Very strange 3D issue with OCZ PC4000 EB

[cantankerous]

Hello to anyone who reads this.

I am having some major issues with running 3D applications on my system. System is exactly as in sig and has been for 1.5 years except for the cpu which was just upgraded a couple weeks ago. I went from a 3200+ Winchester to the X2 4800+. I was using the same ram being OCZ PC4000EB Platinum Edition on both cpus however due to a really weak Winchester I wasn't able to run the ram any higher than 225mhz. The X2 is much stronger and is allowing me to run the ram at its desired stock frequency of 250mhz. Here is the issue. I can Orthos all night long, dual super pi 32M, OCCT, SnM etc however anything 3D and I get very odd issues that mimic video card instablity but it isn't the video card causing it. The screen will flicker black, sometimes half the screen will be cut off only showing a portion of the screen and in worst cases and quite often the monitors will shut off showing no signal. This mimics a vpu recover caused by the video card.

I have been using the same video card, with the same clocks and the same volts to a T on both cpus and this has never happened until the new cpu/higher ram speed increase. I originally thought it was the video card acting up however lowering clocks, raising volts etc had the same results. Anywhere from 20 mins to a few hours, totally random. I decided to try to lower the ram speed to see how it did. At this point it was only at 254mhz being a measily 4mhz above stock. Upon lowering the ram speed I noticed the problem getting less and less until at a low ram speed the problem was gone. Video card was left untouched as was cpu speed as I usually put the ram on a divider to enable the cpu speed to be left high. I have no clue what is going on but something is strange with this ram and 3D applications. I can barely do stock stable without a risk of issues.

I am running Tony's 406BTA bios and it has been cleared many times with the same results. Timings/volts on the ram doing nothing either as even up to 2.75v I cannot keep the ram much above stock without issues.

If anyone can help with some info it would be really, really appreciated. These memory problems are not your typical and it is very hard to diagnose when they mimic video card issues.

I greatly appreciate the long read, any help would be awesome. If you need to know specifics on anything else please do let me know. I look forward to some favourable responses.

Andrew

[jimmyz]

just a guess but did you try locking the pci express freq. in bios? my abit mobo on auto will raise my freq. if i get over 110 my xtx acts up. some mobos also throw more voltage to your card as they raise the freq ( if set to auto).

might not be the problem but it's worth a shot. try to lock it at 100 but try your same settings on cpu and ram.
hope it helps.

[mercyless]

have you tried dfi official bios,and memtest?
mine run flawlessly at the rated speed, around 265 and on, 3D is jerky/unstable
it looks an rma time. head over tohttp://www.bleedinedge.com/forum/forumdisplay.php?f=46

[mercyless]

just a guess but did you try locking the pci express freq. in bios? my abit mobo on auto will raise my freq. if i get over 110 my xtx acts up. some mobos also throw more voltage to your card as they raise the freq ( if set to auto).

might not be the problem but it's worth a shot. try to lock it at 100 but try your same settings on cpu and ram.
hope it helps.

dont think its pci-x as these are locked on dfi mobos

[fhpchris]

bad ram, CE5/6 starts having 3d issues before it becomes unstable in pi.

RMA.

[bachus_anonym]

Very well known issue with Infineon BE or CE-5/CE-6 based modules, which as far as I remember, is what EB had been made of (CE-6).

Past certain clock, usually above 250MHz 5-3-3-3.0, system would experience very weird issues in 3D apps: freezeing, screen corruption, etc. This has been covered a while ago and I had mentioned that many times bofore, when testing DDR RAM. There were better weeks of this IC, that would be all stable up to 270-280 on average, but in general this is something about IC that makes this happen. I've tested, I think, four different 2GB kits based on CE-5/CE-6 (Corsair, G.Skill, OCZ) and all were showing same symptoms - SP2004 stability would be about 10+MHz higher than 3D. Loosening timmings to 8-4-4-4.0, etc. did not help much...

We've learned to live with it and I'd advise you do the same Unless, OCZ will let you RMA it if this kit will is definetely NOT be all stable at 250MHz

[kemist]

its counter intuitive but try giving as little volts as possible, some ce5/6 have shown the highest clocks at low volts, ce-5/6 like low voltage and will die from high volts

[cantankerous]

Hi guys, thanks for all the responses, they are very much appreciated.

The DFI does lock the PCIE frequency so there is not much I can do there. I have it left at 105mhz default and that is they way it has been since day 1 on this board without issues.

I have read around long ago, before I even got this ram, that 3D stability was less than system stability when it comes to this ram however not to the extent of my system. Most people get at least 260, most get between 260-265 3D stable and you are right bachus, usually another 10mhz higher for system stability as a whole outside of 3D. I can barely do stock without issues in 3D. Very dissapointing as it is very much so affecting the OC of my cpu as most mem dividers keep wanting to throw the ram well above, or well below 250mhz if I want to keep my cpu speed high. I would like to run the ram as high as I can along with the cpu but don't like buying PC4000 based ram just to have to put it below 250mhz to play games which is what I mainly use this system for. I am going to do some extensive testing at stock 250mhz some more to see how it holds up. I do not want to do a dishonest RMA however if I keep having issues at stock speeds I am going to have to RMA to get some piece of mind when using my system in gaming/3D benching.

I have the volts on my DFI set to 2.60v which reads as 2.66v through the bios. The OCZ page states this ram at 2.7-2.8v so I think I am low enough on volts however raising it does nothing to help.

