Memory bad??

MrBillMrBill Missouri Member
edited June 2004 in Hardware
I have never ran memtest86 before, but I have a system that I can't even load Win98 on without getting BSOD errors.

I downloaded memtest86 ver3 bootable ISO. Created the CD and booted to it. Not knowing what to expect, my screen starting filling up with information.

At the end of the first test, I had 29,650 errors. At the end of test #3, I had 175,000 errors.

Do you think my ram is bad? :D

I don't remember what brand it is. Some PC2700 I bought at CompUSA (free after rebate ;) )


Funny thing is, this ram was working in another system, NF7-S, (with another module just like it) without any problems. But tested in my KD7 as a single stick, it seems to have a MAJOR problem.

I put an old stick of OCZ PC3000 in it and no errors through test #3.....

Comments

  • Straight_ManStraight_Man Geeky, in my own way Naples, FL Icrontian
    edited January 2004
    RAM is bad or incompatible. Possible, fairly possible though not certain, that the RAM needs different timings to work on that board. IF you have not tuned the heck out of board BIOS, reset CMOS. IF you HAVE in fact tuned heck out of CMOS with BIOS Setup, then write down settings. Then reset CMOS. THen, after either choice above, enter BIOS when machine first comes up, and choose to do two things:

    First, you want default settings loaded. BIOS should auto-reboot. When your box comes back up from THAT, I myself woudl favor entering Plug-N-Play part of setup, tell BIOS it has a PNP O\S (enable that, could be Yes choice), AND to reset all settings in same submenu (essentially, you are asking it to recalc resources for everything and put itself in a mode that 98 can talk to and negotitate things, so 98 might find all sorts of new hardware and\or take one HECK of a long time to come up (seen boxes take 5-10 minutes to resync to a BIOS with those choices). THEN save settings, often this is ESC to get to main BIOS setup menu and then f10 to save. Again, BIOS will autoreboot.

    See if with BAD RAM in, the box actually boots into Windows or safe mode in windows. If yes, then restart Windows. THEN reenter BIOS setup on reboot,and change boot order to CD-ROM first, save settigns and reboot.

    If no go into safe mode, ram BAD or unusable with that BIOS version on that board. A 98 that is on a system with a BIOS told it has a non-PNP O\S can resdet itself to such an odd config that it is bogging and heisitating a lot, and drivers might not go in right until that is fixed. Resetting CMOS will force BIOS to resync RAM settings by SPD, in other words by querying the module for what it wants,and give you best chance to isolate down between true bad RAM and just mistimed RAM in BIOS Setup settings that are stuck into CMOS table. If you do all this and the silly RAM stick still does nto work, I would say check seating of stick, try a different RAM socket, and then RMA the thing if you can and none of this helps.

    Note that if you have a BIOS that is disabled, the changes will change back, and if you managed to short the CMOS battery contacts both at once, the CMOS cell might have done a sudden self-drain or just alarge percentage partial drain and the result might also be a bad CMOS table. In this case, you might get to give the motherboard a new battery.

    One installers thing to note-- do NOT slide aboard around with CMOS cell\battery enabled. IF you have a CMOS cell enabled, you need to slant board down toward port side, get ti aligned so none of the ground contact fingers get bnent down and short things (they are grounds that should rest on top of the ports and NOT hang down over sockets\ports), THEN lower other side of board down. IF you do this right, board will never touch any of the contacts on underside to studs and short things, but might drag across the ground rings you put screws through to go onto studs which should be very carefully placed so they only align with holes with metal rings around them. This shorting of things sometimes happens because some dealers ship with CMOS jumper enabled and folks slide board across studs and make unwanted contacts while doing so.

    TECH HUMOR NOTE, RELLY HAPPENED: I also had one guy, an A+ test prep student, mount his nice new mobo with ZERO studs, right onto the mtetal backplane. It booted once per battery, battery drained when box was plugged in as that metla plate is a GROUND plane and the battery was instantly grounded and discharged in about 10 sec max. He brought the thing to me, I put one CMOS cell in, one boot and no more, then I decided to unscrew board with power off. At that point, I did a mini-lecture to half the class about THOU must USE metal studs.... Instructor for class got a rare huge grin, and I got out of a week's class work (was there, but had free run of place to help others for a week, so long as I let instructor know what was up-- and finished that course in HALF the rated time also).

    New boards usually like all possible ground points that are ringed studded, not just 1 or 2. They can have 2-3 partial LAYERS for ground, each on different parts of board. Leave a critical ground unstudded, part of board acts defective or wonky. And was install-- not board itself-- problem in those cases.

    How does this last relate to issue??? Power stepdown circuits and RAM sockets have ground points, bad ground bonding for either can give strange apparent RAM failures that do not replicate from machine to machine while RAM is to mfr spec. Second, when troubleshooting, have PSU screwed onto case frame, PLEASE. Safety ground is through metal case of PSU to a ground wire fastened inside case to both case and of PSU and ground of the cord for PSU. With new boards, this is kinda critical to get results you can count on being accurate.

    John.
  • MrBillMrBill Missouri Member
    edited January 2004
    I could not install Windows with this stick of ram, so I could not try Safe Mode. ;) I put the OCZ ram in and installed Windows. It's running right now......


    As for the RAM timings, I had it set to SPD, but I did not clear CMOS after installing it. I may give the "bad" stick a try in my KX7 board as I have PC2100 in it now. Too tired right now...
  • MrBillMrBill Missouri Member
    edited June 2004
    Well....after struggling with installing Win2K on my new Xeon board (received my replacement CPU) for several hours, I remembered that I might have had a bad stick of ram. I ram memtest and sure enough....THOUSANDS of errors. I took another stick of PNY PC2700 from another box and re-ran memtest. NO ERRORS!

    Hopefully my dually will be up and running shortly. :)

    PNY has lifetime warranty so I'll be getting this bad stick replaced. It has caused me enough headaches!
  • mmonninmmonnin Centreville, VA
    edited June 2004
    Yep its bad. See if you can get another stick at least. I wouldnt spend much on getting another replacement stick like that one tho.
  • GobblesGobbles Ventura California
    edited June 2004
    I dont trust mem test... ive let it run all night on known bad memory and it never reported errors. We are talking sticks that only show half their size to the bios, cause post errors etc...

    I must be doing something wrong...

    gobbles
  • edited June 2004
    Gobbles wrote:
    I dont trust mem test... ive let it run all night on known bad memory and it never reported errors. We are talking sticks that only show half their size to the bios, cause post errors etc...

    I must be doing something wrong...

    gobbles

    The density of the ram might have been incompatible with the motherboard's chipset. I know for a fact that that is a major problem with buying ram for an old BX chipset board nowdays, the ram chip density is too high and the chipet doesn't know how to use it correctly.
Sign In or Register to comment.