tgl@a.sei.cmu.edu (Tom Lane) (01/21/86)
Apparently the new 800k drives are supposed to be twice as fast as the old drives. Does anybody know how Apple did this? I can think of two possibilities: either the disc actually spins twice as fast, or they have somehow doubled the throughput by simultaneously reading/writing on both heads. Neither of these thoughts is appealing. The first alternative is likely to break some copy-protection schemes (anything that's timing-sensitive could have trouble); the implication is that some copy-protected programs could not be loaded from an 800k drive. (Not to mention purely hardware considerations such as reduced error margins...) The second alternative would imply some rather tricky buffering schemes; the speedup would only exist for a few standard disc access patterns, and there would be NO speedup on old (single-sided) discs. And if they buffer writes, there is a risk of failing to complete the write operation... Can anyone on the net shed more light on this question? tom lane (ARPA: lane@a.cs.cmu.edu)
e-smith@utah-cs.UUCP (Eric L. Smith) (01/23/86)
In article <216@a.sei.cmu.edu> tgl@a.sei.cmu.edu.UUCP writes: >Apparently the new 800k drives are supposed to be twice as fast >as the old drives. Does anybody know how Apple did this? > ... > tom lane (ARPA: lane@a.cs.cmu.edu) I played with one for about 15 minutes last night, and it seemed to me that the data transfer rate is probably unchanged. I attribute the speedup to two factors: twice as much data available without stepping the head (i.e., improved seek latency), and improvements to the file system (it didn't appear to have to re-read the directory as often). Of course, this is only my opinion after playing with it for a short time... -- ------------------------------------------------------------------------------- Eric L. Smith (801) 581-8100 e-smith@utah-cs.arpa ...decvax!utah-cs!e-smith 3118 Merrill Engineering, University of Utah, Salt Lake City, UT 84112 The opinions expressed herein do not necessarily represent those of the University of Utah, my friends, enemies, computer, or even me. :-) At Shoplifter's there's no waiting in long lines at checkout counters, because we don't have checkout counters. We just follow you to your car and read you your rights. -- Firesign Theatre's "Eat Or Be Eaten"
bhyde@inmet.UUCP (01/23/86)
I presume the drives are faster by virtue of improved seek time. Most all that "grinding" noise your drives make is the heads screwing back and forth. If you peak into the slot, with a flash light, you can watch the light reflect off the front of the head. It is amazing, in most disk bound applications it never stops to read anything, it just hopelessly tries to saw the disk in half and put it out of it's misery. - ben hyde, cambridge
bart@reed.UUCP (Bart Massey) (01/24/86)
> Apparently the new 800k drives are supposed to be twice as fast > as the old drives. Does anybody know how Apple did this? > > I can think of two possibilities: either the disc actually spins > twice as fast, or they have somehow doubled the throughput by > simultaneously reading/writing on both heads. > ... > The second alternative would imply some rather tricky buffering > schemes; the speedup would only exist for a few standard disc > access patterns, and there would be NO speedup on old > (single-sided) discs. And if they buffer writes, there is a > risk of failing to complete the write operation... > > Can anyone on the net shed more light on this question? > > tom lane (ARPA: lane@a.cs.cmu.edu) I can't speak for the Mac directly, but on every other machine I've ever seen with dual heads, the sectors are on alternate sides, so that 0 and 1 are on opposite sides of the disk, then 2 and 3, then 4 and 5, etc. On the Mac floppy, I believe a logical block (the smallest unit readable by user-level software) is two sectors, so the speedup would occur for all disk access patterns, and buffering would be straightforward. If this is the source of the speedup, there would be none for old disks, but presumably one would only care about single-sided disks for compatibility reasons, not for day-to-day operation. Finally, no write operations should fail regardless of buffering, as the determination of the "sidedness" of the disk should occur at mount time... Bart Massey
north@apple.UUCP (Donald N. North) (01/25/86)
In article <216@a.sei.cmu.edu> tgl@a.sei.cmu.edu.UUCP writes: >Apparently the new 800k drives are supposed to be twice as fast >as the old drives. Does anybody know how Apple did this? > There are two reasons: 1) The 800k drives do their own speed control in their onboard micro. For the 400k drives, the 68k processor fills the infamous 'disk pwm' buffer with the appropriate values (determined at boot time) to set the speed of the disk (track dependent), then had to wait in a timing loop until it knew the speed was correct. This whole process took some time. 2) The 800k drives have a ready line to tell the 68k when they are ready to transfer data. In the 400k drive system, the 68k spends a lot of time in timing loops to determine when the drive will be ready to transfer data. The media format is the same between the 400k and 800k drives (except the 800k's use both sides, of course). Hope this answers your question. Don North Apple Computer, Inc. Advanced Development Group UUCP: {nsc,dual,idi,voder,ucbvax!mtxinu}!apple!north CSNET: north@apple.CSNET, north%apple@CSNET-RELAY
berry@tolerant.UUCP (David Berry) (01/28/86)
According to something I saw somewhere (delphi I believe) the 800K drives
appear faster because the drive speed and seeking, etc is all performed
by a processor in the drive rather than being performed by the Mac's 68K,
thus allowing more coprocessing to happen. In addition the drive can now
tell the 68K when it is up to speed and ready to transfer rather than the
68K having to watch the drive and figure out when it seems to be ready.
BTW, it isn't clear to me how well all this is going to work with copy
protection schemes which work by stuffing strange numbers in to the PWM
that controls disk rotation speed. Since this is no longer used (the
information is fed to the drive processor, which controls the rotation
speed) I suspect it may break them.
--
David W. Berry
{ucbvax,pyramid,idsvax,bene,oliveb}!tolerant!berry
I'm only here for the beer.smelser@wang.UUCP (pri=8 Craig Smelser) (02/04/86)
> According to something I saw somewhere (delphi I believe) the 800K drives > appear faster because the drive speed and seeking, etc is all performed > by a processor in the drive... > > BTW, it isn't clear to me how well all this is going to work with copy > protection schemes which work by stuffing strange numbers in to the PWM > that controls disk rotation speed. Since this is no longer used (the > information is fed to the drive processor, which controls the rotation > speed) I suspect it may break them. In the May Software Supplement, Apple warned developers that this would happen. You might think that six months (July to January) would be long enough for people like MicroSoft to issue updates, but I don't think they care. When I bought Multiplan in January, it was still version 1.02 with all its obnoxious copy-protection errors (trying to eject non-floppy disks, etc.) and packaged with *Finder 1.0*.
callen@ada-uts.UUCP (02/10/86)
wang!smelser writes: >In the May Software Supplement, Apple warned developers that this would >happen. You might think that six months (July to January) would be long >enough for people like MicroSoft to issue updates, but I don't think they >care. When I bought Multiplan in January, it was still version 1.02 >with all its obnoxious copy-protection errors (trying to eject non-floppy >disks, etc.) and packaged with *Finder 1.0*. Uh, pardon me, but Microsoft doesn't use disk speed games for copy protect on the WORD/FILE/MULTIPLAN/CHART programs - and if you haven't figured out how to get around their copy protection, you either haven't tried or you don't have a copy of Fedit :-) Copy protection DOES suck, though... Jerry Callen ...ihnp4!inmet!ada-uts!callen P.S. Don't send me mail asking me "how to do it". I won't answer. Purchase a Microsoft program and figure it out yourself (like I did).