g-verbru@rocky.cs.wisc.edu. (Rob Verbrugghe) (04/10/89)
Well now I'm a little confused on this whole memory speed issue. Now the
Mac +/SE runs at 8 Mhz (roughly) and requires 150 ns memory for a no wait
state operation.
The Mac II runs at 16 Mhz and requires 120 ns memory for no wait state
operation. But the 68020 takes more cyclces per memory access, so the
memory speed does not have to be doubled.
The Max IIx / SE/30 also runs at 16 Mhz and requires the same 120 ns memory
for no wait state operation. But, the 68030 is different than the '020
in the number of memory cycles required to access memory.
Somewhere in here the Apple Memory Chip vs the 68851 Virtual memory chip
figures in in the rate at which memory is accessed. I belive that the
68851 was slower than the Apple chip because it might have to do Virtual
Memory translation, and that takes time. The 68030 has some 68851
functionality built in, but since its on chip, it may not take as long
as a separate memory unit.
1/8 Mhz == 125 ns 2 cycles == 250 ns
1/16 Mhz == 62.5 ns 2 cycles == 125 ns 3 cycles == 187.5 ns
Now I see a RAM CACHE board being advertised that claims to give a
15% to 30% speed improvement to a Mac II. If there are no wait states
currently in the Mac II, and we do not adjust the speed of the processor,
then faster memory shouldn't help. Right?
Someone is confused. I suspect that it is me. Could some kind soul
set me straight on the facts here? Feel free to get technical, if I
can't figure out what you've said, I can get someone to translate.
Thanks buckets,
Rob Verbrugghe g-verbru@cs.wisc.edu {backbone}!uwvax!rocky!g-verbru
"So you think you know great, nail-biting excitement, you think you know truly
abject fear, you think you know total and complete despair, you think you know
the incredibly degenerate underside of this world we live in, and the ridicu-
lously despicable lengths that your fellow man can sink to, more rotten, more
putrid than the lowest form of fungus.... Oh. You are a sorcerer as well.
Then perhaps you do."
-- Further Conversations with Ebenezum, Volume III
"There are those who claim that magic is like the tide; that it swells and
fades over the surface of the earth, collecting in concentrated pools here and
there, almost disappearing from other spots, leaving them parched for wonder.
There are also those who believe that if you stick your fingers up your nose
and blow, it will increase your intelligence."
-- The Teachings of Ebenezum, Volume VIIalexis@ccnysci.UUCP (Alexis Rosen) (04/13/89)
Yup, there sure is a lot of confusion...
Whoever said the Mac II runs no waits? I _wish_ it did. Apple was inexcusable
cheap and lazy not to put a cache in. Not putting it in the IIcx, IIx, and
SE/30 is just sickening. Precisely because that $300 cache board (which would
cost Apple all of maybe $40 to manufacture on the motherboard) really does
make a 20-30% speed difference.
The Mac II runs 2 wait states, regardless of which MMU you use (the bogus
Apple one, or the 851). Most people think the IIx gets its speedup from the
added data cache, but in fact it comes mostly from the fact that with the
integrated MMU, it runs at only one wait state. The same goes for the SE/30
and the Mac IIcx.
I have the cache board. It's an excellent product which does exactly what
it is supposed to do. If you have a II, you should buy one. It generally
makes your Mac II equall to the IIx in speed, maybe a hair faster on some
things.
(BTW, I am certain about the II. My information about the '030 machines
comes from a variety of sources including two apple engineers, and I have
yet to see them proved wrong.)
