lgreen@pnet01.cts.com (Lawrence Greenwald) (01/15/91)
I've always been curious about the 9 Meg limit on the 2000. What would happen
if I were to have more than 9 meg in the machine, say in one of the following
combinations:
1) 1meg motherboard memory
2meg A2052 card
8meg GVP Series II w/RAM
---
11 meg physical RAM
2) 1meg motherboard memory
8meg GVP Series II w/RAM
4meg 32-bit memort (A2630)
---
13 meg physical RAM
This is not currently in my machine *BUT* the potential for these combinations
is quite real.
What would happen? Machine refuse to boot? Software go crazy?
Your opinions most welcome.
Larry Greenwald
UUCP: {hplabs!hp-sdd sdcsvax nosc}!crash!pnet01!lgreen
ARPA: crash!pnet01!lgreen@nosc.mil
INET: lgreen@pnet01.cts.com
SNAIL:4545 Collwood Blvd, #52 San Diego, CA 92115
"I'm looking over a three-leaf clover that I overlooked be-three!" -Bugs Bunnypeterk@cbmger.UUCP (Peter Kittel GERMANY) (01/16/91)
In article <6931@crash.cts.com> lgreen@pnet01.cts.com (Lawrence Greenwald) writes: >I've always been curious about the 9 Meg limit on the 2000. What would happen >if I were to have more than 9 meg in the machine, say in one of the following >combinations: > >1) 1meg motherboard memory > 2meg A2052 card > 8meg GVP Series II w/RAM > --- > 11 meg physical RAM > >2) 1meg motherboard memory > 8meg GVP Series II w/RAM > 4meg 32-bit memort (A2630) > --- > 13 meg physical RAM > >What would happen? Machine refuse to boot? Software go crazy? Everytime the last card in the row would refuse to autoconfig. Its memory simply wouldn't show up. For Autoconfiguration, the system only has the 8 MB pool of address space and when a certain board doesn't fit into this space or its remainder then it will not get activated. -- Best regards, Dr. Peter Kittel // E-Mail to \\ Only my personal opinions... Commodore Frankfurt, Germany \X/ {uunet|pyramid|rutgers}!cbmvax!cbmger!peterk
jeremym@brahms.udel.edu (Jeremy A Moskowitz) (01/16/91)
It would simply access 9 megs and that's it. No bugs, no freaking
out, no nothing ... Someone asked AmigaWorld that same question,
and that's what they said.
-j
--
E Pluribus // Contacts: jeremym@brahms.udel.edu or jeremym@chopin.udel.edu or
Unix // jeremy@freezer.it.udel.edu (line 1 = jeremym)
\\ // --->Monitor of comp.sys.amiga.emulations<---
\X/ 2001 Dalmations - My stars, its full of dogs...dave@cs.arizona.edu (Dave P. Schaumann) (01/16/91)
In article <744@cbmger.UUCP>, peterk@cbmger.UUCP (Peter Kittel GERMANY) writes: |In article <6931@crash.cts.com|lgreen@pnet01.cts.com (Lawrence Greenwald) writes: ||I've always been curious about the 9 Meg limit on the 2000. What would happen ||if I were to have more than 9 meg in the machine, say in one of the following ||combinations: || ||1) 1meg motherboard memory || 2meg A2052 card || 8meg GVP Series II w/RAM || --- || 11 meg physical RAM || ||2) 1meg motherboard memory || 8meg GVP Series II w/RAM || 4meg 32-bit memort (A2630) || --- || 13 meg physical RAM || ||What would happen? Machine refuse to boot? Software go crazy? | |[ extra memory would not be recognised ] How about writing some kind of device driver that would recognise the additional RAM and use it as another RAM: disk? Sort of like the memory expansion boxes that are around for the C64/128 line (sorry, I forgot the part #). Of course, this is a purely academic interest on my part, as I am still working to get across the A500/1Meg barrier. :( |Best regards, Dr. Peter Kittel // E-Mail to \\ Only my personal opinions... |Commodore Frankfurt, Germany \X/ {uunet|pyramid|rutgers}!cbmvax!cbmger!peterk Dave Schaumann | And then -- what then? Then, future... dave@cs.arizona.edu | -Weather Report
peck@ral.rpi.edu (Joseph Peck) (01/16/91)
In article <744@cbmger.UUCP> peterk@cbmger.UUCP (Peter Kittel GERMANY) writes: >In article <6931@crash.cts.com> lgreen@pnet01.cts.com (Lawrence Greenwald) writes: >>I've always been curious about the 9 Meg limit on the 2000. What would happen >>if I were to have more than 9 meg in the machine, say in one of the following >>combinations: >> >>1) 1meg motherboard memory >> 2meg A2052 card >> 8meg GVP Series II w/RAM >> --- >> 11 meg physical RAM >> >>2) 1meg motherboard memory >> 8meg GVP Series II w/RAM >> 4meg 32-bit memort (A2630) >> --- >> 13 meg physical RAM >> >>What would happen? Machine refuse to boot? Software go crazy? > >Everytime the last card in the row would refuse to autoconfig. >Its memory simply wouldn't show up. For Autoconfiguration, >the system only has the 8 MB pool of address space and when a >certain board doesn't fit into this space or its remainder then >it will not get activated. True enough for