Return to BSD News archive
Newsgroups: comp.unix.bsd.freebsd.misc Path: euryale.cc.adfa.oz.au!newshost.anu.edu.au!harbinger.cc.monash.edu.au!news.mira.net.au!inquo!vyzynz!news.dacom.co.kr!arclight.uoregon.edu!chi-news.cic.net!news.wctc.net!news.new-york.net!ritz.mordor.com!ritz From: ritz@ritz.mordor.com (Chris Mauritz) Subject: Re: ALR Pentium Pro and FreeBSD? X-Newsreader: TIN [UNIX 1.3 950824BETA PL0] Lines: 27 Organization: Mordor International Message-ID: <DquMLF.7LE@ritz.mordor.com> References: <4m5u4b$f1d@news.service.uci.edu> <DqpGnK.7DE@ritz.mordor.com> <4mat1c$qpu@shellx.best.com> <Dqsp97.J59@ritz.mordor.com> <4mdk01$7vr@shellx.best.com> Date: Fri, 3 May 1996 21:34:26 GMT Russell Carter (rcarter@shellx.best.com) wrote: : In article <Dqsp97.J59@ritz.mordor.com>, : Chris Mauritz <ritz@ritz.mordor.com> wrote: : >: >Unless Intel is shipping fixed orion chipsets you're going to find : >: >yourself limited to about 4mb/sec of cpu to memory bandwidth...yech. : >: ^^^^^^^^^^^^^ : >: PCI controller to memory : > : >Oops. However, it is effectively the same since the PCI controller : >sits between the cpu and non-cache memory, right? : : Oops, I was vague. PCI controller card. : : Memory bandwidth is affected quite a bit by the same chipset, but : is still about 40 MB/s for writes and over 100 MB/s for reads. Triton I : is about 80 and 80, for instance. The reason most benchmarks don't : detect the memory bandwidth slowdown is because they run primarily : out of cache, and the P6 second level cache is very very good. : : Not that any of this appears to have affected sales. NT is an : excellent method to slow down any hardware, and that's what P6s are being : sold for. Heh. That's like buying a Ferrari to pull a u-haul trailer. :-) Chris -- Christopher Mauritz | For info on internet access: ritz@mordor.com | finger/mail info@ritz.mordor.com OR Mordor International | http://www.mordor.com/ 201/212/718 internet access | Modem: (201)433-7343,(212)843-3451