Skip to content

Conversation

markshannon
Copy link
Member

@markshannon markshannon commented Apr 19, 2024

Stats show about a 10% increase in traces executed and uops executed, with a corresponding reduction in tier 1 instructions executed.

@markshannon
Copy link
Member Author

CI failures are not new.

Copy link
Member

@gvanrossum gvanrossum left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LG, but what about the benchmark claiming it's 1% slower?

@@ -689,7 +689,7 @@ translate_bytecode_to_trace(
if (expansion->nuops > 0) {
// Reserve space for nuops (+ _SET_IP + _EXIT_TRACE)
int nuops = expansion->nuops;
RESERVE(nuops);
RESERVE(nuops + 1); /* One extra for exit */
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I'm taking this is for gh-117434, right?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It seems likely to be the same issue, but I haven't checked.

@markshannon
Copy link
Member Author

LG, but what about the benchmark claiming it's 1% slower?

That's the tier 2 interpreter. The generated code for tier 1 is unchanged.

@markshannon markshannon merged commit 77cd042 into python:main Apr 24, 2024
@markshannon markshannon deleted the convert-deopts-to-exits branch August 6, 2024 10:16
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants