E6600 OC Stability Difficulties(getting worse)-Question about CPU/RAM voltage, etc.

C

CalvinGT

Limp Gawd
Joined
May 7, 2007
Messages
251
Hello all, long, long time lurker and first time poster. I recently upgraded virtually my entire system and have been trying to tweak my system for the last few weeks. The problem seems to be that I am actually less stable now than I was seemingly a week ago. I have a few questions and want to list my current setup.

E6600 (currently at 3.3ghz)
Tuniq 120
2 GB Mushkin DDR2 800 RAM@ 4-5-4-11
EVGA 680i SLI-LT A1
Seagate 7200.10 320GB HDD
EVGA 8800GTS 320 Superclocked (overclocked to 600/1000)

So, I initially had this sucker OCd to 3.3 at mother board settings of about 1.38V. Orthos passed for over 2-3 hours. I tried to get up to 3.5, 3.6, and 3.7 at various times. I was able to get 3.6 stable Orthos for a couple hours at 1.525v and with the RAM at motherboar default timings (5-5-5-18?). I decided I was more comfortable with an OC of about 3.3ghz and went back to that level. However, over the last week or so I have been getting crashes in games after anywhere from 30 mins to 2 hours. I also occasionally get Orthos running for several hours with no issues, and occasionally for 5 minutes and the computer hard locks, which is very odd-Orthos at the beginning always loudly, very loudly beeped and said unstable and I shut it off and changed the voltages. Now, every time I have a problem, they entire computer hard locks. (edit, just tried at CPU 1.4025 and RAM 2.2, computer hard locked after about 50 minutes with Orthos, temps never cracked 48)

The mystery to me here is that things seem to be getting worse. I had Orthos stable for over four hours at motherboard settings of 1.38125 or some such and RAM at 2.1 (default) about a week ago. Now, that gives me hard locks anywhere from 3 minutes to over an hour or two in Orthos. No beeping, just hard locks. The hard locks persist all the way up to 1.4v CPU and 2.2v RAM. I am currently at 1.4025 and 2.2 but have yet to try it.

I do not understand a few things-why is Orthos causing a hard lock rather than the loud beep that indicates failure? That just totally baffles me as it seemingly changed overnight! Also why is my stability worse now than before, and what can I do to up this stability? Also, should Orthos be run with absolutely nothing else in the background for accurate results? Should my RAM voltage be upped to 2.2 (as I have it) when I am forcing it at 4-5-4-12 when I am at 3.3? Also, what is the difference between the voltage that I am inputing in the motherboard and the actual the motherboard is reporting-because when I put it at ~1.4, the motherboard in the bios tends to read around 1.35 or so.

Sorry for all the questions-the last time I overclocked was my celeron 300 in college and I am slowly trying to get back into the game.

For the record, CoreTemp and TAT show my idle temps at around 24-28 depending on voltage (up to 1.425) at idle, and anywhere from 45-49 under load. I have only cracked the 50s and into the mid 50s when the voltage was above 1.5.

Thanks so much for any help!
 
Have you considered running memtest on the system, if for nothing else then to rule out that it's a memory issue?
 
J_Devil said:
What's your power supply and CPU batch?
Click to expand...

My PSU is the newest model Mushkin 550 True Wattage. I will check the CPU Batch in CPU-Z after work and post. I also have not tried memtest because I didn't think this could be a memory error-could forcing the RAM to run in 4-5-4-11 at OC speeds do this? I figured it wouldn't since these Mushkin sticks are very highly regarded.

Thanks guys, I will run those tests and get that info after work!
 
My stystem appeared unstable after several crashes also. I re-installed the OS and things seemed to be fine again. Memtest is a good way to rule out the memory.
 
Ok, I seem to be stable at 1.4025 and 2.2 RAM, but I feel like those voltages are much higher than should be necessary for this overclock. Could there simply be a large disparity between the set voltages on the motherboard and the actual voltages (there does seem to be inside the bios display)?

