Date: Wed, 11 Nov 1998 23:34:22 +0100 From: Martin Cracauer <cracauer@cons.org> To: Matthew Dillon <dillon@apollo.backplane.com> Cc: =?iso-8859-1?Q?S=F8ren_Schmidt?= <sos@freebsd.dk>, grog@lemis.com, hackers@FreeBSD.ORG Subject: Re: SCSI vs. DMA33.. Message-ID: <19981111233422.A10862@cons.org> In-Reply-To: <199811111932.LAA22898@apollo.backplane.com>; from Matthew Dillon on Wed, Nov 11, 1998 at 11:32:15AM -0800 References: <19981111164136.A29513@cons.org> <199811111730.SAA01133@freebsd.dk> <19981111184459.A10448@cons.org> <199811111932.LAA22898@apollo.backplane.com>
next in thread | previous in thread | raw e-mail | index | archive | help
In <199811111932.LAA22898@apollo.backplane.com>, Matthew Dillon wrote: > You guys are all doing something wrong. First, everyone is using > tiny dd's to test. I'm testing with bonnie inside a filesystem, using a RAM-limited kernel (i.e. 15 MB RAM for 50 MB files). Bonnie's seek test is somewhat interesting even for those multiple-process accesses. As long as you have really independent busses for your IDE drives, IDE may got quite fast. And BTW, if you use benchmarks writing to /dev/null for inter-system comarisions you'll quickly notice that this isn't the end of the story. It is not new for me that one UW SCSI channel has enough bandwidth for the task at hand. However, 4 SCSI drives and 1 UW SCSI controller and still much more expensive than 1 or 2 Promise IDE cards (depends on whether your mainboard has UIDE) and 4 IDE drives. It's bad that the Promise controllers don't work as they should and don't allow simultaneous access to both channels. But for now I am not sure that's god given and maybe there's a workaround and that's why we're discussing these things here. [benchmarks deleted] > This is one of several reasons why you use SCSI if you need performance. > You can't beat it for linear performance, and you can beat it for > transactional performance either. If you have just one drive then > sure, IDE will have similar linear performance. If you have two drives > you may even see similar linear performance reading, but I would guess > that you wouldn't see it writing. My tests so far see the IDE ccd writing as fast at it reads. The numbers aren't impressive since I can't get past the 2-channel limit for now. > If you have several drives, > SCSI beats the shit out of IDE reading and writing. As long as you have one bus per drive, I am not convinced that IDE must be worse here. > For non-linear performance under load, the one-drive case will again > yield similar results SCSI vs IDE. But the moment you start trying > to random read two IDE devices on a single IDE controller, SCSI instantly > scales to the two drives while the two IDE devices will not perform much > better then one. After that, IDE goes into the shitpile for > non-linear performance tests due to its command serialization. A SCSI > bus with 4 drives on it will scale 4x for non-linear performance. Two > IDE busses with 4 drives total will only scale 2x for non-linear > performance. A SCSI bus with 8 drives on it will scale 8x for non-linear > performance. You didn't read what I wrote. I am aware of these issue and have 4 drives, 4 channels on two cards. That it doesn't work because the two channels on each Promise card block each other is a different matter that was not foreseeable. Martin -- %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% Martin Cracauer <cracauer@cons.org> http://www.cons.org/cracauer Tel.: (private) +4940 5221829 Fax.: (private) +4940 5228536 Paper: (private) Waldstrasse 200, 22846 Norderstedt, Germany To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?19981111233422.A10862>