I thank you all again for your responses and if anyone has anything else to say please do share.

[Revv23]

yeah if they cant do 250 even with a rev-e cpu i think its rma time.

[fhpchris]

Very well known issue with Infineon BE or CE-5/CE-6 based modules, which as far as I remember, is what EB had been made of (CE-6).

Past certain clock, usually above 250MHz 5-3-3-3.0, system would experience very weird issues in 3D apps: freezeing, screen corruption, etc. This has been covered a while ago and I had mentioned that many times bofore, when testing DDR RAM. There were better weeks of this IC, that would be all stable up to 270-280 on average, but in general this is something about IC that makes this happen. I've tested, I think, four different 2GB kits based on CE-5/CE-6 (Corsair, G.Skill, OCZ) and all were showing same symptoms - SP2004 stability would be about 10+MHz higher than 3D. Loosening timmings to 8-4-4-4.0, etc. did not help much...

We've learned to live with it and I'd advise you do the same Unless, OCZ will let you RMA it if this kit will is definetely NOT be all stable at 250MHz

Somethings will help with this problem,

1) having an FX with a good memmory controler
2) a good board, like the DFI RD580
3) running the 250:200 divider to run low HTT and high ram speeds
4) using only the voltage you need.

I had 3dstability almost at the exact speeds My stuff was stable in 32m, so its just as dependant on your setup as the ram.

My 4000EB did like 260 mhz 3-3-2 @ 2.59v, so yours are bad

[kemist]

cantankerous, try giving the lowest volts possible, i remember big toe saw higher clocks with extremely low voltage (2.3-2.4v)

[cantankerous]

I will try 2.5v. That is the lowest my board will go. My board overvolts the ram so even then it will be 2.55 or so. I will let you know though I don't think it is volts I will still give it an honest effort.

[dnottis]

Very well known issue with Infineon BE or CE-5/CE-6 based modules, which as far as I remember, is what EB had been made of (CE-6).

Past certain clock, usually above 250MHz 5-3-3-3.0, system would experience very weird issues in 3D apps: freezeing, screen corruption, etc. This has been covered a while ago and I had mentioned that many times bofore, when testing DDR RAM. There were better weeks of this IC, that would be all stable up to 270-280 on average, but in general this is something about IC that makes this happen. I've tested, I think, four different 2GB kits based on CE-5/CE-6 (Corsair, G.Skill, OCZ) and all were showing same symptoms - SP2004 stability would be about 10+MHz higher than 3D. Loosening timmings to 8-4-4-4.0, etc. did not help much...

We've learned to live with it and I'd advise you do the same Unless, OCZ will let you RMA it if this kit will is definetely NOT be all stable at 250MHz

Yes, but that came to play about 270 Mhz. With my kit it was 268.. that was the max it would run 3d stable, about 280 superPI / memtest stable.

If you cannot run 250 stable you have bad memory. I could even run 250 2.5-3-6-2 with mine.

[bachus_anonym]

As not all CE-6/CE-5 modules were even SP2004 stable at 270, same goes for 3D. It varied from kit to kit and from IC week to IC week... But 3D stability issues were very rarely experienced below 250MHz mark.
I had a early CE-6 G.Skill that was stable at slightly above 250MHz and I also had this years' CE-5 Muskin that was rock 3D solid at 280Mhz and Corsair @ 271MHz 3D... The fact is, it varied a lot. That goes not only for OCZ EB line, but also Mushkin Redline, G.Skill' and others.

Bottom line is, those modules were acting like this, just a matter of @ what Mhz, but this rarely occured below 250MHz, although If you look back into the threads in XB, you will see (I'm sure you remember that) that there were many people complaining about problems at stock (not only OCZ, but Mushkin and G.Skill as well). This issue has caught memory guys by surprise as this could not be caught when binning modules (they don't test each kit in 3D benchmarks), unless you were doing it above it's rated speed, e.g. DDR-500 part binned for, e.g. DDR-520, but still sold as DDR-500. As the higher 2D stability, the higher 3D as well, therefore avoiding 3D problems at stock PC4000...

[bestmancajun]

I can push mine to 255 3d stable at a 1:1 ratio. I can only hit 2.76 @ DDR500 with out 3d problems but can prime up to 2.83. I use 2,3,2,7 DDR400 I can do 2.83 24hour prime/3d stable.

[cantankerous]

bestmancajun your sticks sound a lot like mine, but even a tad better. 255 is all for a premium kit like this, I can't even reach that. I have seen many kits hit at least 260 and that is what I was hoping for. Just really dissapointed I guess.

[STEvil]

got some airflow for them?

[bestmancajun]

I had ballistix before, they died... I'm very happy with my EB's still, even though I RMA'd a kit that would lag durring 3d at stock. I run them less then stock so I can get the extra 70Mhz out of my cpu, where I get max performance and stablity. My memory controller is crap plus the darn thing will not work in a expert board... The cpu goes farther on 2x 512MB, was getting stabilty at 2.87Ghz 2,2,2,5 @ DDR516.

[cantankerous]

I too am happy with this ram don't get me wrong. I guess I just expected more stability at such low fsb around stock for the price I payed on these things, especially since many, many others are getting at least 8-10mhz more out of their sticks.

[nike]

is ther a known issue w/ tccd? my tccds not stable@pi but stable@daily ganming/usage .. rmad ure stixx yet?