case 1, but for case 2 isn't the 4meg memory on the card in the CPU slot? Since I am 99.9999% sure of this (okay Dave, make me look silly now :), the only intelligent place to place this memory in the address space is outside of the 16meg limit of the 68000. This is especially true since I vaguely remember Dave mentioning that there are sufficient address lines for up to 64meg of 32 bit ram..... So, the end result (assuming that I am correct :) is that it is quite easy to put 13 megs in a 2000..... > >-- >Best regards, Dr. Peter Kittel // E-Mail to \\ Only my personal opinions... >Commodore Frankfurt, Germany \X/ {uunet|pyramid|rutgers}!cbmvax!cbmger!peterk Joe Peck peck@ral.rpi.edu An Amiga Engineer wanna'be....
peck@ral.rpi.edu (Joseph Peck) (01/17/91)
In article <677@caslon.cs.arizona.edu> dave@cs.arizona.edu (Dave P. Schaumann) writes: >In article <744@cbmger.UUCP>, peterk@cbmger.UUCP (Peter Kittel GERMANY) writes: >|In article <6931@crash.cts.com|lgreen@pnet01.cts.com (Lawrence Greenwald) writes: >||I've always been curious about the 9 Meg limit on the 2000. What would happen >||if I were to have more than 9 meg in the machine, say in one of the following >||combinations: >|| >||1) 1meg motherboard memory >|| 2meg A2052 card >|| 8meg GVP Series II w/RAM >|| --- >|| 11 meg physical RAM >|| >||2) 1meg motherboard memory >|| 8meg GVP Series II w/RAM >|| 4meg 32-bit memort (A2630) >|| --- >|| 13 meg physical RAM >|| >||What would happen? Machine refuse to boot? Software go crazy? >| >|[ extra memory would not be recognised ] > >How about writing some kind of device driver that would recognise the >additional RAM and use it as another RAM: disk? Sort of like the memory >expansion boxes that are around for the C64/128 line (sorry, I forgot the >part #). The biggest problem that you face is the lack of address lines for that much ram. There are a couple of solutions to the problem: 1) Add a little extra hardware/software and set up your memory in banks. The problem here is that EVERY program that you run must be aware of the bank switching, and, ummm, this is a dumb idea on the amiga (except for maybe chip ram, but still.... ). 2) Purchase an '020 or '030 (or soon to be '040) accelerator that supports 32 bit memory outside of the 68000 memory space. I believe that all of them do. 3) Build the Lucas/Frances accelerator, which allows at least another 4meg of 32 bit memory, and it is probably pretty easy to add more. 4) Finally, if you are looking for the least expensive way to add more memory, and you want to avoid as many of the clock speed/processor incompatabilities as possible, build the "Joe Peck's Incredibly Nifty 68012 Memory Board Expansion Project!" (I am probably going to regret mentioning this :) A little known processor in the Motorola 68000 family is the 68012. It is identical to the 68010, with the exception of having additional address pins for a 2 gigabyte space. A little circuit board with some additional decoding circuitry would be needed, as well as your own memory board. Done correctly, the processor daughter board would sit in the processor socket on the motherboard, and the memory board would sit in a Zorro II slot. A small ribbon cable would carry the additional address lines and decoding circuitry between the two. *BEFORE* I get 18 gazillion requests for this hack, all I have is some of the logic worked out in my *head*. The two main obstacles in my path are these: I don't know if the 68012 is even available any more. I have never designed or even looked at a memory board for the Amiga in my entire life. Oh yeah, one more. I plan on getting a 3000 (or a 3500 :), and that would satisfy my memory requirements. However, if anyone is vaguely interested in what I am thinking, at some point in time I can post or email another paragraph or two.... Right now I am finishing a commercial game (which all of you should run out and purchase. Maybe even several since it will be so good. It doesn't even break any programming rules (yet :). ) > >Of course, this is a purely academic interest on my part, as I am still working >to get across the A500/1Meg barrier. :( > > >|Best regards, Dr. Peter Kittel // E-Mail to \\ Only my personal opinions... >|Commodore Frankfurt, Germany \X/ {uunet|pyramid|rutgers}!cbmvax!cbmger!peterk > >Dave Schaumann | And then -- what then? Then, future... >dave@cs.arizona.edu | -Weather Report Joe Peck peck@ral.rpi.edu P.S. I have also designed a 14Mhz hack that I believe takes care of the E-clock problem. No schematics yet, just a logic diagram and a brief description..... At some point I will get around to posting it....
