Author Topic: Debugging the Power Mac 9600 G4  (Read 11540 times)

Offline GorfTheChosen

  • Enthusiast Member
  • ***
  • Posts: 93
  • New Member
Debugging the Power Mac 9600 G4
« on: February 25, 2020, 07:05:32 PM »
So I've mentioned this system in a couple of other threads.

The short version of the backstory is that I acquired this unit back in the day as a new (at the time) unit ... so one-owner (IIRC) It was near or maybe even past EOL and its discontinuation date at the time I picked it up.

Think I was probably holding off on moving to a G3 because of software compatibility issues and because for what I was doing (motion graphics and animation) the G4 was on the horizon and would have offered some further advantages with Altivec over the G3 (or so I thought) while allowing developers to further sort out any issues with their code.

If I recall correctly it had a fairly short useful life before I moved up to a dual processor Gigabit Ethernet model.

After that the 9600 was mostly relegated to back up machine status and a node on the render farm.

At some point or another in there I picked up a Sonnet G4/450 upgrade card for it, added additional RAM (probably from other machines of its generation), etc.

So the RAM has been a bit of a hodgepodge ... some big DIMMs with lots of low density chips (pulled at the present time) combined with the better stuff (fewer, higher density chips) ... but its what I had.

Trying to get the unit back up running and functional has been kind of a nightmare, due to how unstable it has been. Some of that has possibly been due to 3rd party SCSI cards I had installed (now removed) ... will save troublehshooting those until after I get the machine running stably.

After tracking down some of the problems down to a pair of DIMMS that were returning lots of errors when tested, I'm now down to eight 128MB DIMMS which all use fewer, high density chips.

As covered in previous posts:

1. The machine is booting and running with a variety of different PCI cards installed (but no SCSI cards)

2. The machine seems to currently freeze/lock up doing two things: trying to write to a CD or DVD using Toast 5.2.3 (although this was occurring with Toast 5.2 as well) using either a Sony CD-RW (SCSI) or a Pioneer DVR-106D (Firewire), and using Fetch to transfer large files (several hundred MB) across the network @ 100Mb speeds.

I'm using TestMem, which was written by Mark Granger I believe (of EIAS fame), to test the installed memory. Basically what it does is repeatedly test memory in passes until you kill it. It does this by generating a random seed number and then writing that to all (unused) installed memory and then reporting each memory address where the value that was actually read back was different than the value that should have been written.

Each miss counts as an error, and the values (original and result) and the memory location are reported in a form something like:

"At address xxx, a value of yyy was reported, and it should have been zzz."

In the initial testing the bad RAM was returning many, many errors (easily in the hundreds)

After better than 12+ hour run last night, on around 1GB of RAM, the error count was 12 when I checked it a couple of hours ago.
 
That isn't "too bad" (probably less than 1 per hour) ... but it's not zero either.

Here's my question, owing to my own ignorance and lack of technical expertise:

Can the addresses where the errors occurred be translated into a location (memory slot A5 for example) where the suspect DIMM likely is ... or does the way the OS software handle memory management preclude that ?

Here are some of the address locations that were reported as having errors during this last overnight run:

$2BF4BC94
$2783DA5C
$2783DA5C
$2BF3DC94
$2BEFAC94
$2BE79C94
$2783DA5C
$2BEEEC94

I couldn't get the rest (of the 12 errors reported) because the machine is basically unresponsive while its testing (window is not scrollable) and the rest of the addresses had scrolled off the screen at the top. (The only option you get is to command-period it to kill it, at which point the window disappears - I should probably check and see if it creates a log file in the prefs folder or something)

The reason I'm asking is that I'd prefer not to spend possibly the next 4 or 5 days trying to track down which DIMM is actually bad if I don't need to.

Thanks in advance for any advice or insights.


Offline GorfTheChosen

  • Enthusiast Member
  • ***
  • Posts: 93
  • New Member

Offline GorfTheChosen

  • Enthusiast Member
  • ***
  • Posts: 93
  • New Member
Re: Debugging the Power Mac 9600 G4
« Reply #3 on: February 25, 2020, 08:42:46 PM »
Truth be told, the likelihood is very high that this machine will become "surplus to my needs" in fairly short order.

At that time I hope I can find it a good home with someone who has a need for it and can utilize the slots.

