Date: Fri, 25 Apr 1997 11:53:12 -0700 From: Paul Traina <pst@jnx.com> To: Jim Shankland <jas@flyingfox.com> Cc: dg@root.com, cmsedore@mailbox.syr.edu, dennis@etinc.com, freebsd-hackers@freebsd.org, isp@freebsd.org Subject: Re: Router statistics Message-ID: <199704251853.LAA02132@base.jnx.com> In-Reply-To: Your message of "Thu, 24 Apr 1997 22:19:21 PDT." <199704250519.WAA17037@biggusdiskus.flyingfox.com>
next in thread | previous in thread | raw e-mail | index | archive | help
There's a lot less cruft needed to flow packets out a fxp. From: Jim Shankland <jas@flyingfox.com> Subject: Re: Router statistics Paul Traina <pst@jnx.com> writes: > > dg@root.com (David Greenman) writes: > > > > Wcarchive does an average of 3500 pps with a peak of around > > 5000 pps. The average data rate is around 20-25Mbps, with the > > machine around 50% idle. This is using the Intel PCI > > Pro/100B...reduce the idle time to about 30% if you're using a > > DEC/de card. > > Yep, the fxp driver is /much/ more efficient than the de > driver. Oh well, fxp's are cheaper too. :-) (Sigh) Is this an attribute of the driver, or of the respective chips? I've sort of settled on 21140-based cards, and I'd hate to buy Intel stuff, as it only encourages them to take over more of the world than they already have. But for some applications, I really need high throughput and efficiency. And the card *is* cheaper. Does there have to be a switch to the fxp driver in my future? All right, I'm done whining for now. Jim Shankland Flying Fox Computer Systems, Inc.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199704251853.LAA02132>