Date: Fri, 25 Jan 2002 22:58:34 -0800 From: Kent Stewart <kstewart@owt.com> To: Neil McGann <neil@neilmcgann.co.uk> Cc: freebsd-questions@FreeBSD.ORG Subject: Re: UDMA and 100Mbit NIC speed issues Message-ID: <3C52539A.3090704@owt.com> References: <5.1.0.14.0.20020125204321.00a0f4c0@pop.ntlworld.com> <3C51D185.2090600@owt.com> <5.1.0.14.0.20020125235403.00a1f930@pop.ntlworld.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Neil McGann wrote: > At 15:32 25/01/02 -0800, you wrote: > >> Kent Stewart wrote: >> >> I am curious what kind of setup you have. Since we are not seeing the >> same transfer rate between FreeBSD machines, I have to assume you have >> something configured differently than I do. An ifconfig on coral shows >> >> sis0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> mtu 1500 >> inet 192.168.0.17 netmask 0xffffff00 broadcast 192.168.0.255 >> ether 00:07:95:0b:0c:31 >> media: Ethernet autoselect (100baseTX <full-duplex>) >> status: active >> >> This is basically a default setup. The only difference on opal is the >> internal ether address and IP. >> >> I have all of my systems hooked up to 10/100 baseT switches and are >> all running full-duplex. They are on different floors. I have nomad >> and opal on the same switch and coral is in the basement and on the >> second switch. Nomad uses an Intel Pro 10/100+ and coral and opal are >> using the SiS-900 builtin NIC, which was just added to the 4.5 support. > > > ifconfig shows identical parameters to yours, except I have inet6 > addresses too (it's an out-of-the-box stock config, I haven't really > tweaked it much yet). > > Does hw.atamodes show you definitely dma? On coral hw.atamodes: dma,---,pio,---,dma,---,dma,---, on opal hw.atamodes: dma,dma,dma,dma, This also covers the arrangement. Coral has each HD on it's own controler. The pio mode is the TDK CD-R/RW writer. The last two are on a Promise Ultra-100. Opal has 3 HDs and a UDMA-66 DVD on the 2 standard controllers on the SiS-735 based mb. FWIW, opal will do a buildworld in 957.195u 263.416s 21:11.71 95.9% 1366+1576k 11744+3435io 3031pf+0w and coral does it in 960.366u 264.268s 22:09.61 92.1% 1360+1563k 44724+3406io 3013pf+0w Each of them have /usr/obj and /usr/src on different partitions on differen HDs from / & /usr. The fact that opal will finish almost a minute faster was kind of surprising. I had also tested doing buildworlds with a no -j to -j6, in steps of 2, and the no -j was the fastest wall clock time on both systems and -j2 was the slowest. Around -j4 the wall clock time started increasing. I had 3C905 on one and an Intel Pro 10/100+ on the other and when Bill Paul made the SiS-900 work, it was fastest on both machines. I was building ports on the two AMD 1600's and moving the distfiles and the packages to the slower machines. Before the SiS-900 worked the larger memory cache on the Intel always seemed to be 1-2MB/s faster than the 3C905xs. -- Kent Stewart Richland, WA mailto:kbstew99@hotmail.com http://users.owt.com/kstewart/index.html FreeBSD News http://daily.daemonnews.org/ To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-questions" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3C52539A.3090704>