Return to BSD News archive
Path: euryale.cc.adfa.oz.au!newshost.anu.edu.au!harbinger.cc.monash.edu.au!news.mel.connect.com.au!munnari.OZ.AU!ihnp4.ucsd.edu!qualcomm.com!qualcomm.com!not-for-mail From: drew@qualcomm.com (Drew Eckhardt) Newsgroups: comp.unix.bsd.freebsd.misc,comp.periphs.scsi Subject: Re: SCSI-2 Bus Hangs Date: 26 Dec 1995 15:21:50 -0700 Organization: QUALCOMM, Incorporated; San Diego, CA, USA Lines: 17 Message-ID: <4bpshu$asf@qualcomm.com> References: <4bju5p$ihj@wilma.widomaker.com> <4bnidn$lej@almond.elite.net> NNTP-Posting-Host: redcloud.qualcomm.com Keywords: SCSI-2 Lockup Xref: euryale.cc.adfa.oz.au comp.unix.bsd.freebsd.misc:11472 comp.periphs.scsi:43840 In article <4bnidn$lej@almond.elite.net>, Nate Lawson <nate@elite.net> wrote: >Also, the NCR chipset is absolute junk. NCR 53c8xx family chips aren't affected as much as recent Adaptec products by sub-optimal cabling, handle error-free I/O operations with no host CPU intervention, have minimal per-SCSI command overhead, no arbitrary limits on the number of scatter/gather segments, and end up being quite fast. Under FreeBSD, a 486-66 with modern 7200RPM drive can sustain over 6500K/sec through a FFS filesystem. Aggregate transfer rates in excess of 20M/sec through the filesystem have been sustained with multiple NCR chips and spindles. The NCR design also requires almost no off-chip support hardware, AND results in a very cost-effective (I've seen '810 boards for $53, and '825 boards for $110) solution. In summary, I'd say the hardware is excellent. -- Four boxes : soap, ballot, jury, ammo. | Work: drew@Qualcomm.COM Use in that order. | Play: drew@PoohSticks.ORG