mdr@reed.UUCP (Mike Rutenberg) (04/18/89)
Does anyone know anything about the IPI (intelligent peripheral interface)? I understand it to be similar to a much faster SCSI. I believe it is used by BiiN and some of the superworkstation makers. Are there chips for it? Mike -- Mike Rutenberg Reed College, Portland Oregon (503)239-4434 (home) BITNET: mdr@reed.bitnet UUCP: uunet!tektronix!reed!mdr Note: The preceding remarks are intended to represent no known organization
Anonymous@hpcupt1.HP.COM (Anonymous Author) (04/19/89)
A good explanation is in the April 13, 1989 issue of Electronic Design, pp 129-132. IPI is a four-layer ANSI interface. You will see IPI-1 (state machine and procedures for device selection), IPI-2 (device-level interface), and IPI-3 (supplies the device-generic I/O bus interface). IPI-0 is a 16-bit parallel bus running at 5 MHz implemented in a variety of ways. Joe Martinka Hewlett-Packard
Anonymous@hpcupt1.HP.COM (Anonymous Author) (04/19/89)
/ hpcupt1:comp.arch / Anonymous@hpcupt1.HP.COM (Anonymous Author) / 8:23 am Apr 19, 1989 / A good explanation is in the April 13, 1989 issue of Electronic Design, pp 129-132. IPI is a four-layer ANSI interface. You will see IPI-1 (state machine and procedures for device selection), IPI-2 (device-level interface), and IPI-3 (supplies the device-generic I/O bus interface). IPI-0 is a 16-bit parallel bus running at 5 MHz implemented in a variety of ways. Joe Martinka Hewlett-Packard ----------
gnb@melba.bby.oz (Gregory N. Bond) (04/26/89)
The new Sun servers (SPARCservers?) use IPI disks for the high end
boxes, according to the product launch here yesterday. This
would
indicate that they are faster even than SMD disks on the VME bus.
How does Sun do the IPI host adapter?
Greg.--
Gregory Bond, Burdett Buckeridge & Young Ltd, Melbourne, Australia
Internet: gnb@melba.bby.oz.au non-MX: gnb%melba.bby.oz@uunet.uu.net
Uucp: {uunet,mnetor,pyramid,ubc-vision,ukc,mcvax,...}!munnari!melba.bby.oz!gnbaglew@mcdurb.Urbana.Gould.COM (04/29/89)
>Does anyone know anything about the IPI (intelligent peripheral interface)? >I understand it to be similar to a much faster SCSI. >I believe it is used by BiiN and some of the superworkstation makers. > >Are there chips for it? > >Mike Gould CSD went with IPI for its NPL line, expecting IPI to quickly become popular. Unfortunately for Gould, SCSI came along, and peripheral manufacturers chose to do SCSI first. For a long time Gould was just about the only computer vendor that had IPI as the main I/O bus. I think that Gould's disk controllers may still be the fastest IPI controllers around (I don't think that it was Gould's intention to get into the business of making them, but they had to). Lately, SCSI has been pooping out, and IPI has been picking up steam again.
prc@maxim.ERBE.SE (Robert Claeson) (05/02/89)
In article <28200303@mcdurb>, aglew@mcdurb.Urbana.Gould.COM writes: > Lately, SCSI has been pooping out, and IPI has been picking up steam > again. How well does IPI perform when compared to synchronous SCSI? -- Robert Claeson, ERBE DATA AB, P.O. Box 77, S-175 22 Jarfalla, Sweden Tel: +46 (0)758-202 50 Fax: +46 (0)758-197 20 EUnet: rclaeson@ERBE.SE uucp: {uunet,enea}!erbe.se!rclaeson ARPAnet: rclaeson%ERBE.SE@uunet.UU.NET BITNET: rclaeson@ERBE.SE
markb@denali.SGI.COM (Mark Bradley) (05/04/89)
In article <648@maxim.ERBE.SE>, prc@maxim.ERBE.SE (Robert Claeson) writes: > > How well does IPI perform when compared to synchronous SCSI? > -- > Robert Claeson, ERBE DATA AB, P.O. Box 77, S-175 22 Jarfalla, Sweden > Tel: +46 (0)758-202 50 Fax: +46 (0)758-197 20 > EUnet: rclaeson@ERBE.SE uucp: {uunet,enea}!erbe.se!rclaeson > ARPAnet: rclaeson%ERBE.SE@uunet.UU.NET BITNET: rclaeson@ERBE.SE If you look at specs alone, IPI 2 is faster than synchronous SCSI. The available devices are faster, too. IPI level 2 theoretical xfer rate ~12 MB/sec. Devices announced 2.4-6.0 MB/sec. This compares favorably with SCSI's 4.0 MB/sec. upper limit. Unless you want to go to the cumbersome SCSI 2 hardware, but then where do you get the drives. Then there is striping, arrays, parallel xfer technology, etc. If you do things right you can obtain some pretty impressive performance out of either. markb -- Mark Bradley "Faster, faster, until the thrill of IO Subsystems speed overcomes the fear of death." Silicon Graphics Computer Systems Mountain View, CA ---Hunter S. Thompson
aglew@mcdurb.Urbana.Gould.COM (05/04/89)
>> Lately, SCSI has been pooping out, and IPI has been picking up steam >> again. > >How well does IPI perform when compared to synchronous SCSI? I have no data of my own to give, so I can't say. And I'm not sure that pasting together the tables in the magazines is appropriate. However, as I understand it, IPI is a bit more intelligent, allows more useful channel programs, than SCSI - even SCSI-2. But I haven't checked out SCSI-2 much.
jwest@pitstop.West.Sun.COM (Jeremy West) (05/04/89)
In article <128@melba.oz>, gnb@melba.bby.oz (Gregory N. Bond) writes: > > > The new Sun servers (SPARCservers?) use IPI disks for the high end > boxes, according to the product launch here yesterday. This would > indicate that they are faster even than SMD disks on the VME bus. > > How does Sun do the IPI host adapter? > Sun has its own design of IPI controller on a 9U VME card. It has an on-board 68020 and 1 Mb of read-ahead cache and has some good optimisation algorithms. It can take faster disks than are currently available and does fast burst accesses on the VME bus. Throughput is about twice as good as SMD, cabling is simpler, up to 8 drives per controller and 4 controllers per system with 1 Gb (formatted) drives to give max of 32 Gb on SPARCserver390 systems. Price is similar to SMD. Adrian Cockcroft Sun Cambridge UK TSE sun!sunuk!acockcroft (Borrowing Jerry West's account at Mountain View to get at USENET)