1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Geil Ultra failing Memtest

Discussion in 'Memory' started by manveruppd, 25 Jul 2006.

  1. manveruppd

    Wise Guy

    Joined: 24 Feb 2004

    Posts: 1,086

    Location: Leeds/Cyprus

    As title, I ran Memtest86 (the bootable floppy version) to test my 2x512MB Geil Ultra DDR400. I'm on an NF7-S 2.0, with an XP-M which for Memtest purposes I ran at 200x10 (and at default VCore) as I wanted to test the RAM on its default FSB of 200MHz. To test I would boot up Memtest86 and look Test 5.

    First I tested with VDIMM at 2.6V (default) and the RAM at 2.5-7-7-11. On the first run the computer froze midway through the 3rd pass. Memtest86 didn't report any errors, it just froze. On the second run it did 5 passes with no problem.

    Then I ran the RAM at 2-4-4-8 (its default timings), with VDIMM at 2.9 (the maximum that RAM is supposed to be rated at), and chipset voltage at 1.8V (achievable by dropping chipset voltage to 1.4 - my board is one of the buggy NF7s which does that;)). I got hundreds of errors starting 98% of the way through the very 1st pass.

    Then I dropped VDIMM back at 2.6V, keeping the same timings. Midway through the first pass the computer froze again. I ran it again, "dropping" the NB voltage to default, and it did 6 passes with no problem.

    Is it RMA time?
     
  2. manveruppd

    Wise Guy

    Joined: 24 Feb 2004

    Posts: 1,086

    Location: Leeds/Cyprus

    Oops... think I was wrong about this RAM's SPD: the BIOS reports 2-4-4-8, but CPUz says it's 2-3-3-[?] (not sure about the TRas, anyone know how I can find it out? And how can I be sure whether the BIOS reading or CPUZ has it right?
     
  3. Dutch Guy

    Capodecina

    Joined: 18 Oct 2002

    Posts: 24,563

    Location: Amsterdam,The Netherlands

    A few observations after reading your story
    - Default timings are not 2.5-7-7-11 or 2-4-4-8 but most likely as CPUZ suggested, I would use a TRas of 8 (add up the other timings and use that as TRas)

    Errors in test 1 are usually CPU FSB or VCore related so have nothing to do with the RAM but with the mobo or CPU.

    Errors in test 5 are usually memory related so you prolly got them due to the very relaxed timings you ran at.

    Try at 200Mhz 2-3-3-8 with 2.8V or 2.9V
     
  4. manveruppd

    Wise Guy

    Joined: 24 Feb 2004

    Posts: 1,086

    Location: Leeds/Cyprus

    Thanks for your reply!
    I did 2 quick passes at 2-3-3-7 at default Volts and got no errors (not an exhaustive test I know, but a good indication). I also did 2-7-7-11 at 2.8V and it ran error-free for 8 passes till I stopped it. Will next test 2-3-3-8 at 2.8V. I do remember reading on here that the NF7 overvolts the VDIMM, which would explain why all tests I ran at 2.9V failed (at any timings) even though this Geil is rated at 2.9V...

    PS. All tests are just looping test 5! :)

    PPS. Testing at increased chipset voltage fails at all settings, even though a few months ago I actually observed increased stability at Prime95 when raising it. I'm guessing it's overheating due to the hot weather. My room's like a greenhouse atm. I've actually fitted a fan on a Zalman bracket over the NB and it's still failing!

    PPPS. Wow, you're fast! :p
     
    Last edited: 25 Jul 2006
  5. Dutch Guy

    Capodecina

    Joined: 18 Oct 2002

    Posts: 24,563

    Location: Amsterdam,The Netherlands

    Yep, my NF7-S overcolted the RAM by up to 0.1V

    Also there is no need to run at 2-7-7-11 as that are way too slack timings.
     
  6. manveruppd

    Wise Guy

    Joined: 24 Feb 2004

    Posts: 1,086

    Location: Leeds/Cyprus

    I know, but those are (apparently) the most efficient timings for the NF7, so I wanted to test whether it would be stable in normal operation before I tested it at its maximum rating. I doubt I'll EVER run it at tighter timings than that - the only reason I'm testing at its SPD rating is because I might upgrade and sell it in a few months time, so I want to make sure it's running fine before passing it on to someone else.
     
  7. Dutch Guy

    Capodecina

    Joined: 18 Oct 2002

    Posts: 24,563

    Location: Amsterdam,The Netherlands

    A TRas of 11 gives alightly more bandwith than 5 - 8 on a nForce 2 system.

    Also remember that lower timings = faster operation ;)
     
  8. manveruppd

    Wise Guy

    Joined: 24 Feb 2004

    Posts: 1,086

    Location: Leeds/Cyprus

    Hmm... running at VDIMM 2.8 and 2-3-3-8, test 5 looped 41 times error-free, and then the computer hung on the 42nd!

