Date: Tue, 10 Nov 2009 13:23:35 -0700 (MST) From: "M. Warner Losh" <imp@bsdimp.com> To: rpaulo@freebsd.org Cc: freebsd-arm@freebsd.org Subject: Re: SCHED_ULE? Message-ID: <20091110.132335.1735450040.imp@bsdimp.com> In-Reply-To: <43EE151E-574E-4A17-9946-0BC5A6B3BC69@freebsd.org> References: <20091109102704.GA75988@ci0.org> <4AF8EE61.8060502@errno.com> <43EE151E-574E-4A17-9946-0BC5A6B3BC69@freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
In message: <43EE151E-574E-4A17-9946-0BC5A6B3BC69@freebsd.org>
Rui Paulo <rpaulo@freebsd.org> writes:
: Hi,
:
: On 10 Nov 2009, at 04:38, Sam Leffler wrote:
:
: > Olivier Houchard wrote:
: >> On Sun, Nov 08, 2009 at 01:30:48PM +0000, Rui Paulo wrote:
: >>> Hi,
: >>> I guess this has been discussed in the past but, can't we turn on ULE on ARM embedded systems ? What's the bottleneck or performance regression (assuming there's one) ?
: >>>
: >>> --
: >>> Rui Paulo
: >> Hi,
: >> At one point ULE was buggy on arm, but I think it's been fixed like years ago,
: >> so it should be safe to use it.
: >
: > Last I measured it was slower than 4BSD on my xscale boards (not much but measurable). This was mostly doing network packet pushing (wired+wireless).
:
: I also tested this on the Cambria board and I concluded the same (I used sysbench).
This isn't too surprising, since it is optimized for multicore and
4BSD grew up on single-core systems...
Warner
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20091110.132335.1735450040.imp>