Either way, my E6600 batch number is L631F577. I am going to make a Memtest cd tonight and see what I can see.

Thanks for any other input guys!
 
CalvinGT said:
Ok, I seem to be stable at 1.4025 and 2.2 RAM, but I feel like those voltages are much higher than should be necessary for this overclock. Could there simply be a large disparity between the set voltages on the motherboard and the actual voltages (there does seem to be inside the bios display)?

Either way, my E6600 batch number is L631F577. I am going to make a Memtest cd tonight and see what I can see.

Thanks for any other input guys!
Click to expand...

Look on the bright side, you are stable. My machine (sig) is running 3.2GHz with 1.35vCore set in BIOS and 2.1V on the memory, so I do not belive you are that far from what is needed in terms of voltage to run the current batch of E6600s. As far as actual voltage i loose around 0.05 volts on the core (measured with asus probe II), I don't know how much the EVGA boards drop.

As for your memory voltages I don't think 2.2V is all that bad, what are your sticks rated at?
 
Hey Darth,
Unfortunately, I have some bad news. I got home from work late, wrote down that batch number, and started playing some Quake 4 and other random things. Eventually I got a hard lock, so I think "shit not again" and reboot and run orthos. Stable for about 45 minutes, so I am thinking-this is ridiculous, its time to run Memtest. Without changing anything, I ran the newest version of Memtest. Bam. Wow. 9 errors in one pass.

So i went ahead and reset the CPU FSB to standard 9x266 and defaulted all the voltages and reboot. Whoa. Memtest went crazy with errors. I then decided to go ahead and reset everything in the bios and run it again. This time I got through one pass with no errors-I grew suspicious and went into the bios and looked at the settings. Sure enough, it had defaulted my RAM (http://www.newegg.com/Product/Product.aspx?Item=N82E16820146673) to 5-5-5-18, which is ok I guess except the RAM is rated to run at 4-5-4-11! So, unhappy with this I manually reset the timings in the bios, first by setting it to expert and entering the values myself, the second time by enabling the SLI memory feature (which defaulted the RAM to the correct values).

After doing each of these things I ran memtest. Each time I got 9 errors in one pass. I have not had time to confirm that the RAM gets no errors at 5-5-5-18, but I think its fair to say at this point that my RAM is dead/bad/something, right? I am very frustrated right now. Love my new build, love the speed of my computer and my snazzy case, but I have been a fastidiously careful overclocker, have NEVER overclocked the RAM, and ended up with dead sticks ( I also got my first BSOD last night at standard CPU FSB @ 4-5-4-11).

Any thoughts here? Please tell me its not my mobo or something like that. Please! I guess I am going to start the RMA process and see if I can get a gig of RAM cheap from someone to tide me over.

A few quick general questions:
- For a regular overclock (something like 3.2 on 6600) what is the customary procedure in terms of RAM timings. Did I make a mistake having the RAM on expert and forcing 4-5-4-11 at 3.2 (2.2v)? Was that what caused my issues? I never figured the RAM wouldn't run that speed when I left it at the default 400mhz, and it certainly had no issues for a while. If that is where I made my error I want to at least learn from my mistake.
 
This is a common problem. Ram works fine at a certain speed for a while and then over time it degrades and a few months later people start having random crashes and a lack of stability. It happens slowly and gradually that it's sometimes hard to put a finger on what exactly the problem is.

Here's my opinion. A lot of high performance memory, like all the modules based on the famous Micron D9GMH chips, began life rated by Micron at DDR2-667, CL5 timings and 1.8 volts.

These chips were binned so the better chips, with some additional voltage, ended up being used at DDR2-800 or DDR2-1000. In Team's top of the line modules, they're feeding 2.40 volts to ICs that were originally designed for 1.8 volts. Long time reliability is an unknown. Team may have a trouble free RMA process but who wants to go through the hassle and expense of buying more memory or spending months trouble shooting a failing computer?

