Conference hydra::amiga_v1

    When  you added the chips, were you in a static free work area? was the
    soldering iron grounded? were YOU grounded?  If you can answer NO to
    any of these questions, your trouble could be anywhere on the board,
    not just with the new chips.
    
    In an old issue of AW they reviewed a program called Dr Ami which
    (according to one reader's comments) found defective chips on his
    accelerator card.  AW did not think the program was worth buying but
    then the reader's comments are just as good as theirs.
    
    Jean

The built-in monitor in the 2620 rom should help you check the RAM
for error and help identify the bad chip. I recall you get to it by
something like L-Amiga/shift-M while holding both mouse buttons...
someone please correct me if I am wrong; it's been a while!

Typing a "?" and a return should get you a list of commands...

	Good luck!

		-SES

Article 5052 of comp.sys.amiga.hardware:
Path: sousa.enet.dec.com!shlump.nac.dec.com!e2big.mko.dec.com!pa.dec.com!news.crl.dec.com!deccrl!bloom-beacon!snorkelwacker.mit.edu!apple!usc!cs.utexas.edu!uunet!cbmvax!daveh
From: [email protected] (Dave Haynie)
Newsgroups: comp.sys.amiga.hardware
Subject: Re: A2620 RAM test program
Keywords: A2620 RAM test
Message-ID: <[email protected]>
Date: 20 Dec 90 22:40:16 GMT
References: <[email protected]> <[email protected]> <[email protected]>
Reply-To: [email protected] (Dave Haynie)
Organization: Commodore, West Chester, PA
Lines: 29

In article <[email protected]> [email protected] writes:
>I have an A2620 with possible RAM problem. What I need is a RAM test
>program, of course preferably one which pointed out a bad chip. When
>I mentioned this a long time ago, Dave H said something about having
>started on such a program once, but never finished it. Has Dave H or
>anyone else written such a program? If not so, maybe you know of RAM
>test programs that at least says if the memory is good or bad but is
>not A2620 specific? It is annoying not to know where the problem is.

The simplest thing would be to try out the memory test in the A2620.  Fire
up the system with the mouse buttons held down, and while still holding them
down, type "M" when the menu appears.  That'll put you in the A2620 monitor,
used by UNIX type in the past for debugging and all.  Now, to run a test
on a 4MB system, you type:

	T 200000 600000

The machine runs patterns through memory and yells if it find an error.  This
is far from ideal, and it doesn't point at the memory chip that looks bad, only
the memory location.

>Thanks in advance,
>Tommy Petersson


-- 
Dave Haynie Commodore-Amiga (Amiga 3000) "The Crew That Never Rests"
   {uunet|pyramid|rutgers}!cbmvax!daveh      PLINK: hazy     BIX: hazy
		"I can't drive 55"	-Sammy Hagar

	Thanks for the ideas. I'm having enough fun figuring this out that it 
almost compensates for the trouble.
	Static precautions were taken but its always possible that something
got zapped in a moment of inattention.
	The monitor is fun to play with and is a useful tool. So far the memory
test runs without complaints. I have been trying to use the monitor to look for
memory responding to wrong addresses and such. I can clear all the 2620 memory
(200000-600000) and then write a pattern into 200000-400000 then scan 400000-
600000 to see if those addresses are still clear then reverse the address ranges
and try again.So far I can access all this memory from the monitor just fine.
I also have removed the other cards one at a time to check for address
conflicts-no help there though. The i (config info) command returns a system 
config that matches what I get with the Showcards program and Sysinfo.
	I have tried running some non Amiga Dos type software and have gotten
a guru 810000005 " Memory list scrambled" running the first Newtek demo reel.
	If I get this fixed I'll put a note in here to let you know what I did 
wrong.

Are you sure you changed the memory size jumper and not the memory speed
jumper (eliminates a wait state as I recall... this may only be on the A2630)?

	_SES

	I have been able to boot my system with the new memory on line by 
deleting Fastmemfirst from my startup command file. I have tried various
memory tests off the Fred Fish disks and so far the only curious thing I've 
seen is that GFXmem shows all the expected memory with only a little bit 
in use til I startup the rest of the system. Gfxmem shows memory usage 
increasing til the system hangs. The Avail command shows all the memory as
well but shows 2 meg used with nothing running on the system.

	I recently picked up a copy of "Dr Ami" a diagnostic program for 
memory and disks. The memory tests run on my 2620 card as long as I set the
start address just above the actual 2620 start address (I'm then testing about
3.99meg). If I start right at 200000 (2620 goes from 200000 to 600000)
the system gurus with a 81000005 memory list scrambled. I am assuming that 
the memory is working but it isn't configuring properly. The 2620 monitor
memory test runs fine on the whole 4 meg.
	
	The "Bug Bytes" column in the latest Amazing/Amiga magazine indicates
that Dave Haynie has said some of the 2620 cards have problems with the second
2 meg of memory. Does anyone know more about this? I think I may zip a letter
off to Commodore asking about this.