. (I still think this is worth doing, though, as it clearly removes unnecessary work from many hot code paths.) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Even if the performance change is negligible, this seems like a worthwhile cleanup.
I noticed one more opportunity like this: in the |
|