labc-3dc@e260-3g.berkeley.edu (Andy McFadden) (04/14/89)
In article <2819@ncsuvx.ncsu.edu> mikes@ncsuvx.ncsu.edu (Michael Steele) writes: [ whack ] > I was >wondering whether I should get a 28ms drive or a 39ms drive? Is the addition >$100 going to be worth the access time? I think not, but was wanting some >more input. A recent issue of Call-A.P.P.L.E. had a hard drive comparison... among those reviewed were two new Sider drives and the InnerDrive. When it came to loading AppleWorks GS, the InnerDrive performed as well or better than the other drives. Why do I mention this? The InnerDrive is rated at 65ms access time; the two Sider drives were 19ms and 20ms. Obviously it didn't perform as well in a seek test, but the point is that access time doesn't mean much in terms of overall performance (kinda like MHz...) BTW, access time is the time it takes to draw the drive head across 1/3 of the total tracks. >Michael Steele mikes@ncsuvx.ncsu.edu -- fadden@cory.berkeley.edu (Andy McFadden) ...!ucbvax!cory!fadden labc-3dc@widow.berkeley.edu
jm7e+@ANDREW.CMU.EDU ("Jeremy G. Mereness") (04/14/89)
> *Excerpts from ext.in.info-apple: 14-Apr-89 Hard drive speed Andy* > *McFadden@labrea.sta (1069)* > A recent issue of Call-A.P.P.L.E. had a hard drive comparison... among those > reviewed were two new Sider drives and the InnerDrive. > When it came to loading AppleWorks GS, the InnerDrive performed as well or > better than the other drives. Why do I mention this? The InnerDrive is rated > at 65ms access time; the two Sider drives were 19ms and 20ms. Obviously it > didn't perform as well in a seek test, but the point is that access time If someone could do a confirm on this, I would appreciate it. I believe the Inner Drive is faster because it includes a RAM based driver. A+, I believe, did a few comparison tests for drives with and without their RAM drivers. The bottleneck of the Apple ][ line is I/O. All reads and writes to the slots and to the video are made at 1 MHz. The Inner Drive gets around this by basing its drivers in RAM instead of on its interface card, thus the drive can figure out what to do with itself 2.5 times faster than a normal hard drive. Consequently, from the benchmarks I have seen in Call-Apple and A+, the Inner Drive is always a winner. Again, if someone more knowledgeable of the Apple innards could confirm this, I would appreciate it. In fact, I have always been curious about just what's going on with the 1MHz slowdown, and if there is ANY way to get around this. I have seen AE produce impossibilities like the PC Transporter and the Transwarp GS. If there was SOME way to reconcile the i/o bottleneck (I believe that anything's possible) without ruining Apple ][ and peripheral compatibility, I think the knowledgeable computer community would give a lot more respect to the machine. I'm always amazed at how fast the Mac does I/O, and the reason is that it does it as fast as it can; no limitations or slow-downs jeremy mereness ============= jm7e+@andrew.cmu.edu (Arpanet) r746jm7e@CMCCVB (vax.... Bitnet)
hentosh@amethyst.bucknell.EDU (04/15/89)
>The bottleneck of the Apple ][ line is I/O. All reads and writes to the slots >and to the video are made at 1 MHz. The Inner Drive gets around this by basing > >jeremy mereness The way I understand it is that only *writes* to *video* will cause a slow down to 1MHz. Reads from video *do not*. Both writes and reads to the *I/O* spaces ($C000-CFFF) cause the processor to slow down. But access the the ROM, where firmware is located, doesn't cause a decrease in speed (in fact it runs faster than a program in fast ram since it doesn't have to slow down for refresh (refresh cuases an 8% decrese in speed)). The engineers did a good job! (the clock runs at 2.8MHz) Too bad another video buffer wasn't added or better yet a hardware memory move for the entire shadowed (well at least video ) area when switching to shadowed access. This brings me back to a queston asked awhile ago on the net, but got no response. If you access slow ram from native mode. The FPI syncs up with the MEGA II, well for how long does it stay synced for one operation, for part of an operation or for more than one operation so that it won't have to sync up for consecutive calls. Any IIgs engineers out there? Bob -- InterNet: hentosh@amethyst.bucknell.edu | 'Ever have deja vu and amnesia BITNET : hentoshr@bknlvms.bitnet | at the same time?' AppleLink PE : RobertH128 | -- Steve Wright
hentosh@amethyst.bucknell.EDU (04/15/89)
Just a thought, maybe you can get around the slow down by building and interface card in the memory expansion slot (expanded memory and fast I/O). And then write a driver for it, and patch the startup so that the memory manager will a fixed block on that section so that nothing disrupts that section. Sounds like a good idea to me, anyone see a problem with it? I might do build one someday. Not now though, too much school work. Realize that I gave a full 32 seconds to the above idea, so please tell me of any problems you see with it! Any IIgs gods out there? Bob -- InterNet: hentosh@amethyst.bucknell.edu | 'Ever have deja vu and amnesia BITNET : hentoshr@bknlvms.bitnet | at the same time?' AppleLink PE : RobertH128 | -- Steve Wright
labc-3dc@e260-3f.berkeley.edu (Andy McFadden) (04/16/89)
In article <sYFSoRy00Xoo4-gl52@andrew.cmu.edu> jm7e+@ANDREW.CMU.EDU ("Jeremy G. Mereness") writes: > [I write:] >> A recent issue of Call-A.P.P.L.E. had a hard drive comparison... among those >> reviewed were two new Sider drives and the InnerDrive. >> When it came to loading AppleWorks GS, the InnerDrive performed as well or >> better than the other drives. [blather] From April 1989 Call-A.P.P.L.E. : ***** Performance Results: Applied Ingenuity FCP Sider Intuition Systems Inner Drive 20 (68ms) D9/C96 (18ms) WARP Q84 (19ms) w/Driver wo/Driver w/Driver wo/Driver wo/Driver 1) 23:22 28:45 21:20 27:09 26:47 2) 1:22:15 3:04:40 1:39:31 3:19:43 2:35:07 3) 38:90 40:76 8:11 8:87 7:46 Where test 1 = GS/OS cold start to load test 2 = AppleWorks GS load test 3 = 100K seek/thrash (read blk $0, blk $989f, blk $1, blk $989e...) ***** Clearly the Intuition Systems drive out-performs the other two without a special GS/OS driver (important for ProDOS 8 applications). What do these cost? AI InnerDrive - $499 (20MB) / $650 (40MB) FCP Sider D9 (87MB) - $1695 FCP Sider C96T (a D9 with a tape backup unit) - $2495 IS WARP Q84 (84MB) - $1299 DISCLAIMER: I don't work for anybody. I own an InnerDrive, and I'm happy. >jeremy mereness -- fadden@cory.berkeley.edu (Andy McFadden) ...!ucbvax!cory!fadden labc-3dc@widow.berkeley.edu