Page 1 of 2 1 2 LastLast
Results 1 to 20 of 29

Thread: Crash City - 8600

  1. #1
    Join Date
    Apr 2001
    Location
    Newton, MA, USA
    Posts
    22

    Default

    I am running what was originally an 8600/300 with the following configuration:

    1. 3 internal hard drives, the original 4 gig Quantum, a Seagate Barracuda 4 gig, and a Hitachi 18.2 gig.

    2. Upgraded processor: Crescendo G3/500

    3. Upgraded video card: ATI Radeon PCI card w 32 megs of VRAM

    4. Sonnet PLI card for USB 1.0 and FireWire ports (no devices attached yet).

    5. 416 megs of installed RAM

    6. System 9.1

    7. 56K V.90 USR Sportster dialup (high speed access is coming up my street slowly)

    My system was 9.0.4 but it became unstable several weeks ago and I tried reinstalling 9.0.4 and then upgrading to 9.1.

    I crash about ten minutes into using a browser (within 10" with Netscape 4.7.9, within 5" with IE 5.x, and within 10" with Opera (beta for PPC).

    MacsBug tells me that I have one of two problems at crash: either too long a word in a user memory space, or a corrupt system heap. I don't get this unless I am browsing the Internet, but I get it consistently when I am browsing.

    If I try to use Outlook Express 5.0.2 or 5.0.3 with IE, I crash almost immediately.

    I have tried all the remedies suggested on various forums:

    I've reinstalled my system, reformatted my hard drives, checked the media and volume integrity, trashed preferences files for everything in sight, run with minimal extensions, reinstalled each piece of software (always with no extensions running, or following the installation instructions). I tried upgrading systems. I checked for vestigial system parts or legacy files. I've reseated each of the memory chips, run diagnostics on the system with various sets of chips in or out, and I've re-seated each PLI card. I replaced the battery. I tried zapping the PRAM via hardware and via software, and I tried resetting the CUDA switch. I've checked the wiring and the cables, and reseated each. I installed an additional internal exhaust fan (very small) so that the processor runs at 35-39 degrees C. I tried reinstalling the drivers for my printer, my modem, and each other device connected. I've tried prayer, drink, sex, abstinence, and sin.

    Nothing works.

    Help!!!!!!

    Pete.

  2. #2
    Join Date
    Aug 2000
    Location
    Concord, CA
    Posts
    7,056

    Default

    Pete,

    My sympathies. You have done a lot of troubleshooting. Four things stick out for me:

    1.When you started having the problems with OS9.0.4, had you installed any hardware at that time?

    2.I'm not familiar with the 8600 but isn't there some kind of board on it (I am having a brain block and can't remember the name) that is famous for coming loose, particularly when working inside the machine?

    3.Are you running the latest Sonnet software?

    4.Memory intermittent. Your comment about system heap corrupt sounds like one or more of your RAM sticks has a problem. Obviously it passes the rudimentary RAM startup test but a more thorough test is in order, such as the Newer GaugePRO overnite or the xlr8 PowerControl memory test while running PowerFrax in the background overnite. They are on the Gurus ftp site. This possibility is my best guess for the problem. BTW, when running the tests, in Energy Saver CP, set all three settings to Never. Not sure about the Newer test but the xlr8 test will stall if something goes to sleep. k

  3. #3
    Join Date
    Jun 2000
    Posts
    769

    Default

    Hey Pete,

    Even though Apple seems to have abandoned us legacy OS 9.1 type folks, there were some essential system software updates to OS 9.1, especially to CarbonLib and MRJ. See what you get by trolling with the Software Update Control Panel. Might help.

  4. #4
    Join Date
    Aug 2001
    Location
    Grangeville, ID USA
    Posts
    9,124

    Default

    For what it's worth. From what you've said, one thing I can think of is termination. You have made major additions to your SCSI bus, is the last device terminated? Are any other devices terminated? Do you have any 80 pin SCA devices that you adapted down to the 50 pin internal and while were at it do you have any adapters at all? Are any of the devices set to supply terminator power? SCSI is finicky about this.

    I gather from your list that all your drives are installed on the fast/wide internal bus so I guess PCI card problems aren't the issue. What model drives are you running?

    Another thing to check might be firmware upgrades for particularly the modem but also from Sonnet, you might have a incompatability with 9.1.

    Just a couple ideas, hope it helps.