    I've never had this happen to me while running Memtest86 before... usually either it'll run for hours until I stop it or it will start producing errors from the 1st pass - this seemingly random freezing is inexplicable to me, and I'm wondering whether it's not something to do with the RAM at all... I think I'll need to run Prime95 at small FFTs a few hours to test my CPU, though I've always run it at 2.2 or 2.4 completely stable, so it would be strange if it started hanging at 2GHz...
     
  9. Dutch Guy

    Capodecina

    Joined: 18 Oct 2002

    Posts: 24,563

    Location: Amsterdam,The Netherlands

    I would blame the RAM if it would produce errors, if it is hanging I would indeed suspect the CPU or the mobo (northbridge)
     
  10. manveruppd

    Wise Guy

    Joined: 24 Feb 2004

    Posts: 1,086

    Location: Leeds/Cyprus

    Hmm... Priming at Small FFTs failed after 1h 46'... I'm guessing the processor may indeed be starting to go!
     
  11. Dutch Guy

    Capodecina

    Joined: 18 Oct 2002

    Posts: 24,563

    Location: Amsterdam,The Netherlands

    Increase the VCore slightly, maybe the CPU is somehow used to higher VCore (if that is even possible :confused: )
     
  12. james.miller

    Capodecina

    Joined: 17 Aug 2003

    Posts: 19,431

    Location: Woburn Sand Dunes

    im having the same probloem with geil ultra-x pc4400 and my opteron. passes memtest in dos for hours, fails prime it 1 second flat in dos and wont boot to windows at all. with geil value ram the pc is fine.

    the ultra-x does it on 3 differen pcs, but that isnt enough to warrent an rma from ocuk :o
     
  13. manveruppd

    Wise Guy

    Joined: 24 Feb 2004

    Posts: 1,086

    Location: Leeds/Cyprus

    I'm convinced it's the heat in my room to be honest... CPU 48 C idle, whereas before summer it just about got to 45 under load!

    Geil gives lifetime warranties, so you don't need to go through OcUK!
     
  14. Dutch Guy

    Capodecina

    Joined: 18 Oct 2002

    Posts: 24,563

    Location: Amsterdam,The Netherlands

    The heat can make Prime95 fail.

    Maybe clean the CPU cooler?
     
  15. manveruppd

    Wise Guy

    Joined: 24 Feb 2004

    Posts: 1,086

    Location: Leeds/Cyprus

    Yep, I'm certain it's overheating, although according to MBM5 it never got above 52C even while priming, at which temperature I would've thought it'd be ok... :confused:

    Case internals got a dusting recently, so short of re-seating the HSF the only thing I can do is wait for the heat wave to pass before trying again... Thanks for the help! :)
     
  16. Cyber-Mav

    Capodecina

    Joined: 30 Jul 2005

    Posts: 15,528

    Location: Midlands

    dodgey ram?
     
  17. manveruppd

    Wise Guy

    Joined: 24 Feb 2004

    Posts: 1,086

    Location: Leeds/Cyprus

    That's what I thought, but not according to Memtest... I don't think it would simply hang completely if it was just the RAM.
     
  18. Nami

    Banned

    Joined: 29 Sep 2005

    Posts: 141

    Location: Shrewsbury, UK

    I know this thread is a little old now and you may be sorted, but I just had a similar experience with my PC.

    I was getting instability so kept backing off my overclock until I was at stock and it was mostly stable (but not as it shoud be). Memtest showed no errors, but my PC would not boot to windows when using CAS 2.5. Other CAS latencies were fine and dandy, apart from when overclocking. Prime blend tests would not progress. I notice on my rig that when it's all healthy, a prime blend test starts on 99/100% RAM usage but low CPU usage, then after a while the CPU kicks in to gear and goes 100% as well. When I was experiencing these problems, the CPU never kicked into gear.

    When I tried one stick of RAM at a time and forced it to use CAS 2.5, sure enough one stick worked fine and the other didn't work at all. Bingo, faulty RAM stick. But memtest did not highlight this regardless of what timings and test etc.

    I'm just about to RMA the RAM now. I thought this information may be useful for you as well as potentially you could have the same issue! I'm still baffled as to why memtest didn't show any errors at CAS 2.5 :confused:
     
  19. manveruppd

    Wise Guy

    Joined: 24 Feb 2004

    Posts: 1,086

    Location: Leeds/Cyprus

    Testing them one at a time is a good idea actually, thanks! I haven't done any more testing at all since I've been waiting for the weather to cool down! :p
     
  20. Nami

    Banned

    Joined: 29 Sep 2005

    Posts: 141

    Location: Shrewsbury, UK

    Yea and now it's raining all the time :mad: :p ;)