Date: Wed, 25 Feb 2009 20:33:37 +0800 From: Siquijor Philips <siquijorphilips@gmail.com> To: ivoras@freebsd.org Cc: freebsd-net@freebsd.org Subject: Re: Questions on processing smaller frame size Message-ID: <a27b90e40902250433x6d83e4b1q3165d551c2647c5c@mail.gmail.com> In-Reply-To: <a27b90e40902250239x414d3fa7l12d291d278162377@mail.gmail.com> References: <a27b90e40902242314w12c15fddma43e1cd5afec8938@mail.gmail.com> <20090225075310.GA85904@svzserv.kemerovo.su> <a27b90e40902250239x414d3fa7l12d291d278162377@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Hello Ivan, This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enig69E41D4C44B97AD296C94242 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: quoted-printable Siquijor Philips wrote: > Hello Eugene, >=20 >> Traffic bandwidth does not matter (or much less), PPS rate matters. >> Packets drop due to high pps rate. Higher packet size, lesser pps >> saturates link and pps just can't grow high. It can with smaller packe= ts. >> >=20 > All the test scenarios here are bombarded with 1-Gig of network > traffic. When packet drops due to high pps rate, meaning to say that > the current FreeBSD system can't still handle this kind of situation > with high packet rate?=20 > Not unlikely. See other similar findings by other users, usually also > with em cards. Ok, let me check. > Or just it depends on your hardware? I just > can't imagine that with 2x quad-core system processing on high packet > rate, average CPU utilization consumes a total of 98%. >Total across all CPUs? With 64-byte and 128-byte frame, the total average CPU utilization will vary from 92-98%. Below is one of the top output. CPU # 6,7,5,4 and 0 are in 0% idle state already. By default, Chelsio NIC were using MSI/MSI-X interrupt on multiple RX/TX queues. # top -S PID USERNAME THR PRI NICE SIZE RES STATE C TIME WCPU COMMAND 338 root 1 -68 - 0K 16K CPU7 1 639:53 98.93% irq262: cxgbc 340 root 1 -68 - 0K 16K CPU4 0 631:19 98.19% irq264: cxgbc 337 root 1 -68 - 0K 16K CPU6 3 642:28 98.10% irq261: cxgbc 339 root 1 -68 - 0K 16K RUN 1 616:07 96.63% irq263: cxgbc 336 root 1 -68 - 0K 16K CPU3 2 621:11 90.33% irq260: cxgbc 335 root 1 -68 - 0K 16K CPU0 2 633:18 89.50% irq259: cxgbc 334 root 1 -68 - 0K 16K CPU1 3 642:27 88.87% irq258: cxgbc 333 root 1 -68 - 0K 16K CPU5 1 648:13 88.57% irq257: cxgbc 341 root 1 -83 - 0K 16K RUN 0 157:14 13.53% cxgbsp 16 root 1 171 ki31 0K 16K RUN 1 484:55 8.59% idle: cpu1 15 root 1 171 ki31 0K 16K RUN 2 483:39 7.76% idle: cpu2 14 root 1 171 ki31 0K 16K RUN 3 490:02 7.37% idle: cpu3 11 root 1 171 ki31 0K 16K RUN 6 485:50 0.00% idle: cpu6 10 root 1 171 ki31 0K 16K RUN 7 484:51 0.00% idle: cpu7 12 root 1 171 ki31 0K 16K RUN 5 475:38 0.00% idle: cpu5 13 root 1 171 ki31 0K 16K RUN 4 412:14 0.00% idle: cpu4 17 root 1 171 ki31 0K 16K RUN 0 409:06 0.00% idle: cpu0 342 root 1 -83 - 0K 16K RUN 4 155:04 0.00% cxgbsp > Try reducing the number of CPUs, it might help by reducing contention. Ok, I'll try. Regards, Siquijor --------------enig69E41D4C44B97AD296C94242 Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.6 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFJpSIBldnAQVacBcgRAvGvAJ99YslOqGaklehf6uQjLrAEm/hJ6gCgyTUd rf3LDBNfsymm+jxbN0WHyU0= =mjmo -----END PGP SIGNATURE----- --------------enig69E41D4C44B97AD296C94242--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?a27b90e40902250433x6d83e4b1q3165d551c2647c5c>