Re: 2.2.1 SMP Scalability to 4-way

urbanski@us.ibm.com
Sun, 31 Jan 1999 09:52:29 -0700

The applications are Blast and hmmer from http://blast.wustl.edu and
http://hmmer.wustl.edu. Blast 2.0 is the first application I was talking
about, but unfortulately isn't provided as open-source so there's probably
not much that can be done. Hmmer is though. Solaris is currently runnning
Hmmer on the same hardware about 100% faster than Linux - and I'm not sure
why. Hammer does compile with pthreads enabled. I can provide you with
the model files I am using if you are interested in doing the comparison
yourself.

Jay Urbanski
Netfinity Systems Engineer
IBM Advanced Technical Support
MCSE, PSE, Certified Solaris Systems Administrator
(817)962-3597 TL 522-3597
(817)962-7307 fax
(800)413-9093 pager
urbanski@us.ibm.com

alan@lxorguk.ukuu.org.uk (Alan Cox) on 01/31/99 11:39:17 AM

To: Jay Urbanski/Dallas/IBM
cc:
Subject: Re: 2.2.1 SMP Scalability to 4-way

> assumed this application operated similarly on both operating systems,
but
> it does not. On Solaris it spawns a single multi-threaded process which
(I
> assume) uses the correct number of threads. On Linux, it spawns multiple
> instances of a process - but uses the wrong number! If given the
parameter
> P=2, which is supposed to mean 2 CPU's it spawns 3 processes. So at P=4
it
> was spawning 5 processes. Consider this a retraction.

I'd be very interested to see how it compared with that fixed and both
threaded / both forking.

Alan

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/