After discussing this with a colleague, you do seem to be measuring a build
that
fundamentally looks like debug build. In release mode, GetOutputInfo only
gets
called once per optimized code object. So you are replacing a O(ln(n) * m)
with
a O(ln(n) * n + m * ln(n)) algorithm (where n == number of deopt points and
m ==
number of opt/deopts cycles per SharedFunctionInfo). For a small m--which
something V8 strives for anyway--your patch likely makes the complexity and
performance worse.
https://codereview.chromium.org/334713011/
--
--
v8-dev mailing list
[email protected]
http://groups.google.com/group/v8-dev
---
You received this message because you are subscribed to the Google Groups "v8-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
