A4000 and Me!
13 messages in this thread
Greetings,
I have a question. I have an A4000 and I have two things I wish to do.
The first is to copy the kickstart ROMs into fast memory. The CPU command
supplied with AmigaDOS 3.0 doesn't seem to work, and neither does the
Set040 program in the AmigaTech Library.
The second wish is to be able to find out why my A4000 is losing track of
time in the tune of 6-10 minutes a day? Andy Finkel told me that there
was a jumper that I might want to change that had to do with the time
being ticked by the Power Supply (or something like that). Alas, this
jumper is undocumented.
I would appreciate any response.
-mark=
Just out of curiousity, why do you want to copy the kickstart
ROMs to fast memory?
–Doug
It is my understanding that the kickstart image loaded into memory
provides faster access to the system routines. It certainly was the case
with my A3000.
-mark=
Are you sure that's true on the A4000? Seems that with a 68040
built-in they would make sure the ROM had 32-bit addressing…
The difference between 16-bit and 32-bit addressing is what
made it faster on the A3000. But you'd need to talk to a
hardware type who knows the A4000 to tell for sure.
–Doug
Doug,
I am not sure–but I think even with 32 bit addressing to the ROM it would
not help as much as putting the beast in memory. Generally speaking ROMS
are much slower to read from than RAM.
I would love to get the expert opinion on this… anyone?
-mark=
Mark,
The ROMs used in the A1200/A4000 are fast ROMs, much faster than those
used in the earlier machines. They are also organized 32 bits wide. In the
A1200 they are supposed to run at zero waits. I don't think the A4000 runs
with zero waits but when you include the CPU caches, putting the ROMs into
RAM really doesn't gain you much speed. Also, with the changes to the MMU
in the 040 building the translation tables is supposed to eat up a rather
large chunk of RAM.
Dean
DKB Software
Dean,
Thanks–I appreciate your comments and the "answer". I can live with
it… I just wanted to squeeze another 150% of performance out this
lightening machine. You can't blame a guy for tryin' eh?
-mark=
Doug and Mark,
The FASTROM command doesn't appear to work on my A4000. I have assumed
that it isn't needed. It doesn't give an error, but also doesn't change
the amount of RAM available, or change the status line given by the CPU
command. I think Doug is at least partialy correct. Hopefully the ROMS
are also fast, in addition to being 32 bits wide.
Ron
Requires very clever use of the MMU. The A4000 MMU is very different from
the A3000 MMU. This the "very clever" part. Since it requires the MMU and
C= has more or less reserved the MMU for OS functions we'll all have to
wait on C= releasing a "cpu fastrom" sort of utility for the A4000. (The
existing cpu and setcup programs will NOT work with the A4000 MMU.)
{@_@}
The ROMs are a LOT slower than RAM; you'll see an improvement moving
kickstart to RAM (the fastest RAM you have, ideally).
That said, I dunno how to do it if set040 and CPU don't work.
Bob
Mark,
There's a command written by Greg Tibbs and included with the PP&S 040
cards that will copy Kickstart to 32-bit RAM on the A4000. Guess what,
though? The speedup is almost NIL! Between the fact that the ROMs are
32-bit and the fact that most ROM routines get cached by the 040, what
copying Kickstart to RAM does primarily is waste 512K of RAM… The
command is called FastROM040, and I've seen it on BIX and GEnie, if you
want to try it. Just put it before SetPatch.
Isn't there a little pot next to the clock for adjusting its speed? There
was on my Rejuvenator, and I thought the 2000 had one as well…
Denny,
We called Commodore on my 'clock' problem. The response was
'interesting', they wanted to swap the motherboard! 😉
Yea know.. I think I can live with this sub-space clock until around
Febuary or March of next year. Then I can warranty out both the clock and
IDE! 🙂
(grin!)
-mark=
Hehe, I think my 4000 is going to have a sudden failure about Feb or March
myself. But considering my luck, they would just stick in a IDE version
anyway 8^)