Overclock Stability

J

JamesL

Gawd
Joined
Nov 7, 2007
Messages
657
E6750, currently at 400fsb x 8


When I did my first overclock, I bumped the fsb to 350, and then 375, and ran prime95 w/ small FFTs stable for several hours. However, I crashed twice while playing a crysis demo on mid/high, and failed to boot once.
I bumped the fsb up to 385, and then 390, 395, and 400, and it failed to boot. I rebooted it, and it worked fine, ran prime95 for the past 4 hours; 0 warnings, 0 errors. Task manager reports both cores at 100%.

Core temps peak at 58 C, and all voltages are at stock - FSB ratio 1:1.


Do the crashes and boot-failure have anything to do with CPU stability? (Do real-life-applications differ from cpu torture tests in this aspect?)
Am I utilizing prime95 correctly?

I have no intention of raising it past 400x8, but 2 boot failures and in-game crashes worry me.
Thanks in advance for any input -
 
Are you running the beta version of Prime 95, that runs 2 at once, or are you using the original version that only runs one thread at a time? If you are using the original, you must open 2, and set the the affinity to 0/1. Or just download orthos, its a nice little GUID for Prime and will run 2 threads as is, so it is using both your cores.

Up the voltage a little bit, stock voltage is nice, but once you start overclocking its nice to give it a little boost due to vdroop, and to even out stabilty. A perfectly stable CPU should be able to pass 24 hours of prime. I have heard, crysis is a great CPU test though, cause if theres any lack of stability Crysis will find it.
 
I have found it best to do other things on your pc while priming. I've had cpu's that would prime all day and night, but if I open a web browser while priming it would lock/reboot/ect.
 
I run 475 * 8, but I'm on water and everything is overvolted. I can run 500 x 8 but its not stable unless the fans are on full blast which gets annoying. Before I put the water in I was running 425 stable with the stock intel cooler. Mobo is an Asus IP35.
 
(Do real-life-applications differ from cpu torture tests in this aspect?)
Click to expand...

Absolutely, the best example is your in-game crashes, Prime etc. do not push the ton of video information from the cpu, through the MCH/Northbridge, to the video card like a game does. Add cooling to your MCH/Northbridge.
 
JamesL said:
E6750, currently at 400fsb x 8


When I did my first overclock, I bumped the fsb to 350, and then 375, and ran prime95 w/ small FFTs stable for several hours. However, I crashed twice while playing a crysis demo on mid/high, and failed to boot once.
I bumped the fsb up to 385, and then 390, 395, and 400, and it failed to boot. I rebooted it, and it worked fine, ran prime95 for the past 4 hours; 0 warnings, 0 errors. Task manager reports both cores at 100%.

Core temps peak at 58 C, and all voltages are at stock - FSB ratio 1:1.


Do the crashes and boot-failure have anything to do with CPU stability? (Do real-life-applications differ from cpu torture tests in this aspect?)
Am I utilizing prime95 correctly?

I have no intention of raising it past 400x8, but 2 boot failures and in-game crashes worry me.
Thanks in advance for any input -
Click to expand...

There's a template somewhere to fill out so you can give us the details of your overclock. Plus, what mobo are you using?

Alot of users here can push to 3.2 at stock but i have only pushed to 3ghz 375x8. I don't do long prime95 tests. Instead, I run crysis, and i compress divx videos as a test. If either crash, my system is not stable.

It sounds to me you need to add a bit of voltage, but that means heat, which means you need a better cooler.

I don't know why u kept pushing your cpu speed after you crashed. I'd stop at the point where the crashing stopped as well.
 
Are you running the beta version of Prime 95, that runs 2 at once, or are you using the original version that only runs one thread at a time?
Click to expand...
Yup, both cores running 100%

I have found it best to do other things on your pc while priming
Click to expand...
I've been using my computer with prime on, half the time. Web browsing for the most part, but also 1 hour of photoshop, and a couple runs through superPi 1m.