Right now, I'm just enjoying the challenge of trying to debug it and get it running stably ...  :)

Offline FdB

  • Platinum Member
  • *****
  • Posts: 666
  • And then...
Re: Debugging the Power Mac 9600 G4
« Reply #4 on: February 25, 2020, 08:57:23 PM »
Well... maybe not brand new RAM. (Seems I've a few sticks leftover 'round here.)
And if it really does come down to finding it a home... check here with Syntho.

(Read his thread / book covering his exploits with the 9600.)
http://macos9lives.com/smforum/index.php/topic,2656.msg15938.html#msg15938
This Must Be The Place

Offline GorfTheChosen

  • Enthusiast Member
  • ***
  • Posts: 93
  • New Member
Re: Debugging the Power Mac 9600 G4
« Reply #5 on: February 25, 2020, 09:18:45 PM »
Well... maybe not brand new RAM. (Seems I've a few sticks leftover 'round here.)
A veritable depot of surplus Mac parts ... as well as useful knowledge on a whole variety of things.

And if it really does come down to finding it a home... check here with Syntho.
Yeah ... saw his previous comment about wanting a "backup 9600" ... just in case.

Never hurts.

(Read his thread / book covering his exploits with the 9600.)
http://macos9lives.com/smforum/index.php/topic,2656.msg15938.html#msg15938
Yup ... I have ... twice ...  ;D

And will probably do so again - along with the other thread you referenced on MDD AcBel PSU's.

It takes a few runs over for me to lose the panic of needing to find a solution to a problem and for things to really start sinking in.

And speaking of PSU's, as of this afternoon I have a Fedex tracking number for the PSU's ... they should be here by this coming Friday.

Offline GorfTheChosen

  • Enthusiast Member
  • ***
  • Posts: 93
  • New Member
Re: Debugging the Power Mac 9600 G4
« Reply #6 on: February 27, 2020, 02:40:56 PM »
After disabling speculative access for the Sonnet G4 upgrade the number of memory errors reported during testing is now down to 6 over a 14+ hour run.

I did locate the original install floppy for the ADS Firewire drive case that the DVD-R is in, with an INIT and a configuration tool on the floppy. Installed the INIT - which the documentation states is specifically for HDDs - and then used the tool to set the unit to operate in ATAPI mode - choices were Hard Drive, ATAPI or Optical, or Auto-Configure. (Without the INIT installed, the config tool doesn't even see the drive)

Made no difference ... Toast still locks up the computer while trying to burn to that drive.

The documentation also mentions to check the ADS website in the future for additional INITs which support ATAPI devices ... so probably going to be doing a search in the Wayback Machine to see if it captured anything. Might be time to try swapping out the CompUSA USB/Firewire card for the ADS Firewire card that came with the drive case kit.

Did further research yesterday on the StarTech Gigabit NIC and based on the chipset it uses (RealTek) it may well function under OS 9 ... once I install a set of drivers for it. So I need to install those and test it this evening.

Offline GorfTheChosen

  • Enthusiast Member
  • ***
  • Posts: 93
  • New Member
Re: Debugging the Power Mac 9600 G4
« Reply #7 on: February 28, 2020, 08:45:01 PM »
Got the StarTech Gigabit NIC functioning in the 9600 and OS 9 ... after I figured out I had installed the wrong drivers for the chipset initially.

Once I got that sorted out it worked fine ... at least at 100Base speeds.

The current modem/router has only 100Base Ethernet ... but I have another one that will hopefully work I need to hook up and test out that is 1000Base.

If nothing else I can most likely use the LinkSYS gigabit switch internally for the local LAN ... and then uplink that to the current modem/router if the other isn't functional.

Offline Syntho

  • Platinum Member
  • *****
  • Posts: 1325
Re: Debugging the Power Mac 9600 G4
« Reply #8 on: February 29, 2020, 11:26:00 AM »
That long thread I posted -- I wish I could go back and edit it since a few minor things are wrong, plus I have more to add. But I don't have the power to edit after so much time  :-\

I can't comment much on the CD burner issue, but I have two machines with two different burners and they work fine with no problems. As for using Fetch, I've discovered that it's the devil for all of my 9600s. If you look at my post history carefully you'll see me posting about issues with it. Too long of a story to post here, but eventually I decided to plain just give up on it altogether. It works for ok for small files every now and then but it's best to delete it altogether. I never did replace the FTP client because even other FTP clients gave me issues.