rwm@atronx.OCUnix.On.Ca (Russell McOrmond) (01/17/91)
In a message posted on 15 Jan 91 06:56:15 GMT, lgreen@pnet01.cts.com (Lawrence Greenwald) wrote: LG>What would happen? Machine refuse to boot? Software go crazy? In both cases, you would end up with 9M of ram - Autoconfig will only configure the first 8M of RAM (P.S. You ARE allowed to have 2M or CHIP so the actual maximum for an A2000 is 10M) :Later
rbabel@babylon.rmt.sub.org (Ralph Babel) (01/17/91)
It might be interesting to note that GVP have a new combination of '030 accelerator board plus 32-bit memory piggy-back board for the A2000 that can be populated with a maximum of 4 megs of autoconfig RAM in Zorro-II space and an additional 16 meg of "autoconfig" RAM outside Zorro-II space. So currently I have 21 megs of autoconfig RAM in my A2000 and still have 4 megs of Zorro-II space left, e.g. for my A2088-bridgeboard. This memory is _real_ main-memory, not just a big RAM-disk. The board can also be populated with 8 megs of pure Zorro-II RAM and numerous other configurations. Ralph
daveh@cbmvax.commodore.com (Dave Haynie) (01/17/91)
In article <677@caslon.cs.arizona.edu> dave@cs.arizona.edu (Dave P. Schaumann) writes: >In article <744@cbmger.UUCP>, peterk@cbmger.UUCP (Peter Kittel GERMANY) writes: >|In article <6931@crash.cts.com|lgreen@pnet01.cts.com (Lawrence Greenwald) writes: >||I've always been curious about the 9 Meg limit on the 2000. What would happen >||if I were to have more than 9 meg in the machine... >|[ extra memory would not be recognised ] >How about writing some kind of device driver that would recognise the >additional RAM and use it as another RAM: disk? Sort of like the memory >expansion boxes that are around for the C64/128 line (sorry, I forgot the >part #). Nope, that's not going to work. Briefly, here's how the AUTOCONFIG logic see things. It loops through boards, each of which in turn asserts its configuration register block in the 64K slot starting at $00e80000. Each board indicates the memory space it requires in this configuration information, and usually the OS decides where that board goes, and writes that configuration address to the board. If there isn't enough space, the board can't be added to the system at all. Most boards support the "shutup" feature, which allows the OS to make a board simply disappear, allowing it to get to any boards after which may be small enough to still fit in the available memory space. The C128 memory expansion devices are specifically designed to support banked memory (they actually have a DMA device which does fast block memory movements faster than the C128 can do them). You could build something like this for the Amiga, of course, but you can't talk a generic memory board and make it bank switch like this. >|Best regards, Dr. Peter Kittel // E-Mail to \\ Only my personal opinions... >Dave Schaumann | And then -- what then? Then, future... -- Dave Haynie Commodore-Amiga (Amiga 3000) "The Crew That Never Rests" {uunet|pyramid|rutgers}!cbmvax!daveh PLINK: hazy BIX: hazy "Don't worry, 'bout a thing. 'Cause every little thing, gonna be alright" -Bob Marley
daveh@cbmvax.commodore.com (Dave Haynie) (01/17/91)
In article <1991Jan16.030721.22132@mtiame.mtia.oz> jep@mtiame.mtia.oz (Jesper Peterson) writes: >>2) 1meg motherboard memory >> 8meg GVP Series II w/RAM >> 4meg 32-bit memort (A2630) >> --- >> 13 meg physical RAM >This should work because the 4M on the A2630 are not on the bus. 'Fraid not. While the A2630 RAM isn't on the expansion bus, it is in fact autoconfiguring memory, in the same logical address space as expansion bus memory. This is necessary to allow expansion bus devices DMA access to the A2630's memory. Motherboard memory isn't autoconfiguring memory, so that's not an issue. In the above case, the GVP memory probably gets shut up, though it would get added if you dropped it down to 4 megs. >Jesper. -- Dave Haynie Commodore-Amiga (Amiga 3000) "The Crew That Never Rests" {uunet|pyramid|rutgers}!cbmvax!daveh PLINK: hazy BIX: hazy "Don't worry, 'bout a thing. 'Cause every little thing, gonna be alright" -Bob Marley
peck@ral.rpi.edu (Joseph Peck) (01/19/91)
