OCZ Value VX Stability issues

[odie]

Hi

I got myself a pair of Value VX (part number OCZ4001024WV3DC-K) and I've been having a lot of problems getting them stable in gaming mostly. They memtest quite well. For instance 245 is error free (test 5 and full test range, multiple loops) at 3.33v, but was horridly unstable in windows.

Voltage didn't help. Looser timings didn't help. Slightly lower overclocks didn't help. HEAPS lower overclocks didn't help.

What i found to make the most difference so far has been using weak drive strengths. Right now I use 7 or 5 and a decent number of apps (gaming mostly) are fairly stable for decent periods of time. 3Dmark05 and 06 are fine with this. NFS:MW (been playing it recently) is ok. Farcry seems fine. Source based games crash fairly quick however and so does UT2004 so its not exactly stable yet. Superpi, prime (haven't run it for a decent amount of time yet). OCCT and other stress testers are fine.

Some info.

I run an opteron 144, DFI Expert (tried both orig bios and e702-2bt), Antec Neo 550, 7800GT
I have active cooling over the ram, as well as right now a desk fan.

http://i48.photobucket.com/albums/f245/eertu/co_VVXtrouble310106.jpg

I'm having a lot of trouble getting over the last stability hump. I've been systematically going through trying different tref's and the like with no luck.

I have access to another similar pc to mine (x2 3800, a8n sli, gskill gbzx, neo 550) for testing. unfortunetely same PSU. So I can test out some things suggested. Although I already tried it and the system blue screened at 210 htt.


ANY help from someone experienced or someone who has had the same sort of problem before to fix this or even just find the culprit would be greatly appreciated.

[[XC] Jaco]

you can try to swap the sticks . This helps sometimes.
What about more vdimm ? what happens when you give it 3,4v

[odie]

yeah i tried swapping. tried new slots. tried them singly. not really any difference. voltage does nothing.

[markost]

Have you tried active cooling on the sticks. Under normal windows use there isn't much load on the ram but when video editing or especially gaming the temps ramp right up. There could be as much as 20 degrees C difference when you have good air circulation around them.

[markost]

Have you tried active cooling on the sticks. Under normal windows use there isn't much load on the ram but when video editing or especially gaming the temps ramp right up. There could be as much as 20 degrees C difference when you have good air circulation around them.

Sorry bud. didn't see that you already have got active cooling

[markost]

have you tried setting Bypass Max to 7. lower latency dimms tend to prefer the higher setting. Dynamic Idle cycle timer could be disabled to try to give more stability.

[odie]

tried both i *think* (on the bypass max, hard to remember)

um. while i'm at it i'll try to remember everything i've tried.

tried in the other pc as mentioned but that crash could mean anything. i would like to try this ram in MY pc with the x2 3800 but i can't (housemate wont let me go that far).

disconnected all unnesessary devices from power. i dont think its PSU anyway since its so selective about which apps it crashes in.

tried lax timings, 8-4-4-2.5, 8-3-3-2 etc.

tried more voltage. less voltage. more LDT voltage. more chipset voltage.

tried a range of trfc. 7,12. 7,14. 7,16. 10,24. a few others. also tried a whole bunch of trp, twr etc values.

tried weak and normal drive strengths. found weak to allow some better stability.

trying different trefs. not much difference between them so far.

tried running 2t, tried enabled or disabled dynamic idle cycle counter and interleave.

tried a few different async latencies and read preambles. tried different idle cycles.

edit : and just to make sure i tried bypass max of 7. it didnt crash as bad as it usually does if that means anything. hehe. crash to desktop instead of complete freeze.

[markost]

seems like you are on a sticky one there. Have you run Prime 95
http://www.mersenne.org/freesoft.htm
I've attached an excel file with my settings/comments on it. They are for VX4000. hope it can provide an insight.
I've found that Bigtoe and Andy from OCZ are always helpfull when things get like this.

[odie]

i primed my cpu extensively with the ram on a divider. i haven't done a decent prime on the ram. my guess is that it would fail after a while like all my games are. all the games listed above are ... well semi-stable. if i run them a while they crash.

it is indeed a strange problem.

i think i'll prime it tonight.

thanks for the help so far.

[NightCrawler™]

Maybe this trick will help you... It worked for me like a charm.
I had stability issues when OC'ing the whole system... plus some temperature problems. I stumbled on a thread and someone pointed out there's a more stable way for 3.3v on your memory..

http://www.xtremesystems.org/forums/attachment.php?attachmentid=42057&d=1136829276

I'm able to run very tight timings atm.... still testing though.... and my temps seem to have dropped a bit too...

NCâ„¢

[Markus_d]

try dymanic idle cycle: disabled
and set drive strenght to 8
and drive skew:1

try using Tref: 1168

[markost]

Normally if your ram is dicky it will fail Prime pretty quickly. A mate is running the same setup as mine (except he's using 4000 plat 2 gig) and he just dropped an FX60 into it and now his ram has gone from 280 FSB to 230. Maybe it could be the memory controller on your chip. When your system crashes, does it crash to the desktop, reboot or BSOD.

