Date: Sat, 15 Aug 1998 20:23:15 +0000 (GMT) From: Terry Lambert <tlambert@primenet.com> To: mike@smith.net.au (Mike Smith) Cc: thorpej@nas.nasa.gov, hackers@FreeBSD.ORG Subject: Re: TESTERS WANTED for new ATAPI CD/CDR/CDRW driver. Message-ID: <199808152023.NAA20511@usr01.primenet.com> In-Reply-To: <199808151153.LAA01317@word.smith.net.au> from "Mike Smith" at Aug 15, 98 11:53:23 am
next in thread | previous in thread | raw e-mail | index | archive | help
> I'd be curious to know if anyone's measured the cost of the SCSI<->ATA > translation for ATA disks, since that's where performance is really an > issue. Nobody cares how efficient access to a Zip disk is, but an > extra 10% overhead on ATA disk transactions would really hurt. > > Command passthrough doesn't (can't) work directly, as many fields are > different sizes. I think what you mean is that there's no need for any > intelligence in the translation, which is fair enough. A question I have is this: If I'm passing SCSI commands down, is it possible to actually support multiple outstanding tagged commands? I know that the EIDE specification allows for this, but I am unaware of any IDE disk drives that support it. If I could get multiple outstanding transactions at a cost of 10% overhead, all I need is two of them and my system is running 80% faster than it was... For a single user workstation running a curses program on the console, this is not a problem. For any other load, this should generally be a win, especially if swapping is occurring. The inability of most IDE hardware to support multiple outstanding tagged commands is the primary reason I still recommend SCSI to people who are running light loads and are never going to need more than one disk and one CDROM (though "never" is a very long time indeed... for systems with most Intel chipsets, there is a limit to 2 simultaneous PCI bus masters before the negotiation fails, meaning I'm limited to 4 IDE devices no matter what). Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers. 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?199808152023.NAA20511>