I have a gigabit ethernet card as well and I couldn't even get the machine to recognize it no matter what so I went back to the onboard ethernet. While much slower, it's rock solid. I made a point to mention in the thread that a 9600 is usually best left as stock as possible. The more cards, addons and mods I tried, the more hell I experienced. The only real mod I'm using is a PCI card for an SSD drive which I record music onto. I learned that to get the system stable, you MUST boot off of an actual SCSI hard drive on the internal (or external) SCSI bus. I was trying to boot off the SSD but many problems cleared up when I went back to a SCSI drive for booting. That's another long story.

I actually did have some success and no apparent issues with certain CPU upgrades, but one brand of them I'd stay away from. I think I posted more about that in the mega thread.

As far as RAM, I put in 768mb on my main machine (advertised limit), but even that is overkill for my purposes. I'd be happy with 256mb. Actually, I'd be happy with 128mb.

The 9600 is by far my favorite computer ever. Just recently I decided to get another 9600 going for misc. purposes. Eventually I'll have it dedicated to 2trk stuff, like editing samples, maybe mastering, etc. I'm pretty sure after posting that thread, I've driven up the 2nd hand market value of 9600 machines by a good bit. Maybe I should've kept it a secret  :D
« Last Edit: February 29, 2020, 04:55:40 PM by Syntho »

Offline Syntho

  • Platinum Member
  • *****
  • Posts: 1325
Re: Debugging the Power Mac 9600 G4
« Reply #9 on: February 29, 2020, 04:50:49 PM »
Oh, one more thing. About RAM, I bought some brand spanking new RAM for my 9600 from OWC. Very shortly after I bought a few sticks, they go out of stock. I emailed them and they said they won't be manufactured anymore  >:(

What I do now is just hold onto what I've got. Some sticks eventually tested as bad, some sticks were good. I really don't like mixing-matching different sticks of RAM (bad experiences with it), especially because of the... I forget what it's called but it's the pairing thing across the slots.

If I have 128mb in a machine, it pretty much suffices for what I need to do. 256mb is preferable, but in my main mixing/production machine, I leave all of my 768mb of OWC RAM in it. The others are there with less just as backups and if my main machine goes down I'll probably switch the guts, including the RAM into a backup.

Offline Syntho

  • Platinum Member
  • *****
  • Posts: 1325
Re: Debugging the Power Mac 9600 G4
« Reply #10 on: February 29, 2020, 05:08:43 PM »
Yet another thing: I just took a look back, and turns out, the Sonnet upgrades were the ones that gave me trouble. I remember not being able to use MOTU software AT ALL using the Sonnet upgrade. The NewerTech MaxPower ones are most stable. I also did some searching at the Digidesign forum archive, and people running systems back in the 90s and early 2000s had the same experience: NewerTechs are rock solid and the Sonnets caused problems. Not necessarily running Pro Tools, but there are posts about the Sonnet CPU upgrade making certain software and hardware go haywire. Even people that worked at Digidesign came out and said to go with a NewerTech since those are the ones they tested their systems with.

If there's ever going to be a CPU upgrade in my system, it's gonna be a NewerTech, but I just leave them stock anyway  -afro-

Offline GorfTheChosen

  • Enthusiast Member
  • ***
  • Posts: 93
  • New Member
Re: Debugging the Power Mac 9600 G4
« Reply #11 on: February 29, 2020, 06:25:54 PM »
That long thread I posted -- I wish I could go back and edit it since a few minor things are wrong, plus I have more to add. But I don't have the power to edit after so much time  :-\

I know the feeling.

I can't comment much on the CD burner issue, but I have two machines with two different burners and they work fine with no problems.

Yeah ... I think my CD burner issues got sorted out by handling problems with the SCSI chain and the "speculative access". Looking back on it, I recall now that a couple of the cables I used were kinda suspect 20 years ago ...

I think I may still have problems though with that Pioneer DVR in the Firewire case ... will have to test that on another machine and see if it's something specific to the 9600 ... or actually a problem with the burner itself.

