OCZ Enhanced Bandwidth PC3500 OC woes

[illusionz3]

For the last week or so I've been trying to tweak my newest system. It is a 3.0e, p4c800-e deluxe and 2x512 ocz 3500 EB. I'm having some real difficulty getting a stable overclock. I can get up to 3.2 Ghz at 1:1 with 2.5-3-2-8 timings. Anything past that is a mess. I boot into memtest86 and it usually fails test 5 at the very end. It gets to 99% and then reports around 1000 errors. I have no idea what the problem is here. I thought maybe I have a weak cpu so I tried some oc'ing with 5:4. I could get stable up to 3.7 or so. I ran prime95 for 10 hours error free at 3.6 with 5:4. Can anyone help me out? I bought this ram because I heard good things from people and the good reviews from anandtech. I was really hoping to get at least 3.6 1:1 with my setup Thanks for any help.

 

[MercenaryForHire]

What voltage are you feeding those? If it's still the stock (2.5v?) that's impressive.

- M4H

 

[illusionz3]

With a 5:4 barrier I was running 2.5-2-2-5, 240 FSB, 2.65v (Haven't run tests at 2.55v yet). With 1:1, I've tried 2.55v - 2.85v pretty unsuccessfully, with having to give 2.85v for 217 FSB, 1:1.

 

[3LEMENT0]

try relaxing the timings and upping the voltage, then if they pass the test then lower the timings. Trial and error method would work ok....good luck

 

[illusionz3]

I've tried a lot of combinations of timings and voltage, I shall keep trying though
Does the AGP voltage have anything to do with the memory, sometimes memtest doesn't like to start when i have 2.85v on the memory and 1.7v on the agp. It runs with 2.75/1.7 though, just wondering.

 

[myocardia]

That memory won't run that fast at 1:1. It's the OCZ PC3700 Gold, Rev. 2 that's so awesome. Besides, with any P4, running a memory divider doesn't hurt performance at all, as long as you can run tighter timings with the divider.

 

[illusionz3]

I don't kow about that. I've seen a couple reviews on the 3500 EB models that have gotten to 250+. I know not to expect the same results but I thought they would at least be able to run 240 FSB at even lax timings. Oh well

 

[myocardia]

Overclocking memory is exactly like overclocking processors. Just because my processor will do X.Xghz at default voltage, doesn't necessarily mean that yours is going to be able to do the same. And like I said in my last post, with a P4, running a memory divider hurts nothing, as long as you can run tighter timings with the divider enabled.

 

[illusionz3]

Yea. I think I'm going to try to get 250 5:4 stable. Thanks for your help.

 

[myocardia]

Okay, it took me long enough, but here is the link to what I was telling you about a memory divider not affecting performance with P4's: link to Thugsrook's testing. In case you don't know, Thugsrook is one of the most knowlegdeable and well-respected people on this forum.

 

[jhurst]

If you want to extreme overclock your memory, get some PC4000 so you can be sure it will run @ least 250FSB 1:1. For what you paid for the OCZ EB, I'm almost positive you could have found HyperX PC4000 cheaper.

 

[illusionz3]

Thanks for the link. Do you think I should settle for 2.5-2-2-5 with what I have now or switch to 2-2-2-5 ram?

 

[illusionz3]

Originally posted by: jhurst
If you want to extreme overclock your memory, get some PC4000 so you can be sure it will run @ least 250FSB 1:1. For what you paid for the OCZ EB, I'm almost positive you could have found HyperX PC4000 cheaper.

Yes, I probably could have. I think I'm going to go down the 5:4 road though.

 

[myocardia]

You don't have to have 2-2-2-5 with a P4. 2.5-3-3-6/7 would still be fast enough, as long as you're getting 400 more mhz out of your processor.

 

[illusionz3]

Sounds good, thanks a lot myocardia.

 

[smahoney]

The CAS 2 at 200 will beat out the CAS 3 at 250 in most real world benchmarks, unless you can get some CAS 2.5 at 250 which is really good to get. If you are thinking of swapping, I would recommend the Corsair XL 2-2-2-5 memory and run 5:4.

 

[Jhatfie]

That's weird that you have had trouble with the OCZ EB's. I've heard tons of success stories, myself included. And regardless of what some people may think, the EB's at a high FSB running cas2.5 is generally as fast as anything else running cas2. Mostly because A64's and P4's, the ras-cas delay is as or more important than cas. Now with the old XP's cas is the most important. My OCZ PC3500EB's (I am running one 512MB and one 256MB stick) run 240FSB 2.5-2-2-10 (Pretty much ideal timings for my A64) @ 2.9v with no errors in memtest. They also will run up to 275FSB with 3-4-3 timings at 3.1v (some memtest errors). Right now I have them at 250 with 3-2-2 timings at 3.0v error free. I am going to see if they will do 2.5-2-2 @ 250 error free with 3.1v when I have time.

Have you tried just using one stick? It is possible one of them may be bad and holding back your overclock.

 

[illusionz3]

Well well well. I did some testing. I've been running 3.6 Ghz stable with 5:4. I took out one stick and ran memtest with 1:1. Worked fine. Swapped sticks, memtest doesn't even load. So I guess I have a bad stick. *calls rma*