wdw@aucs.UUCP (Bill Wilder) (01/26/88)
I notice some of the more recent high performance disk controllers use
the ESDI interface and support disk interleaving of 1:1. I understand
the improvements in transfer rate that result from a 1:1 interleave.
Are there any other benefits to be had from the ESDI interface?
Specifically, I have a non-ESDI disk controller that also supports 1:1
interleave due to double-buffering within the controller itself. Isn't
this controller going to deliver maximum performance or are there
drives available that run faster than 3600 R.P.M. or support higher
data densities that only ESDI interfaces can keep up with.
Just curious. Thanks for any info you can offer.
--
UUCP: {uunet|watmath|utai|garfield}!dalcs!aucs!wdw
BITNET: WDW@Acadia
Internet: WDW%Acadia.BITNET@CUNYVM.CUNY.EDUpaul@vixie.UUCP (Paul Vixie Esq) (01/28/88)
In article <788@aucs.UUCP> wdw@aucs.UUCP (Bill Wilder) writes: >I notice some of the more recent high performance disk controllers use >the ESDI interface and support disk interleaving of 1:1. I understand >the improvements in transfer rate that result from a 1:1 interleave. >Are there any other benefits to be had from the ESDI interface? ESDI disk configurations do not all have 1:1 interleave. Part of it depends on the processor speed, bus speed, etc, since somebody has to be able to "catch" data coming off the drive. If you use 1:1 and the rest of the system can't keep up, you wait 1/3600th of a second for the next sector to come around AGAIN when you're ready for it. If you use 2:1 and you only really needed 1.1:1, you wait 0.9*((1/3600)/34) second for the sector to come around, which is bad but not AS bad. 1:1 that you can keep up with is to be preferred, but it depends on more than the drive and controller. ESDI's principle benefits over the MFM/ST506 are: different encoding scheme (RLL 2,7 perhaps?) yielding 34x512 bytes/track instead of 17x512; hard- sectored drives are available which are faster and more reliable than soft- sectored drives for various reasons; the language spoken between the controller and drive allows for more than 1024 cylinders (many ST506-type controllers cannot deal with more than 1024 cylinders); there is a way for the controller to ask the drive to report its geometry -- so you don't need remember the (cylinders,heads,sectors) count in the driver or in the disk label. Also, because of the different encoding scheme used, the transfer rate between drive and controller is about 11 mbits/sec instead of ST506's 5 mbits/sec. Lastly, ESDI lets you have 7 or 8 disks on a controller instead of 4. Not a useful enhancement until someone comes up with a multi-threaded controller and implements it in ECL :-). -- Paul A Vixie Esq paul%vixie@uunet.uu.net {uunet,ptsfa,hoptoad}!vixie!paul San Francisco, (415) 647-7023
abh@POGO.CAMELOT.CS.CMU.EDU (Andrew Hastings) (01/29/88)
In article <794@vixie.UUCP>, paul@vixie.UUCP (Paul Vixie Esq) writes: > ... If you use 1:1 and the rest of the > system can't keep up, you wait 1/3600th of a second for the next sector to > come around AGAIN when you're ready for it. Disks usually rotate at 3600 rpm, so you'd wait 1/3600 minute, or 1/60 second.
karsh@nicmad.UUCP (Bruce Karsh) (01/29/88)
In article <794@vixie.UUCP> paul@vixie.UUCP (Paul Vixie Esq) writes: > >ESDI disk configurations do not all have 1:1 interleave. Part of it depends >on the processor speed, bus speed, etc, since somebody has to be able to >"catch" data coming off the drive. If you use 1:1 and the rest of the >system can't keep up, you wait 1/3600th of a second for the next sector to >come around AGAIN when you're ready for it. Wow! Where do you get disks that fast? :-). Seriously, I think that 1/3600 *minutes* is a typical duration for a disk rotation. Most hard disks spin at about 60 revs per second. -- Bruce Karsh {uwvax,ihnp4}!nicmad!karsh
paul@vixie.UUCP (Paul Vixie Esq) (01/30/88)
In article <747@PT.CS.CMU.EDU> abh@POGO.CAMELOT.CS.CMU.EDU (Andrew Hastings) writes: #>In article <794@vixie.UUCP>, paul@vixie.UUCP (Paul Vixie Esq) writes: #>> ... If you use 1:1 and the rest of the #>> system can't keep up, you wait 1/3600th of a second for the next sector to #>> come around AGAIN when you're ready for it. #> #> Disks usually rotate at 3600 rpm, so you'd wait 1/3600 minute, #>or 1/60 second. I knew that :-). Ooops. 1/3600th of a second would be pretty fast, and nothing to complain about... Sorry, folks. -- Paul A Vixie Esq paul%vixie@uunet.uu.net {uunet,ptsfa,hoptoad}!vixie!paul San Francisco, (415) 647-7023