As for using Fetch, I've discovered that it's the devil for all of my 9600s.  If you look at my post history carefully you'll see me posting about issues with it.

Hmm ... interesting.

I'm a longtime user ... since at least version 4, maybe earlier. As a long-time license holder, I actually just paid to upgrade it to latest version in the last month (which will probably be THE LAST version) ... always been rock solid for me ... up until this last issue - which I suspect was not related to Fetch itself, but rather to Sonnet's software.

Too long of a story to post here, but eventually I decided to plain just give up on it altogether. It works for ok for small files every now and then but it's best to delete it altogether.

Define "small files" ... I did two 500MB+ file transfers (@ 100Mb Full Duplex) and I don't consider those small. I'm looking for to seeing how it does @ 1000Mb.

I never did replace the FTP client because even other FTP clients gave me issues.

I'm thinking maybe it wasn't the FTP clients ...  ::)

Here's one thing I can tell you about Fetch (in OS 9 on the 9600) from watching it in operation on the 9600: Fetch's "Fido" (the dog icon) will "run" while it's transferring files and then periodically it will stop (appearing to freeze), making it look like it's hung. It's actually not.

What's actually happening is the system (or maybe even Fetch itself) is doing something in the background that you can't see. If you give it a little bit (15 - 30 seconds or more - maybe a minute) Fido will eventually start chasing the bone again and the file transfer will complete.

It's kind of maddening, because a similar thing (Fido apparently freezes) can happen when the program (or the OS) HAS actually locked up the machine.

I have a gigabit ethernet card as well and I couldn't even get the machine to recognize it no matter what so I went back to the onboard ethernet.
What card is it ?

While much slower, it's rock solid.

Yup - onboard ethernet is solid here as well. I'll have more info about faster speeds than 10Base once I do further testing - but I suspect my problems were attributable to the inability to disable "Speculative Access" with Sonnet's software. Hopefully I've eliminated that by using PowerLogix's Cache Control software.

I made a point to mention in the thread that a 9600 is usually best left as stock as possible. The more cards, addons and mods I tried, the more hell I experienced.

Yeah ... I went down that road ... about 20 years ago, when it was my main machine. That was part of why it got tossed aside and replaced with a new Powerbook Pismo G3.

I suspect I was in too big of hurry and skimped on troubleshooting because I had work to deliver (and the impatience of youth of course ;D)

The only real mod I'm using is a PCI card for an SSD drive which I record music onto. I learned that to get the system stable, you MUST boot off of an actual SCSI hard drive on the internal (or external) SCSI bus. I was trying to boot off the SSD but many problems cleared up when I went back to a SCSI drive for booting. That's another long story.

Booting it off an SSD on a SATA card ... or an SSD hanging on an ATA card with a bridge board ?

Could make a difference ...  ;)

BTW - I think I saw you make mention of the stock SCSI boot drive being really noisy or whiny ... mine is actually really quiet (It's a 4 gig IBM, 50-pin Fast and Narrow) 

Having said that though though, I've got a pair of IBM 9 gig, a pair of IBM 4 gig, and pair of IBM 2 gig - all Ultra Fast and 68-pin Wide - and they are all the devil, noise-wise. It's like your sitting in the chair in the dentists office ...
 
I actually did have some success and no apparent issues with certain CPU upgrades, but one brand of them I'd stay away from. I think I posted more about that in the mega thread.

IIRC it was Sonnet right ?

I'm suspecting that the Sonnet INIT/software is the problem here ... using the Powerlogix software seems to have handled a number of issues for me. But the proof will be really giving the machine a good workout with some long, multi-day renders ... let it chew on something for a week or so and see what happens ...

As far as RAM, I put in 768mb on my machines (advertised limit), but even that is overkill for my purposes. I'd be happy with 256mb. Actually, I'd be happy with 128mb.

For what I did (and what I will do, once I get everything back up and running) RAM and disk speed was something you could never get enough of.

The 9600 is by far my favorite computer ever.


I'm a fan myself ... that's why I still have it after 20 years.

Just recently I decided to get another 9600 going for misc. purposes. Eventually I'll have a machine dedicated to 2trk purposes, like editing samples, maybe mastering, etc. I'm pretty sure after posting that thread, I've driven up the 2nd hand market value of 9600 machines by a good bit. Maybe I should've kept it a secret  :D

LOL ... hindsight is always 20/20 ...

