Date: Thu, 30 Nov 2006 17:31:08 -0500 From: Kris Kennaway <kris@obsecurity.org> To: Miles Nordin <carton@Ivy.NET> Cc: freebsd-sparc64@freebsd.org Subject: Re: Netra T1 105 (sparc64) optimization Message-ID: <20061130223107.GA61593@xor.obsecurity.org> In-Reply-To: <oqvekw609g.fsf@castrovalva.Ivy.NET> References: <755cb9fc0611290724q127f006va84f3457c48443b6@mail.gmail.com> <oq8xht7biy.fsf@castrovalva.Ivy.NET> <456E8D8B.8010102@orel.ru> <755cb9fc0611300454s4d28ad14rd402cba8388d49@mail.gmail.com> <14989d6e0611300700n3ac073bayc4584f2b1020ee61@mail.gmail.com> <755cb9fc0611300839y6deed1ceqf452018cc2f73737@mail.gmail.com> <oqzma869j9.fsf@castrovalva.Ivy.NET> <755cb9fc0611301208w1ff93987m88885b3b59444373@mail.gmail.com> <oqvekw609g.fsf@castrovalva.Ivy.NET>
next in thread | previous in thread | raw e-mail | index | archive | help
--6c2NcOVqGQ03X4Wi Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Content-Transfer-Encoding: quoted-printable On Thu, Nov 30, 2006 at 05:24:43PM -0500, Miles Nordin wrote: > >>>>> "av" =3D=3D Alexandre Vieira <nullpt@gmail.com> writes: >=20 > av> Without polling it transfers like 8MB/s, with polling it does > av> less than this and there is twice the latency. >=20 > that's not what i expected on both counts. >=20 > but the supposed benefit of polling is increase in pps capacity, not > reduced latency. sparc64 doesn't do preemption, so interrupt latency is longer than it should be. Presumably that is why polling helps so much. Kris --6c2NcOVqGQ03X4Wi Content-Type: application/pgp-signature Content-Disposition: inline -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.5 (FreeBSD) iD8DBQFFb1urWry0BWjoQKURAmh8AKCbh5NUeUZI0spuY5CGYNqVxVzgUQCgsXSL kiqIvDS9HdPCPoM7PScb5mQ= =RQCs -----END PGP SIGNATURE----- --6c2NcOVqGQ03X4Wi--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20061130223107.GA61593>