[markost]

Maybe this trick will help you... It worked for me like a charm.
I had stability issues when OC'ing the whole system... plus some temperature problems. I stumbled on a thread and someone pointed out there's a more stable way for 3.3v on your memory..

http://www.xtremesystems.org/forums/attachment.php?attachmentid=42057&d=1136829276

I'm able to run very tight timings atm.... still testing though.... and my temps seem to have dropped a bit too...

NCâ„¢

The Expert doesn't use jumpers for VDIMM. It's all done in BIOS. A positive design revision.

[NightCrawler™]

Neither does the SLI-DR.... In the BIOS it's either up to 3.2v or up to 4.0v depending on your jumper setting...
This trick only gives you an extra option to fixate it @3.3v...

NC™

[odie]

Normally if your ram is dicky it will fail Prime pretty quickly. A mate is running the same setup as mine (except he's using 4000 plat 2 gig) and he just dropped an FX60 into it and now his ram has gone from 280 FSB to 230. Maybe it could be the memory controller on your chip. When your system crashes, does it crash to the desktop, reboot or BSOD.

i've considered that too but dont really have a means to test atm. i REALLY hope its not.

when it crashes it depends on a few settings. the only one it never does is reboot. BSOD is next, hardly does that. usually its either a freeze, sometimes with horrible sounds (sometimes wierdly i can turn numlock on and off but not ctrl alt del or anything else), and more frequently a program crash to desktop.

i don't think it is the mem controller because i had some ocz premier sticks in here before and they ran fine at a whole bunch of settings. 6-3-3-2.5 at 200 all the way to 230, and if i kept them longer probably higher with tweaking. the value vx :banana::banana::banana::banana:s out way before that, weak, loose, more or less vcore.

[markost]

Tried overclocking the RAM only?

[odie]

using multipliers on the cpu?

nar. its stable at 2.6, 2.7 and 2.8 with the right vcore. i didnt really think it necassary. 2.6 with stock.

although might give it a go.

edit : ok tried multipliers. ran my processor pretty close to stock. no change.

ran prime large ftt, 9 and a half hours so far and no fail. going to stop it now.

[odie]

well had a bit of time testing today.

posting this from 1 stick of gskill 2gbzx at 230 tight timings

reason for 1 stick? unluckily for my mate during my testing i found the other stick to be faulty.

my ram went in his machine and exhibited much the same results as what happened on mine.

[markost]

Sounds ominous for the ram!
You are not alone!
I just put in the OCZ BH5 today and both sticks are bad. Won't even load windows at stock speed. Just goes to show that even a brand new set can be faulty.

[odie]

well thats not good.

[markost]

You should give Team Xtreem BH5 a go. OPB has got it running at 277 FSB @ 2.2.2.5.

Mine runs 265 @2.2.2.5 prime stable.

Goes harder than all the OCZ VX/BH5 which I've used before

[markost]

Neither does the SLI-DR.... In the BIOS it's either up to 3.2v or up to 4.0v depending on your jumper setting...
This trick only gives you an extra option to fixate it @3.3v...

NCâ„¢

The good thing about Bigtoe's jumper mod is that if you have a power supply with adjustable rails (OCZ powerstreams for eg) you can raise VDIMM via the 3.3v rail to get up to 3.5v if needed. The benefit of the Expert is that there are no jumpers at all for VDIMM. I broke some VX early on using the SLiDR when I wasn't aware of the 3.3v mod. Expensive mistake.

[odie]

i tried the x2 3800 now and the same thing. i also got some more bh-5, same thing.

i think its an issue with the board but i cant pin point what or how or WHY or anything. nobody else seems to have these kinds of problems.

the only untested gear is the PSU and expert mobo atm.

[markost]

Mate

It seems as if you have tried everything. Tony at OCZ has mentioned that CPU-Z 1.31 has been screwing up the SPD chips occasionally but that doesn't explain why your DIMMS play up on other M/B's. Might be worthwhile dropping him an email>>>>> oczguy2@ocztechnology.com

good luck

[odie]

will do.

probably not worth continued dialogue here.

thanks for your help.

[uwackme]

Your motherboard may just have a particularly picky Vtt, and this will limit the max stable MEM speed.

Mine max's at around 255Mhz, and I am sure the VX and BH5 sticks I have can go faster, but NOT on this motherboard. DFI NF4 SLI-DR.

Until there is a fix for Vtt spikes, there will be a limit.

Also, how old are your VXvalue's? Early ones had UTT-CH while alter ones went to UTT-BH chips. There are timing differences between them.

CH chips like 8,16,8,2,2,2,2.0 while BH like 9,14,7,2,2,2,2.0 plus diff's in other timings.

[odie]

they are fairly new. pretty sure they were BH