Been there many, many times myself ... what ya gonna do ?  ;D

Offline GorfTheChosen

  • Enthusiast Member
  • ***
  • Posts: 93
  • New Member
Re: Debugging the Power Mac 9600 G4
« Reply #12 on: February 29, 2020, 06:39:47 PM »
Oh, one more thing. About RAM, I bought some brand spanking new RAM for my 9600 from OWC. Very shortly after I bought a few sticks, they go out of stock. I emailed them and they said they won't be manufactured anymore  >:(
Yeah ... that kinda thing sucks.

What I do now is just hold onto what I've got. Some sticks eventually tested as bad, some sticks were good. I really don't like mixing-matching different sticks of RAM (bad experiences with it), especially because of the... I forget what it's called but it's the pairing thing across the slots.

It's called memory interleaving ... which supposedly delivers a speed increase that is not insignificant (maybe 10% or more ?)

Deal was the same size and speed of DIMM needed to populated in the corresponding slot in both RAM banks (IOW: in both banks A and B) in order for the system to take advantage of it. If they were, it would just happen auto-majikcally ...

I do recall however that back in the day, it was suggested that memory interleaving could prose a problem for (some ?) G4 upgrade cards. Some were advising that if one had problems to make sure that the memory WAS NOT interleaved.

Kinda make sense in way if you look at XPostFacto/PostFactor ...

The author talks (and provides a means) to SLOW things down (particularly on a SCSI chain) during installs ... because it's too fast for OS X.

If I have 128mb in a machine, it pretty much suffices for what I need to do. 256mb is preferable, but in my main mixing/production machine, I leave all of my 768mb of OWC RAM in it. The others are there with less just as backups and if my main machine goes down I'll probably switch the guts, including the RAM into a backup.

Good plan ... makes sense to me ...

Offline Syntho

  • Platinum Member
  • *****
  • Posts: 1325
Re: Debugging the Power Mac 9600 G4
« Reply #13 on: February 29, 2020, 06:40:59 PM »
I think a lot of the problems with Fetch were due to filenames and weird characters which carried over to different FTP clients. I was also running System 7 and OS8.6. I don't remember the details exactly, but I fare much more well by using AFP in the chooser.

I'd have to dig out my box of cards to figure out what the ethernet one was, but I have I think what is a Sonnet SATA PCI card in my main 9600. There are some old posts of mine that cover this more in detail, but just one example of funky business you'll encounter with not using a SCSI HD is that certain MOTU software will crash and is unusable (their synth editor was the big one, the name slips my mind at the moment). Their software places an invisible key or authorization file on your main system drive, and when I say main system drive, I'm talking the one that the computer was made to have as the system drive. It's almost as if their software knows which kind of hard drive your computer came with stock, and if that drive isn't in the computer, the software will crash... even if you install their software on another drive!

Offline GorfTheChosen

  • Enthusiast Member
  • ***
  • Posts: 93
  • New Member
Re: Debugging the Power Mac 9600 G4
« Reply #14 on: February 29, 2020, 06:48:11 PM »
Yet another thing: I just took a look back, and turns out, the Sonnet upgrades were the ones that gave me trouble. I remember not being able to use MOTU software AT ALL using the Sonnet upgrade. The NewerTech MaxPower ones are most stable. I also did some searching at the Digidesign forum archive, and people running systems back in the 90s and early 2000s had the same experience: NewerTechs are rock solid and the Sonnets caused problems. Not necessarily running Pro Tools, but there are posts about the Sonnet CPU upgrade making certain software and hardware go haywire. Even people that worked at Digidesign came out and said to go with a NewerTech since those are the ones they tested their systems with.

Yup ... I can certainly attest to problems with Sonnet's processor upgrade cards ... when using their software anyways.

That was the real beauty of Powerlogix's Cache Control ... as I understand it, it worked with ANY manufacturer's G3/G4 upgrade card.

Dunno much about any others besides Sonnet, the only other I've had was a Bottomline Distribution "RailGun" G3 card ... which is how I acquired the Powerlogix Cache Control software.