    Rick



    [This message has been edited by ricks (edited 27 November 2001).]

  5. #5
    Join Date
    Aug 2001
    Location
    Grangeville, ID USA
    Posts
    9,124

    Default

    I think the card Kaye is thinkin' of is the regulator in the G3, not the 8600.

    Rick

  6. #6
    Join Date
    Apr 2001
    Location
    Newton, MA, USA
    Posts
    22

    Default

    Well, kaye hit the nail on the head! The utilities he suggested showed a RAM error which hadn't shown up with any of the more general utilities I had used. I didn't have a chance to test the memory sticks by a process of elimination yet, but I'll do that. Is there a utility that identifies WHICH stick or sticks is/are bad?

    Many thanks!

    Pete.

  7. #7
    Join Date
    Aug 2000
    Location
    Concord, CA
    Posts
    7,056

    Default

    Pete,

    Unfortunately there is no utility that will point to the bad RAM stick. Trial and error shuffling the sticks is the only way. One thing though, a G3-500 will stress RAM much more than your original processor. The RAM stick could be 70ns whereas for most a 60ns or faster is called for with your G3. Look on the RAM chips on a particular stick for -7, -70, 7, or 70 or something like that, meaning 70ns. Hopefully you have mostly 60ns or maybe even 50ns. k

  8. #8
    Join Date
    Apr 2001
    Location
    Newton, MA, USA
    Posts
    22

    Default

    Kaye,

    I spent this weekend testing the RAM sticks in my machine, using the utilities I downloaded at your suggestion. Something peculiar occurred: The problem seems to be isolated in two 128 meg DIMMs. Let's call them A and B.

    I began by taking 2 64 meg DIMMs out of the machine and running the tests. The errors occurred with those out, so I eliminated those temporarily. (Later, I ran the tests adding them in one by one, and they both tested out OK.)

    Next, I took A out and re-ran the tests. No errors. Then I put A back in and took B out. I ran the tests and got the errors. So I assumed that the problem was in 128 meg DIMM A.

    Then I put B back in the machine, same slot as it had come out of, ran the test, and got... 5 errors, first run, then none on 15 subsequent runs. So I tried it in 2 other slots, just to check. No errors.

    Then I put A back in the machine and took B out, just to double check that I hadn't mixed them up. Errors. I tried it another 2 slots. Same errors.

    Then I took A back out and tried B again. This time, no errors. I restarted the machine and tried again. No errors for B, once again.

    So now I'm running the machine with A out, B in. And I get fewer system crashes, but still one or two, leading me to believe that B is also bad.

    So here are a few questions:

    1. I bought the 128 meg DIMMs RAM from a reputable, local Mac dealer (wish I'd bought it from you guys... you seem better-informed about Macs than anyone). They assured me that this RAM had been tested in 8600s and would work fine. It seemed to work fine, for months. Then this crashing started. Is it possible that the RAM went bad? Is it possible that my machine really DOESN'T work with 128 meg DIMMs?

    2. I'm assuming I should replace the 128 meg DIMMs. All my RAM is 60ns, as far as I can tell. The configuration right now is: 2 original 16-meg DIMMs, 2 64 meg DIMMs bought as an upgrade when I bought the 8600, and 1 of my two error-prone 128 meg DIMMs. Given that I have upgraded to the G3/500 processor, should I bite the bullet and buy all new 50ns RAM? If so, all EDO RAM? Where can I find that on MacGurus, or given the fact that I'm dealing with an upgraded machine, should I call and explain the situation instead?

    3. If I buy all 50 ns RAM, I'd be replacing 416 megs. What configuration would you recommend? Perhaps given the age of the machine, I should put less money into it and just buy less RAM. Any suggestions?

    Many thanks for your help in sorting out this problem.

    Pete.

  9. #9
    Join Date
    Aug 2001
    Location
    Grangeville, ID USA
    Posts
    9,124

    Default

    Pete,
    I was running 2x128mb and 4x64mb ram modules in my 8600. 128mb ram should work fine.

    At one point I had an additional 2x64mbs installed, those 2 were the only modules not interleaved, they were the same manufacturer but one was 2x32 and the other was 8x8. Even that never caused me problems before and after upgrading to a XLR8 500mhz G3.