---
Alexis Rosen
alexis@ccnysci.{uucp,bitnet}jmunkki@kampi.hut.fi (Juri Munkki) (04/14/89)
In article <1572@ccnysci.UUCP> alexis@ccnysci.UUCP (Alexis Rosen) writes: >Whoever said the Mac II runs no waits? I _wish_ it did. Apple was inexcusable >cheap and lazy not to put a cache in. Not putting it in the IIcx, IIx, and >SE/30 is just sickening. Precisely because that $300 cache board (which would >cost Apple all of maybe $40 to manufacture on the motherboard) really does >make a 20-30% speed difference. I read that the IIcx has the CPU soldered to the motherboard. I hope this isn't true, since it would make adding a cache a lot harder and probably will cost you a new 68030. Can someone check this out? I would also like to hear the reasoning behind this decision, if they did it. Has anyone made a survey of video boards? Are there any affordable boards with an accelerator or faster RAM? When the Mac II was introduced, I intended to wait for DMA hard disk controllers and 34010-based video cards. The SCSI interface turned out to be fast enough, but it seems silly that the access time to the video RAM is 300 ns (from the point of view of the processor). _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._ | Juri Munkki jmunkki@hut.fi jmunkki@fingate.bitnet I Want Ne | | Helsinki University of Technology Computing Centre My Own XT | ^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^
alexis@ccnysci.UUCP (Alexis Rosen) (04/23/89)
In article <21218@santra.UUCP> jmunkki@kampi.UUCP (Juri Munkki) writes: >I read that the IIcx has the CPU soldered to the motherboard. I hope >this isn't true, since it would make adding a cache a lot harder and >probably will cost you a new 68030. Sadly, this is true. In fact, I don't think there's any way to get the 030 off of the board without destroying both. The 030 isn't just soldered- it's surface-mounted, just like everything else (except, thankfully, the battery). The justification is simple: 1) It cuts down on failure rates due to bad contacts (official reason) 2) It makes cheap third-party processor speed enhancements like caches impossible (real reason) >Has anyone made a survey of video boards? Are there any affordable >boards with an accelerator or faster RAM? When the Mac II was >introduced, I intended to wait for DMA hard disk controllers and >34010-based video cards. The SCSI interface turned out to be fast >enough, but it seems silly that the access time to the video RAM is >300 ns (from the point of view of the processor). The only board I know of with an accelerator is the RasterOps board (I think it has an AMD QPDM on-board). It makes some things MUCH faster, others not at all. Very expensive. On the other hand, there are now three disk speedup products available (and we sure need them!). If you really need fast video, wait for the Mac IIex, which has two 34020s onboard (I don't know how this works, but it seems to imply that the video - at least the main screen - is interated onto the motherboard). It should ship this summer (I hear August MacWorld, but June is possible). --- Alexis Rosen alexis@ccnysci.{uucp,bitnet} alexis@rascal.ics.utexas.edu (last resort)
shap@polya.Stanford.EDU (Jonathan S. Shapiro) (04/24/89)
In article <1691@ccnysci.UUCP> alexis@ccnysci.UUCP (Alexis Rosen) writes: > > The 030 isn't just soldered- it's surface-mounted, just like >everything else... > >The justification is simple: >1) It cuts down on failure rates due to bad contacts (official reason) >2) It makes cheap third-party processor speed enhancements like caches > impossible (real reason) There are lots of legitemate reasons to gripe at Apple. We don't need to fabricate "dark side" theories. The reduction in failure rates, increase in board part density, reduction in cost of manufacturing, and improvement in noise immunity are all sufficient justification for surface mounting parts. Those of you who are used to being able to field-upgrade processors will have to change your expectations. Expect roughly all future machines to be build around surface-mount technology. The cache argument has absolutely nothing to do with it. Apple has no interest in shortchanging their own machine, and it is pretty clear that *they* aren't in the cache business, so it isn't an issue of competition. Lets talk about important things, like why my >$1500 upgrade doesn't include a goddamn DMA chip and reasonable SCSI performance... Jon
suitti@haddock.ima.isc.com (Stephen Uitti) (04/26/89)
>> The 030 isn't just soldered- it's surface-mounted, just like >>everything else... > >There are lots of legitemate reasons to gripe at Apple. We don't need >to fabricate "dark side" theories. >... >Those of you who are used to being able to >field-upgrade processors will have to change your expectations. And besides, i hear you can still buy Mac IIs, which are field-upgradable to current high end Mac speeds. My machine (a Mac II) is obsolete (which means, roughly, that you can get softare for it that works). >Expect roughly all future machines to be built around surface-mount >technology. And a good thing too. I carry a calculator (PSION) with over 64K of non-volitile RAM. Expandible to over 256K RAM. It talks RS-232. It runs for months (yes, that's plural) on a 9 volt transister battery, available at any drug store (you can still get them for free from Radio Shack - every month). It sings! It fits in my pocket. OK, so there aren't many sockets. Shucks. Still, it might be nice if Apple put the CPU on a card in a bus. Even DEC does this sometimes. Stephen.
alexis@ccnysci.UUCP (Alexis Rosen) (05/02/89)
In article <2600017@hcx9> goldfarb@hcx9.UCF.EDU writes: >Unfortunately, the Orchid cache board will not work with Mac IIs that have >the 851 installed. I've confirmed this with Orchid. What do I do now, >Alexis? (Or are you talking about some other product with which I'm not >familiar? If so, please enlighten me.) I've waited to answer this until I was certain. The answer is, all boards shipped after today will be compatible with the MMU. All boards acquired previously will be retrofitted at no additional cost. This just goes to show how good their support is. They also provide a FOUR YEAR WARRANTY. I don't want people to think that things are perfect, though. Virtual will now work with the board, but A/UX will NOT. No idea if it ever will. I feel certain though that when System 7.0 comes along, it will be supported. So unless you plan to run A/UX, it's a steal at $299, and even better when you buy mail-order. I'm still glad I got it, and I've had one since a few days before release. It's a very fine piece of work. --- Alexis Rosen alexis@ccnysci.{uucp,bitnet} alexis@rascal.ics.utexas.edu (last resort)