spencer@eris.UUCP (02/05/87)
I have a question for the people at commodore. I have a friend who has just upgraded his A1000 to 1 meg of internal ram, it is addressed as Chip ram and he has to do something special to make it called fast ram. So the question is when I heared that 2 meg chips were coming out I asked and the story was that there was no upgrade path. Then my friend does this and is talking about adding a power supply to power the rest of the internal 2 meg chip ram, and was going to install all 2 meg. That means that if he doesn't do that trick that makes the amiga address it as fast mem, he would have an amiga with 2 meg of chip ram. Now couldn't that work with the new chips? What else do we need? randy PS. Who has 1 megabit chips for under 15 bucks yet? -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- Randy Spencer P.O. Box 4542 Berkeley CA 94704 (415)284-4740 C#(415)283-5469 I N F I N I T Y spencer%eris@berkeley.edu Now working for |||||||||||::::... . . spencer@USCVAXQ.bitnet But in no way |||||||||||||||::::.. .. . ....ucbvax!eris!spencer Officially representing ||||||||||||:::::... .. s o f t w a r e -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
captain@uhura.UUCP (02/05/87)
In article <2438@jade.BERKELEY.EDU>, spencer@eris.BERKELEY.EDU (Randy Spencer) writes: > > I have a question for the people at commodore. I have a friend who has just > upgraded his A1000 to 1 meg of internal ram, it is addressed as Chip ram and > he has to do something special to make it called fast ram. So the question is > when I heared that 2 meg chips were coming out I asked and the story was that > there was no upgrade path. Then my friend does this and is talking about > adding a power supply to power the rest of the internal 2 meg chip ram, and > was going to install all 2 meg. That means that if he doesn't do that trick > that makes the amiga address it as fast mem, he would have an amiga with > 2 meg of chip ram. Now couldn't that work with the new chips? What else do > we need? No. No. No. He would NOT have chip mem. If he failed to take the adress line over from the fast mem bus he would have unadressable memory. What your friend has done is a common hack currently. He has piggybacked another 512k of memory on top of his chip mem. He then takes one address line over from the expansion bus in order to map the memory into a different bank than the chip mem. What he has now is memory which the computer THINKS is fast memoru, but in fact has all the arbitration slow downs of chip memory. The command he is executing in software is an ADDMEM. He has to do this because etra memory (i.e. fast mem) is supposed to call out on the bus during startup as to how much and where it is. (This is part of the Amiga's autoconfigure, this frees you up from the limitations imposed by a system like the IBM-PC where you have to set dip switches internally). If he failed to do this he would NOT have chip mem there, he would have inacessable memory. In addition, all the Amiga ports are rated for power output in the hardware manual. Developers are assured that they can pull the rated power from the various ports' power pins. Concievably, if you had a system with a mumber ofperipherals, you could experience REAL problems with the extra power you are stealing from inside the machine. (Also, this hack requires you to solder directly to memory chips... yuk!) In conclusion. If you don't mind soldering to chips, your new memory being slow, still being restricted to 512k chip memory (the bus in the a100 has NO more room, so forget about expanding that), possibly pushing the limits of your power supply, and scaring any reasonable maintainence facillity from ever touching your machine again (it will be VERY non-standard), go ahead and do the hack. It is certainly cheaper. Myself, I paid $1500 for my Amiga, I can't afford NOT to spend the extra money to do it right! Jeff Kesselman captain@uhura Disclaimer: These ideas are mostly mine, though I hang about with some bright people and steal their's occasionally. >
grr@cbmvax.UUCP (02/09/87)
In article <2438@jade.BERKELEY.EDU> spencer@eris.BERKELEY.EDU () writes: > >I have a question for the people at commodore. I have a friend who has just >upgraded his A1000 to 1 meg of internal ram, it is addressed as Chip ram and >he has to do something special to make it called fast ram. So the question is >when I heared that 2 meg chips were coming out I asked and the story was that >there was no upgrade path. Then my friend does this and is talking about >adding a power supply to power the rest of the internal 2 meg chip ram, and >was going to install all 2 meg. That means that if he doesn't do that trick >that makes the amiga address it as fast mem, he would have an amiga with >2 meg of chip ram. Now couldn't that work with the new chips? What else do >we need? > >Randy Spencer P.O. Box 4542 Berkeley CA 94704 (415)284-4740 C#(415)283-5469 Well, there's chip memory, and chip memory. The problem is that the custom chips will only generate address for the first 512K of memory, being limited to 9 multiplexed address lines (18 bits == 262144*2 byte/work). The additional memory uses the timing and control provided by Agnus, but is only accessable from the processor side. There is another fix out there that relocates the added memory to C000000 territory, where 1.2 will automatically recognize it and not try to use it as chip memory. This results in a sort of "slow fast ram" configuration, but it works... As to the external supply, no need as long as he remembers that there are only so many amp's coming out of the supply, and if he wants to use more inside, then he can't expect to get the maximum allocation out of all the external ports. The real problem with making any new custom chips that address more chip memory is that there aren't enough pins on the 48-pin agnus chip to add another address line! So either a different package would be needed or some kind of external logic would have to be tacked on - which would realistically require a whole new PC board... -- George Robbins - now working for, uucp: {ihnp4|seismo|rutgers}!cbmvax!grr but no way officially representing arpa: cbmvax!grr@seismo.css.GOV Commodore, Engineering Department fone: 215-431-9255 (only by moonlite)
grr@cbmvax.UUCP (02/17/87)
In article <7@ethz.UUCP> claudio@ethz.UUCP (Claudio Nieder) writes: > >According to the AGNUS PIN ASSIGNMENT printed on page C-2 of the Amiga >Hardware Reference Manual Agnus has two VSS pins (27,41). Is there >really no way to use one of these to add an additional address line ? > claudio You ever try to take a ground pin away from a chip guy? Seriously, lead-frame impedance to ground is a serious problem with LSI chips. The more bus lines you push around, the greater the ground noise seen at the die. This can lead to glitches on the both the signals generated by the chip, and the chips's perception of external signals. As a result, you see packages with multiple ground and power pins and also a tendency to shift the power and ground pins towards the center of long, skinny packages. How many pins you allow is often rule-of-thumb or how many are available, although design rules for gate arrays do give more specific guidelines - lots of pins! -- George Robbins - now working for, uucp: {ihnp4|seismo|rutgers}!cbmvax!grr but no way officially representing arpa: cbmvax!grr@seismo.css.GOV Commodore, Engineering Department fone: 215-431-9255 (only by moonlite)