Memory problem?

Hi,

Ive been running my 2x1GB 8500C5 at 1066Mhz for a couple of months on an ASUS P5N32-E SLI with my ATI X1600pro, no problems at all. I finally got round to replacing my graphics and got a 8800ultra...

In XP i get alot of crashes and in Vista i get alot of graphics driver failures and restarts. I know theres alot of problems with the Vista drivers but my XP crashes started before i installed Vista :/ and those drivers should be stable enough anyway.

So while ive been tampering to make this more stable and have found that if the mem speed is anywhere below 930ish mhz, the stability greatly increases!

Is there a reason i cant run at 1066Mhz with my ultra all of a sudden or does

it sound like slightly ropey ram?

I dont suppose you would lay blame at the graphics card but i wouldnt rule anything out right now

yeah its all drivered up with the 169.25. (169.28on XP) and ive had the crashing probs on both xp and vista on fresh installs :(

I set it back to 1066Mhz earlier and booted up Crysis and, well crashed before it finished showing the EA logo!!

Restated, set bios to 5:4 (allowing for a 10% cpu overclock) that one way or another gives me 940Mhz, which let me have a good sesh on crysis and is still going strong as i type this (at 1066Mhz mem the graphics driver will also crash playing videos and crash just sat in the desktop too).

bad mem or bad graphics then????? i cant decide lol

Got an RMA for the graphics but not sent it back yet because i found slowing the ram helped :/ not ideal tho

O and worth mentioning the crashing seems to have no connection to the cpu OC, the only pattern i can find is linked to the mem speed.

O and, and may be worth mentioning that the EPP on these chips never worked, i set them manual from the SPD because pc would not POST when set to CPU0% setting (or anything other than disable tbh).

Just an update,

i ran memtest on my system all night. was at stock speeds so mem was 1066 and cpu was 2400.

then addded 10%OC to my cpu (gets a little more out of the graphics card) and the bios auto ratio'd my ram, so CPU is 2600odd and mem is 1055Mhz

ran memtest, said my mem was faster, 4100ish rather than 3750 and still passed OK.

Booted vista and its not crashed yet after a couple of hours and a bit of crysis.

But i just know its gonna happen again :(

Does this sort of intermittance make any sense?

yeah it just got stable :)

im sure ill be back at some point but i hope it just burnt itself in or something... :p:

Ok so not as good as i thought, it crashed in halo and cod4 then i changed the mem speed back to the stable settings, CPU=2640, FSB=1172, Mem ratio 4:5, Mem Speed=939.

So back to my original question regarding this mem, can there be an explanation as to why my mem does not stay stable on this system at 1066, but is fine at 939?

Thats one way of looking at it, the other makes sense too.

That is, that it cant really be a software problem if as soon as the mem is changed to 939Mhz, the system seems completely stable (im not ignoring the odd crash, theres simply no unusual freezes, no hangs in xp and no driver resets in vista)

That would indicate a hardware error and at the same time rule out a software one!

Tho both arguments make total sense, i think i need to run at 939Mhz for a week or so (my computers on nearly all the time) just to make sure its as stable as it seems right now!

I was hoping the symptoms would ring a bell but im just as confused as i was.

As for the driver cleaner stuff, it doesnt really apply because both xp then vista were installed on their own pre-formatted partitions!

Ill see what it runs like, any suggestions that might provoke a crash out of the mem at 939Mhz could be useful.

Tho my deepest suspicion is a memory problem, Im just going back to my pre-vista days, running at 1066, when i used to get a wierd low-level memory error in a dialog that i didnt really understand, it instantly crashed the software running but didnt kill windows. but tbh this was so very rare i never looked into it!

Thanks

EDIT 2: The only other, probably irrelivent point that crossed my mind when i got the mem out of the container, is the peaks on the heatspreaders are installed visibly wider on one of the sticks than the other giving the the impression that the sinks were installed differently on each stick :/ tho ive also noticed that neither stick gets that warm.

EDIT: i think this is what we all get for reverse engineering technology from teh UFO at roswell without truly understanding how it works eh? :p

Well at 939Mhz it ran for over 36 hours and crashed out after 3 hours of halo. Same crash.

Ive set to 800Mhz but kept the timings as per the 8500 spec. and am just going to see how that goes.

Cant get away from the relationship between the mem speed and the crash tho :/

Ok so i was running unlinked at 800Mhz but with the 8500 spec timings 5,5,5....

Unlinked because i just wanted to quicky set it to 800 after it crashed at 939.

BUT, after reading your post i though id make both our lives easier, went into the BIOS and loaded BIOS defaults. I was going to run like that (still with a 10% CPU OC but memory was going to be the JEDEC#2 profile at 800Mhz)

The only settings i changed as i reset to default was the enabling the RAID controller on SATA 1&2, so i would be able to get into windows.

So i turned off my comp, getting ready to go back into the BIOS to set boot device prority to my RAID partition, but it didnt POST, just an amber light on my monitor :/

I opened the case and reset the BIOS batery jumper. No POST.

I grabbed my old graphics and whipped that in, still no POST.

I took the ram out of BANK3, still no POST.

I moved the ram from BANK0 to BANK3, still no POST.

I removed the mem from BANK3 and put the original BANK3 stick in BANK0, and it POSTs with 1 gig :o

So i setup the RAID and popped the 8800 back in and loaded windows, restarted, popped the original BANK0 chip into BANK3, but no POST.

So i removed the original BANK0 stick from BANK3 and booted with 1 gig.

Thats where i am as i type this :(

I dont know if my problems were related to this chip being on its way out but i cant think of anything that i just did that would kill it tbh.

Now the 'dead' chip is infront of me i can tell you its CM2X1024-8500C5D ver1.2, 2.2V.