    I get the impression that your computer runs fine without the 2x128s installed, is that true? If so, time to replace them. Keep the old stuff that works, newer faster ram won't show much (if any) in the way of noticable performance increases. As long as the G3 upgrade gets along with the ram, you have a winner.

    Rick


    [This message has been edited by ricks (edited 03 December 2001).]

  10. #10
    Join Date
    Aug 2000
    Location
    Concord, CA
    Posts
    7,056

    Default

    As ricks says, 128MB DIMMs are no problem in your machine. And either FPM or EDO is fine but personally I don't mix the two. My two S900's and my PTP are all FPM sticks because I need FPM for the PTP so I just purchased all FPM. The Crescendo G3-500 or its software may be a little pickier about your RAM than the equivalent xlr8 G3-500 and its software, I don't know.

    These days 50ns RAM is difficult to find and probably expensive. 60ns should be just fine. 70ns could be iffy with a G3-500. It is possible that one of the 128MB sticks has turned bad and I think you have identified the suspect one.

    Memory here http://www.macgurus.com/shoppingcart...i?edodimm.html and here http://www.macgurus.com/shoppingcart...powermacs.html your choice. I don't think I would dump what is working though.

    Another point, you are familiar with the fact that most applications, by default, do not assign enough memory for almost any serious use. Do a Get Info with the application not launched and assign more memory. Depending on the application, sometimes much more memory. A browser is a perfect example of needing much more memory if you use a lot of bells and whistles. k

  11. #11
    Join Date
    Aug 2001
    Posts
    232

    Default

    Keep in mind that a lot of memory pieces are sold with a lifetime warranty. You should take advantage of it if you think it applies.

  12. #12
    Join Date
    Apr 2001
    Location
    Newton, MA, USA
    Posts
    22

    Default

    Thanks, kaye.

    I usually assign 70-100 megs to my browser(s).

    I'll order some replacement RAM this week.

    One last question: If I keep the 128 meg stick that's working in, will the fact that it's not interleaved slow the whole system down?

    Thanks again.

    Pete.

  13. #13
    Join Date
    Oct 2001
    Location
    Varies widely
    Posts
    100

    Default

    70-100 megs for a browser? What kind of plug-ins are you running?

    No browser has any need for more than about 40MB unless you're viewing PDF files with Flash, MP3s, and the Star Wars EP2 trailer embedded in them. Which, of course, isn't possible

    IE 5 runs great in about 30MB for me. Mozilla, same deal. iCab likes 10MB just fine. All this under 9.2.1 on a PB G3/266 w/384MB RAM, so somehow I don't think doubling or tripling my RAM allocation is going to make any difference at all...

    p

  14. #14
    Join Date
    Apr 2001
    Location
    Newton, MA, USA
    Posts
    22

    Default

    In reply to THE PICKLE:

    I'm running mostly Netscape and Opera; since the summer, I can't get IE to run for more than 10 minutes without crashing badly. If I'm running OE at the same time, the crash is almost immediate; certainly within one minute or one click.

    I'm running a few plugins, nothing odd, and I do have to access heavy-duty PDF files for work, plus access our office database (I manage a technical search engine, ITpapers, and sometimes work from home).

    IE runs very briefly if nothing - and I mean nothing - else is running. Inevitably but not predictably, it crashes when I try to use a menu item or to do something else meaningful.

    Despite my rounding up all the usual suspects, updating, and reinstalling Java, IE always manages to do some odd things like disallow a perfectly correct JavaScript and freeze (solution: disable script debugging and re-enable Java features one by one in IE Options). If one didn't know better one would suspect that Microsoft didn't want its products to run with those of Sun or that there was some competition involved.

    I still haven't seen a really reliable way to deal with the IE+OE crash syndrome that began this summer. New version, old version, new installation, new preferences, new incantations, new spells, nothing gets them to work together.

  15. #15
    Join Date
    Aug 2000
    Location
    Concord, CA
    Posts
    7,056

    Default

    Pete,

    I just reread one of your earlier posts. In answer to your question, I'm not positive but I think that you won't lose measureable speed with the single 128MB DIMM (not interleaved) and the rest of your memory interleaved.

    Start with your largest interleavable DIMMs in A4/B4, next largest interleavable in A3/B3, etc., until you run out of pairs, then the 128MB DIMM in either of the remaining slots. k

  16. #16
    tachyon Guest

    Default

    Internet Explorer 5 problems are heavily related to the six-plus Microsoft Extensions in your System Folder. These, I have found, are easily corrupted and deleting them allows Microsoft First Run to "self-heal" and reinstall fresh copies into the System Folder.

