Closed Bug 577884 Opened 15 years ago Closed 14 years ago

TM: inline UnboxInt32

Categories

(Core :: JavaScript Engine, defect)

Product:
Core
Component:
JavaScript Engine
Version:
Other Branch
Platform:
x86
macOS
Type:
defect
Priority:
Not set
Severity:
normal

Tracking

()

Status:
RESOLVED WONTFIX

People

(Reporter: gal, Unassigned)

Details

We have 1084161 calls to it. This would be a serious speedup I think.
I played around with this. I am not seeing the dramatic speedups I was hoping for. I will do more on this once fatval lands. Phi nodes will help too probably.
One possible problem is that diamonds currently interrupt CSE. You could try the patch in bug 542905. But yeah, phi nodes is ideal for this kind of thing.
The call overhead seems significant, so I was hoping just eliminating that for the predominant number of cases should help but it really didn't.
Did you look at the final generated LIR? It's possible the extra labels could be causing CSE to not happen, which might be undoing the improvements that inlining has caused. I have a script that automatically compares the generated LIR for all the SS benchmarks for exactly this kind of thing.
Obsolete with the removal of tracejit.
Status: NEW → RESOLVED
Closed: 14 years ago
Resolution: --- → WONTFIX
You need to log in before you can comment on or make changes to this bug.