I may have a chance to try the memory in another computer tomorrow but i think its dead :p

O and btw, you mentioned using the SLI-MEMORY settings, i.e CPU 0% and such, but the EPP never worked on my chips and i always had to set it manually. Enabling SLI-Ready memory has always resulted in no POST for these particular chips on my mobo :(

Ok, thanks!!

Tho i think you are right, even with the one stick in single channel (and presumably any fault with the other stick is taken out of the equation), tbh it still froze in-game last night :(

Im curious to see if the crash is more frequent at the higher tested frequencies on the new sticks. Because at 1066 the crashes are immediate or within 5 mins and at less than 940 it might run for 3 hours with no problems.

Ill post again once i test this but im beginning to realise im not going to find a solution on this forum and probably not at all for the foreseeable future if other peoples, more drastic problems all over the internet with running nvidia drivers, have been complaining for 12 months. I can at least get my machine to run for 3 hours under load which is more than ALOT of people have been able to do this year with the newer hardware it seems, lookin round the forums :(

Nvidia needs a NVIDIA_GUY to answer the 3000odd posts specific to random and untraceable errors in vista and xp resulting in xp freezes and more-so Vista driver resets.

Thanks RAM_GUY

Hello again!

Just a quick recap for you... Ive got an RMA for some Dominator C5 1066 v1.2 memory. 1 stick seems to have failed, but ive delayed sending it back to you because i was waiting for another 2 gig of the same ram as ive decided to go with 4gb.

Question is, the new ram has arrived and is v1.3 tho all the specs are the same.

Would it be possible to request replacement with specific version?, as in 1.3's instead of the 1.2's im sending back, if it does make a difference?

Your RMA is valid for 30 days right?

Ok, just to say thanks for sorting the replacement pair of ram.

FYI i have 4 modules of 8500C5D Dominator running at 1066Mhz on my P5N32-E. I hear alot of problems involved where 4 sticks of ram have to be run at slower than rated speeds but just to say im really happy this does not seem to be the case on this particular mobo (tho i guess some natural variation applies.)

Thanks again!

Hello again!

Ok so all has been well (apart from 2 BSOD that i was just going to ignore), until yesterday when i was in windows, on youtube, using firefox. I typed in the search box, hit enter and the computer displayed a flash of BSOD before resetting.

On reboot it gave me BIOS checksum error and obv no post, just a booter for the mobo flash software luckily :)

I made a disk to recover the bios, it failed a couple of times leaving ascii artifacts on the display half way through the flash, but did eventually work.

Windows booted but BSOD just before the welcome screen.

I started troubleshooting by taking the 1 pair of ram out of the second dual channel bank. Windows booted.

I swapped the memory i'd taken out with the memory that was stable and windows wouldnt boot.

I put them all back in and went into the bios to check if they would work at their rated settings and voltage rather then the bios standard but noticed the BIOS was reporting 3GB of ram.

So... my mobo has killed another pair of ram :/

i dont think heat was a factor as the case is quite cool and i have the airflow fan on the ram too!

This is the v1.2 that i had replaced, the v1.3 i bought while i was waiting is ok.

Whats the best way to go from here?

Hi again, i delayed sending the duff ram back as ive just not got around to it :(

Story so far is; I bought 2GB, it blew and i RMA'd it.

While i was waiting for the RMA i bought another 2GB.

Recieved RMA and ran with 4GB for a while.

One pair failed again, the newly RMA'd pair.

Since then ive been running 2GB with the failed pair in the draw.

However, Last month or so ive been getting random crashes and freezes. I though it was normal windows trouble but it did get annoying when it would hang the computer while watching any video through any local media player (MPclassic, Media Player 11, VLC), and crash games too, but strangely less often than media players :/

I installed a cut down version of XP and was still getting the problem with little software installed so i downclocked the ram to 850ish and the problem dissapeared for a couple of days. Then i decided to update my BOIS as it says ram compatibility has been improved with 1405, it all went fine, went into the bios to setup raid and booted ok. I went into windows, jotted down the reccomended settings and went into the BIOS to update the ram timings (SLI-READY settings have never worked for me and this ram, i have to reset CMOS every time ive touched it). I changed everything to the spec you rate it as and that was the last time it booted. :/

I managed to blag 2GB of 800Mhz SLI-Ready ram from a mate, i tried the SLI Memory setting in the BIOS to CPU=MAXOC, and it worked perfectly :/ :/

Since then ive left my machine on for 4 days straight, playing videos, downloading, games, idle, cpu&ram stress test and ive not had a single crash at all.

In short... i have 4GB of very fast but very broken memory :( Can i RMA this and get something a little more reliable?

P.S Ive been using the dominator fan since i bought the pair while i was waiting for my first replacement, also worth noting, the dominator fan had one thumb screw that was unusable due to the thread coming away from the body. I couldnt be bothered senting it back for £10 so i just superglued that pin and its secure but not perfect ofc. Your stuff seems to be breaking on me as a matter of course, please make it stop :(

Ok, but fyi, the first two pair and the last two pair died while in 2Gb configuration.

Only the second out the the THREE pairs that died was in 4GB config (which i understood would either work from the outset or not at all depending on the individual mobo?), which did work fine for weeks.

The chips seem to be going through gradual decline before the big pop. Just my observation. And i could count on one hand the number of hours my PC's been turned off in the last week and so far no instability at all running SLI-READY ram with SLI-READY settings which have never worked for the Dominator :(

Are you saying that if they DO work with 4Gb at 1066Mhz, it is likely that they will gradually decline and die? Or that if they dont work at 1066Mhz, you recommend 800Mhz? or both?

Anyway, ill RMA them soon.