From: Godmar Back (gback@cs.utah.edu)
Date: Wed Jan 13 1999 - 15:00:51 EST
>
> Note that GCJ stores the lock pointer in the object (just after the dtable
> pointer). Okay this increases the size of each object a little but saves on that
> hash lookup. I was going to do this to integrate GCJ at some point anyhow,
> though I could modify GCJ to not do this if preferred.
>
Well, Kaffe still does have the USE_LOCK_CACHE you put in.
It's disabled by default, so it might be rusty, but it's there.
Pat has gotten good speedups in Alta by using USE_LOCK_CACHE,
at the cost of using more memory (he doesn't time gc...)
This not only solves the problem of finding the lock when
unlocking it, but it also helps in finding the lock when locking the
object. It does not make it a constant time, as the lock stored in
the object may be stale.
Without having numbers in front of me, I would kind of like to see it
optional in both gcj and kaffe. Let's ask the gcj mailing list what
they have to say.
- Godmar
This archive was generated by hypermail 2b29 : Sat Sep 23 2000 - 19:57:41 EDT