Path: utzoo!utgpu!news-server.csri.toronto.edu!clyde.concordia.ca!uunet!wuarchive!zaphod.mps.ohio-state.edu!mips!prls!philabs!linus!nixbur!nixpbe!peun11!josef From: josef@nixpbe.UUCP (Moellers) Newsgroups: comp.os.cpm Subject: Re: How to speed up Ampro LB+ SCSI? Message-ID: Date: 10 Aug 90 07:21:54 GMT References: <2681@inews.intel.com> <8190004@hp-lsd.COS.HP.COM> <12835@mentor.cc.purdue.edu> Sender: news@nixpbe.UUCP Lines: 33 In <12835@mentor.cc.purdue.edu> wilker@descartes.math.purdue.edu (Clarence Wilkerson) writes: >Could you implement "scsi device to scsi device" transfer without having >to go through >the CPU? This is possible under some circumstances ( e.g. two disks on >same controller ), >but I'm not sure of the generality. From what I know about SCSI, I'd say it depends. (This is standard answer #75534) SCSI distinguishes between initiator and target. The initiator selects a target and then the target requests from the initiator whatever information is needed (command block, data, message) or sends to the initiator whatever information it holds (data, status, message). Usually, hosts are initiators and devices are targets. So, in order to do a "device to device" transfer, You'll have to have one device that can act as an initiator, communication with another device that continues to behave as a target. Some tape drives can do this. You just tell'em to read n blocks of data from target x and then leave it to do it's task. If You were to look at the SCSI bus, You'd see the tape drive selecting the disk, the drive requesting command blocks from the tape, then sending data to the tape, etc. Probably one or the other controller can do a disk-to-disk-copy locally, but that would be very controller specific. -- | Josef Moellers | c/o Nixdorf Computer AG | | USA: mollers.pad@nixbur.uucp | Abt. PXD-S14 | | !USA: mollers.pad@nixpbe.uucp | Heinz-Nixdorf-Ring | | Phone: (+49) 5251 104662 | D-4790 Paderborn |