In article <7017@crash.cts.com> lkoop@pnet01.cts.com (Lamonte Koop) writes: >peck@ral.rpi.edu (Joseph Peck) writes: >>In article <744@cbmger.UUCP> peterk@cbmger.UUCP (Peter Kittel GERMANY) writes: >>>In article <6931@crash.cts.com> lgreen@pnet01.cts.com (Lawrence Greenwald) writes: >>>>What would happen? Machine refuse to boot? Software go crazy? >>> >>>Everytime the last card in the row would refuse to autoconfig. >>>Its memory simply wouldn't show up. For Autoconfiguration, >>>the system only has the 8 MB pool of address space and when a >>>certain board doesn't fit into this space or its remainder then >>>it will not get activated. >> >>True enough for case 1, but for case 2 isn't the 4meg memory on the >>card in the CPU slot? Since I am 99.9999% sure of this (okay Dave, >>make me look silly now :), the only intelligent place to place >>this memory in the address space is outside of the 16meg limit of >>the 68000. This is especially true since I vaguely remember Dave >>mentioning that there are sufficient address lines for up to 64meg >>of 32 bit ram..... > >I think you are confusing the difference between limitations of the CPU >addressing space, and limitations imposed by AmigaOS. AmigaOS limits >AutoConfig memory to 9.5 megs (v1.3). Additional memory can be added, but a >board must be set up in a way as to be put in the system memory pool in a >specific, unused place. [The OS must also be "informed" that it is there as >well]. The memory on the A2630 is AutoConfig, and therefore is tallied as >part of the AutoConfig total limit. No, I wasn't confusing the address space limitations, I just assumed that I wasn't confusing anyone else. Since Dr. Kittel already pointed out that the 4 meg of 32 bit ram IS autoconfiguring, my statement about the 13meg configuration being okay is out the door though..... Anyhow, what I meant by there being insufficient address lines was that Commodore specifications for the 68000 based amiga's included only 8 megabytes of autoconfiguring space. All other areas are basically considered "offlimits" for user type add on's. So even though there are holes in the 68000 addressing space outside of the autoconfiguring area, later versions of the custom chips may use those addresses. So in order to be safe, you shouldn't put things where they don't belong. I do realize that you CAN put things in those holes. In fact, I was considering putting my own register in to provide software control for a 7/14Mhz accelerator that I worked on a while ago. I decided not to because A) My hack is designed for a 500 or 2000, I happen to own a 1000. B) I am finishing a commercial game that will (hopefully) earn me some money, while making a nice hack won't at the moment. C) School inevitably tries to occupy every spare (and not so spare) moment that I have..... > > LaMonte Koop Joe peck@ral.rpi.edu
ford@amix.commodore.com (Mike "Ford" Ditto) (01/23/91)
In article <7016@crash.cts.com> lkoop@pnet01.cts.com (Lamonte Koop) writes: >>The biggest problem that you face is the lack of address lines for that >>much ram. > >No. The 68000 can physically address up to 16 megabytes of RAM. This does >include memory used by >any I/O devices and such on the bus, as the 68000 does use memory mapped I/O. So the problem *is* the number of address lines: All the 16 Meg space has been allocated (look at the Amiga Expansion Specifications), and there are no more address lines on the Zorro II bus. >The limiting factor here is AmigaOS, which limits _AutoConfig_ memory boards >to a total of 9.5 megs, regardless of the 68000's 24-bit address space. It's not the software but the AutoConfig hardware spec that specifies the range of addresses for AutoConfig memory boards. Unix, for example, is subject to the same limits. -=] Ford [=- "Goodbye, goodbye, goodbye, goodbye, (In Real Life: Mike Ditto) goodbye, goodbye, goodbye." ford@amix.commodore.com - Oingo Boingo, "Goodbye, Goodbye" uunet!cbmvax!ditto
david@kessner.denver.co.us (David Kessner) (01/24/91)
In article <890@amix.commodore.com> ford@amix.commodore.com (Mike "Ford" Ditto) writes: >It's not the software but the AutoConfig hardware spec that specifies >the range of addresses for AutoConfig memory boards. Unix, for >example, is subject to the same limits. > > -=] Ford [=- Geez. 9 meg for UNIX? 8 is bare minimum-- I have 12 here on my i386. Does this limit exist on the A3000? David Kessner