"Kenneth D. Merry" wrote:
>
> Wes Peters wrote...
> > Bill Paul wrote:
> > >
> > > Of all the gin joints in all the towns in all the world, Charles Randall
> > > had to walk into mine and say:
> > >
> > > > Bill Paul has developed a driver for the Alteon Tigon 1 and 2 cards.
> > > >
> > > > http://www.freebsd.org/~wpaul/Alteon/
> > > >
> > > > FYI,
> > > > Charles
> > > >
> > > > -----Original Message-----
> > > > From: David Miller [mailto:[EMAIL PROTECTED]]
> > > > Sent: Wednesday, August 18, 1999 1:55 PM
> > > > To: [EMAIL PROTECTED]
> > > > Subject: Gigabit ethernet support?
> > > >
> > > >
> > > > Any supported cards in 3.2.x? The HCL pages don't list any:(
> > >
> > > The ti driver supports several cards, including the Alteon AceNIC,
> > > the 3Com 3c985-SX, the Netgear GA620, the DEC EtherWORKS 1000, the
> > > SGI PCI gigabit ethernet card, the NEC gigabit ethernet card and
> > > possibly some from IBM as well, though I don't know the PCI vendor/device
> > > IDs for those so I can't be sure (if you find them out, you can try
> > > hacking them into the driver). All of these are supported by the same
> > > driver because they're all OEMed from Alteon.
> >
> > We have two of the NetGear GA620's here, and they work quite nicely. I use
> > them for testing throughput via Gig-E on our switches. Mine is running in
> > a lowly PII/233, on a 32-bit x 33 Mhz slot, and can push bits at 320 Mbps.
> > The GA620 will work in any 32 or 64 bit, 33 or 66 Mhz slot. A 64x66 slot
> > would probably speed things up appreciably.
>
> I doubt a faster PCI interface would really speed things up. My guess is
> that you've got some other bottleneck other than PCI bandwidth. Is the CPU
> pegged on either end?
Not on mine, it's running about 45%. The other end is much faster, a PII/400,
and is just discarding the packets, so it's not sweating.
> I would recommend that you make sure you've got a couple of things tweaked,
> they may increase your performance somewhat:
>
> - set your MTU to 9000, unless of course you're going through a switch that
> can't handle it
Not yet, that's part of what I will be developing. ;^) Due to architectural
limitations, we may not be able to support jumbo frames larger than 8k.
> - turn on net.inet.tcp.rfc1323, it enables support for TCP windows larger
> than 64K
OK.
> - increase net.inet.tcp.sendspace and net.inet.tcp.recvspace to 256K.
> You'll have to edit src/sys/socketvar.h and increase SB_MAX. From what
> I've seen (this may not be quite correct, but it's close enough) SB_MAX
> has to be double whatever you want to set sendspace and recvspace to.
> This has the effect of changing the TCP window size to 256K, I think.
> From what I've seen, increasing it to 512K is counterproductive unless
> you've got a card with 1MB of SRAM on board. (The Netgear boards have
> 512K.)
OK.
> And finally, netperf seems to work reasonably well for testing performance:
>
> http://www.netperf.org
Cool. I've been using several tools, since we do our real performance testing
with a SmartBits. I use spray to generate UDP traffic and a hacked-up version
of tcpblast for tcp traffic. I'll clean it up and re-release it one of these
days.
> > They're relatively cheap, too, going for $339 at DataComm Warehouse.
>
> FWIW, NECX has them for $310.
Cool. We don't have an account there; if I get one from DataComm all I
have to do is give them the account number and it shows up on my desk the
next morning. It's MORE convenient than going to the store. ;^)
--
"Where am I, and what am I doing in this handbasket?"
Wes Peters Softweyr LLC
http://softweyr.com/ [EMAIL PROTECTED]
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-hackers" in the body of the message