If there's ever going to be a CPU upgrade in my system, it's gonna be a NewerTech, but I just leave them stock anyway  -afro-

If that works for you, then you go with it ...  ;)

Offline GorfTheChosen

  • Enthusiast Member
  • ***
  • Posts: 93
  • New Member
Re: Debugging the Power Mac 9600 G4
« Reply #15 on: March 11, 2020, 06:47:55 PM »
Decided to beat my head against the wall with the ATTO SCSI card a little bit more today. It's still hanging the machine at the gray startup screen with a cursor freeze. Tried pulling all the cards out one by one,  including pulling the TT128 video card and sticking it in the second slot (B) and the ATTO in the first (A), trying other slots, sticking a terminator on the external port of the card, and connecting the VXA tape drive.

Didn't get anywhere with it, but after I gave up on it I tossed the Adapter 2940UW in there just for s's and g's ... and surprise - it's working again !

Software sees the card, flasher would even let it me flash it ... if I wanted to (I took a pass since it is the last available, version 4.1)

It even sees the VXA-1 tape drive.

Presently have Retro going through and verifying the catalog files for all the VXA tapes while I have it up and running.

FWIW, here's the configuration at the moment:

OS 9.2.2
1G of RAM installed
Sonnet G4/450 upgraded installed and enabled with Power Logix's Cache Control software
Slot A - IMS TwinTurbo 128 video card
Slot B - Startech 8169 Gigabit NIC
Slot C - Wired4DVD MPEG-2 Decoder card
Slot D - Adapter 2940UW (only thing connected to it is the VXA tape drive externally, card termination set to automatic)
Slot E - Acard AEC-6260 ATA66 card (non-RAID, no devices connected to it)
Slot F - CompUSA USB 1.1/1394a card

As far as why it now seems to function, the only things I can offer are:

1. I did zap the PRAM a bunch of times, including during one boot where I zapped 5 times in a row.

2. Reordering/removal of cards (ADS Pyro 1394a has been removed)

3. The machine is laying on it's side, with the side panel removed.

At this point, if it continues to seem/run stable, I'm not terribly inclined to monkey with it much further ... beyond sticking a 120GB SSD in it, running off the ACARD ATA card.

We'll see how it goes though ...  ::)

Offline Syntho

  • Platinum Member
  • *****
  • Posts: 1325
Re: Debugging the Power Mac 9600 G4
« Reply #16 on: March 13, 2020, 08:02:18 PM »
Slot order is crucial. I've read about and have experienced some awfully weird card slot order stuff in my time for sure.

Offline GorfTheChosen

  • Enthusiast Member
  • ***
  • Posts: 93
  • New Member
Re: Debugging the Power Mac 9600 G4
« Reply #17 on: March 13, 2020, 10:30:33 PM »
Slot order is crucial. I've read about and have experienced some awfully weird card slot order stuff in my time for sure.

Yeah ... although the memories are somewhat dim at this point, I seem to remember it being an issue back in the day.

Probably part of why I threw in the towel on the machine so quickly.

I should have the SSD installed this weekend (while also retaining the original stock SCSI HD) so then it will time for a good burn-in to see if it will run stably.

If it does, I probably won't alter the configuration further.

Offline GorfTheChosen

  • Enthusiast Member
  • ***
  • Posts: 93
  • New Member
Re: Debugging the Power Mac 9600 G4
« Reply #18 on: March 16, 2020, 06:23:11 PM »
Got the SSD installed in the 9600 and got OS 9.1 installed on a couple of partitions on it.

Went to upgrade that to 9.2.1 and OS9Helper wouldn't cooperate ... crashed every time I tried to use it.

Not sure what's up with that ... since I think I used it fairly recently with no problems on the internal SCSI drive.

Ultimately I drag-copied the OS 9 System Folder off the internal SCSI HD.

Didn't really notice much difference in boot times but I wasn't clocking it either.

In other news from The Good News Department, turns out the internal Zip drive is not hosed after all.

Turns out a small label had come off one of the disks and was gumming up the works.

Once that was removed with a pair of hemostats everything seems back to normal.

Offline Syntho

  • Platinum Member
  • *****
  • Posts: 1325
Re: Debugging the Power Mac 9600 G4
« Reply #19 on: March 16, 2020, 10:54:20 PM »
You seem to be doing more diagnostics and burn-in testing than even the engineers did at Apple years ago  :-*