    Not too long ago we had an IE5 problem that disappeared when we deactivated the Configuration Manager control panel. IE5 users ought to keep a spare copy of that CP and reinstall it as one of the troubleshooting steps.

  17. #17
    Join Date
    Apr 2001
    Location
    Newton, MA, USA
    Posts
    22

    Default

    Tachyon,

    Thanks for the suggestions. I had already reinstalled all the Microsoft stuff several times, and the problems continued unabated, but the Configuration Manager suggestion is something I hadn't tried... I will do it tonight.

    Pete.

  18. #18
    Join Date
    Oct 2001
    Location
    Varies widely
    Posts
    100

    Default

    WRT the IE/OE problems...

    Don't use OE. It's awful and is just asking for trouble. Use Eudora instead, which has NEVER given me problems, unless you count having to bump up the RAM partition a couple megs back when the Drawing Board theme (anyone remember that?) was all the rage.

    IE is fine with Eudora. iCab is even better (half the RAM usage of IE, with the *same* plug-ins), but a lot of web sites (fortunately none that I access regularly) are too stupid to know that there are browsers besides IE and Netscape, and refuse to work when faced with a browser identity other than those two.

    p

    ------------------
    the pickle
    http://macfaq.org

  19. #19
    Join Date
    Apr 2001
    Location
    Newton, MA, USA
    Posts
    22

    Default

    This gets curiouser and curiouser.

    I got the new RAM sticks, 2 128-meg DIMMs from MacGurus on Thursday. I installed them Saturday morning.

    My configuration is now:
    2 16-meg DIMMs
    2 64-meg DIMMs
    2 128-meg DIMMs.

    I tried to configure them as follows:

    The 2 128-meg DIMMs in A4 & B4.
    The 2 64-meg DIMMs in A3 & B3.
    The 2 16-meg DIMMs in A2 & B2. (They had come with the Mac, installed in A1 & B1).

    Here's the odd part: The DIMM in B1 would not fit tightly into B2. I don't know why; it just wouldn't. Everything looked good, there was nothing in the socket (which I inspected closely, millimeter by millimeter).

    I wound up installing the 2 16-meg DIMMs in A1 & A2, where they're not interleaved, but at least they fit.

    The machine fires up just fine, but now it indicates 544 megs of installed RAM.

    So of course I look at the 2 128-meg DIMMs I've removed, and of course, they're both 128-meg DIMMs, the old, bad ones. I look at every DIMM installed, and I confirm that I have 416 megs installed.

    The Mac still thinks there's another 128-meg DIMM in there.

    I tried re-pressing the CUDA chip and refiring the machine. Same results.

    I then tried running the memory utility that kaye recommended to me way back at the beginning of this RAM saga. Over and over again, everything checks out fine, no errors, 544 megs.

    Now I wouldn't complain, except that at the end of all this, I get the same old crashes I had before. "Too long a word in the memoryspace < ID >."

    I'm starting to wonder whether there is a different problem altogether, perhaps a solder joint that's gone bad.

    One more new symptom: Every once in a while, when the machine hangs and I try to invoke Macsbug, I get a blank screen instead of dropping into Macsbug. To me, that suggests a hardware problem or a bus error. I think kaye's suggestions were correct, but now I think this was only part of the story.

    Once again, any suggestions?

  20. #20
    Join Date
    Aug 2000
    Location
    Concord, CA
    Posts
    7,056

    Default

    Pete,

    Couple of things come to mind. What are you using to confirm how much memory is installed? PowerControl, as you have seen, runs the memory/PowerFrax tests just fine, but the rest of the info is broken with later OS versions. So don't trust its memory total. GaugePro should be accurate as well as Apple System Profiler (ASP) and About This Computer in the Apple Menu from the Finder. All those should agree. Also, if you did a Cuda Reset, Virtual Memory in the Memory CP may be turned on. Most of the above mentioned utilities will report the state of Virtual Memory as well as just checking the Memory CP.

    What tells you that the 2 16-meg DIMMs in A1 & A2 are not interleaved? Don't trust PowerControl for this. Your problems "Too long a word in the memoryspace < ID >." and the behavior of Macsbug, unfortunately I have no idea. Anyone? Wish I could be more helpful. k

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •