Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 18 Sep 2009 15:11:11 -0700
From:      Steve Kargl <sgk@troutmask.apl.washington.edu>
To:        "b. f." <bf1783@googlemail.com>
Cc:        jeff@FreeBSD.org, freebsd-amd64@FreeBSD.org
Subject:   Re: new to amd64
Message-ID:  <20090918221111.GA5821@troutmask.apl.washington.edu>
In-Reply-To: <d873d5be0909181449h55c87dfdp280d1fd3a7d5749a@mail.gmail.com>
References:  <d873d5be0909181449h55c87dfdp280d1fd3a7d5749a@mail.gmail.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On Fri, Sep 18, 2009 at 09:49:57PM +0000, b. f. wrote:
> Steve Kargl wrote:
> >
> >If you are running any floating point intensive applications
> >and these applications are multithreaded, you may want to use
> >the 4BSD scheduler rathar than ULE.  The last time I tested
> >ULE with an MPI application, it display miserable performance
> >on a dual, quad-core opteron system.
> 
> Have you made any attempt to analyze this problem, or discussed it
> with the primary author of ULE?

yes and yes.

> If so, to what do you attribute the poor performance, and do you
> have any ideas on how to improve ULE performance in this context?

The only acceptable solution I found was to switch to 4BSD.

-- 
Steve



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20090918221111.GA5821>