RE: Poor thread performance on Linux vs. Solaris

> From: Bill Soudan [mailto:[email protected]]

> > A coworker of mine played with the notion of removing global lock
> > for the futexes, replacing it with a per-hash-queue lock. He did it
> > on our rtfutex patch, and I don't know how easy would it be for
> > the mainstream futexes to adapt, but it might be worth a try; it
> > will not resolve fully the problem, but it might help a lot.
> 
> Any chance I could get my hands on this code?  From the list archives, I
> found this link:
> 
> http://developer.osdl.org/dev/robustmutexes/
> 
> but the rtfutex code I found still has the global lock.  Am I looking in
> the wrong place, or is this unreleased code? 

He posted a patch to the list [that I still had not time to go
over] where we was moving the lock to the per-queue ... let me
dig the archive ... 
http://lists.osdl.org/pipermail/robustmutexes/2003-August/000011.html

I don't know if that would fully solve it [and still, the rtfutex
patch is kind of unstable]. Maybe you want to wait until Monday,
as he lives in China and they are full blown in weekend now--he
should have a chance to answer your message.

Iñaky Pérez-González -- Not speaking for Intel -- all opinions are my own (and my fault)