It sounds like your memory is simply not capable of running at CL4 anymore. If you can get it to run error free overnight at CL5-5-5-18 then I guess the real problem is another memory manufacturer that has stretched the truth. You can send them back for another pair but a couple of months from now you might be having the same problems again.
 
Thanks very much for the reply Unclewebb. Unfortunately, we are only talking about this RAM working at these speeds for 2 weeks on the dot, and like I said-never overclocked, just set the timings at what it said the defaults were. Very very frustrating. I already submitted and RMA and am thinking about what to buy to tide me over until I get these back-thinking about the Crucial Ballastix or Corsair XMS since they seem pretty resellable.

Either way, you feel like its likely not a motherboard issue? Thats my primary fear since I dont want to have to tear everything out. And while your explanation makes perfect sense, it boggles my mind that RAM I bought at 220$ (with a big rebate and that now has plummeted in price) at such good timings simply will not run at those timings properly after two weeks!

Also-is it normal to have to manually input the RAM timings in the motherboard bios? Have always been curious about that.

You guys are great, glad I finally started posting.
 
I prefer to enter all memory timings into the bios manually. That way you know exactly what you should be getting and you can check it with CPUz to make sure. Setting memory to 2.10 to 2.20 volts is usually pretty safe. It's best to use the maximum that the manufacturer recommends so if you do need to return it you can do so with a straight face.

Are you sure your memory could pass MemTest86+ the day you first installed it? Sometimes there are problems from day 1 but you don't really notice them until later. It's always a good idea to run MemTest86+ overnight to make sure that everything is OK.

Your memory may have been rated at CL4-5-4 but that doesn't mean it will be able to run at those exact timings on every possible board. Mushkin may have used an Intel 965 board for testing and got those timings but it may not run those timings reliably on your 680i board.

In the old DDR days memory manufacturers would rate their memory at one speed for Intel and then with looser timings for AMD. Today DDR2 memory can usually run at better timings on AMD boards compared to Intel because of their integrated memory controller.
 
Thanks again Unclewebb. I honestly have no clue if it would have passed originally or not-I never even was smart enough to think to check until last night. However, I am going to run memtest all night tonight at the 5-5-5-18 settings just for the hell of it. In the meantime, I am glad you don't seem to be indicating that this might be a mobo or CPU issue (thank god for Memtest!!) Hopefully when I get new RAM in I can get back to my good OCs (was very comfortable at 1.425 at 3.4 before the RAM started going crazy).

For the record, I submitted my newegg RMA this morning, and went ahead and ordered the DDR2 1000 Ballistix from Buy.com and paid for the 3-5 day shipping. That RAM looks sweet and I am reading great things about it on these boards, so I will send back the mushkin when that arrives, and then sell one of the two based on how I feel about their performance. Its a tremenodous pain in the ass (and a shame newegg features no cross shipping protection plans), and its always frustrating to have something fail and have to pay 150 out of pocket for a replacement, but at least I should be able to recoup most of my losses by selling one of the two modules (too bad I bought that RAM at 185 AR and its 145 AR now-either way I am going to lose a few bucks!).

Anyhow, thanks for the help again, I will update you on the memtest results, and hopefully have some new OCing results to post next week :)
 
Ok, the Ballistix should get here tomorrow. I have two questions:

1. Should I run the RAM-FSB in linked or unlinked mode? If linked, how do I know how to safely determine how fast to push it with the RAM speed?

2. What latency should I run the RAM if I am overclocking modestly-say at 3.3 or 3.4? Should I just leave it at 5-5-5-15 or can I set it to 4-4-4-12 which many other users have reported is stable at 2.2-2.3v?

Thanks for any help you can give guys, I just want to maximize my OC potential and minimize danger of damaging the RAM.
 
You must log in or register to reply here.
Back
Top