I run 475 * 8, but I'm on water and everything is overvolted. I can run 500 x 8 but its not stable unless the fans are on full blast which gets annoying. Before I put the water in I was running 425 stable with the stock intel cooler. Mobo is an Asus IP35.
Click to expand...
That's one of the reasons I stopped at 3.2. My cpu fan is actually running at 40%, and I turned off all my case fans except one of them, and temperatures rose about 2 degrees. imo, the quietness is worth the extra 2 degrees. Also, I don't want to push my ram past their rated speeds.


Absolutely, the best example is your in-game crashes, Prime etc. do not push the ton of video information from the cpu, through the MCH/Northbridge, to the video card like a game does. Add cooling to your MCH/Northbridge.
Click to expand...

Would it help to re-sit the northbridge w/ AS5?
I actually lowered my cpu fan to 40%(variable.. 70% once it hits 60), and shut off all but one of my case fans, to lower the sound output. This only raised my temps 2 degrees, and I think it was well worth it. I don't want to have to put a fan on the northbridge if I don't have to.



There's a template somewhere to fill out so you can give us the details of your overclock. Plus, what mobo are you using?

It sounds to me you need to add a bit of voltage, but that means heat, which means you need a better cooler.

I don't know why u kept pushing your cpu speed after you crashed. I'd stop at the point where the crashing stopped as well.
Click to expand...

I'm bought an ip35-e since it was the cheapest from newegg with respectable performance.
I knew that the cnps9500 wasn't a really great cooler(performance-wise) when I purchased it, but refurbished(same thing as brand news, with replaced fans), go for really cheap on ebay, and i liked the fact that it had a low center of gravity... weight...
I see areas on hsf where the cpu rubbed against it(around the edges), so i want to lap the cpu/hsf when I have the time, cause I'm worried about the stability during the summer.

I kept pushing my cpu speed because I couldn't re-produce any of the reboots, and was told that the crysis crashes were software-based instabilities, since it only crashed in-between setting changes.
 
Would it help to re-sit the northbridge w/ AS5?
Click to expand...

It would not hurt, and replace the push pins with real hardware.

Simple modifications like this, done with care, vastly improve the reliability and stability of almost every board and cost almost nothing but the time taken to do it right. Yes, it is a pain if you already mounted the board in the case. Consider it as a penance for OCing without first doing proper board thermal prep.

I turned off all my case fans except one of them, and temperatures rose about 2 degrees. imo, the quietness is worth the extra 2 degrees.
Click to expand...

As long as your realize you just increased the operating temp of ALL the components on the board and have no easy way of measuring the effect of reduced case airflow on them would be. In particular I would be concerned about the cpu power regulation circuity around the cpu socket in an OCed situation due to the reported slow cpu fan speeds you mentioned above. Coupled with low case air turnover/airflow, long term reliability and stability under heavy load could be an issue (or not). Load the CPU and let it run an hour, while running quickly pull off the side panel and stick a finger on the MOSFET heatsink just between the cpu socket and the back IO plate. If you can count to 10 with your finger pressed on it, you are good. If you can count to five, consider turning on another fan, if you cant make it to 5 I would recommend more airflow in that area. BTW you are using the mosfet heatsink just to get a general idea of the cooling in that area, the MOSFETs themselves are rated for much higher temps than the cpu, it is the power filitering capacitors in the circuit nearby that I would really be worrying about but there is not a good way of getting a finger or temp probe on them unless you have one of those fancy IR thermometer "guns". Also if you do not keep boards long, less than 2 years, the concern is not as great. I tend to keep boards around 4+ years while OCing the heck out of them, so I tend to be focused on thermals, you know your situation better than I do.
 
I appreciate you bringing that to my attention..
I lowered the cpu fan, because no matter what speed they were running, the fins, even at the base of the heatpipes felt no warmer than my case itself(after all, the cpu is only cooling the hsf, not the cpu directly).
I do plan on getting better fans however, that'll push air with less noise. My cpu fan blows directly on the MOSFET, and it feels only slightly warm(like a chair someone's been sitting on).. is it normal for the mosfet to be that cool?... However, my northbridge seems to be getting extremely hot.. like a metal car, that's been roasting under the sun.
 
Wow... I just replaced the thermal pad on the northbridge, ...
Before, I could easily leave my finger on the heatsink for hours, and just have it feel "hot"..

After I replaced the pads with as5, I burnt my finger on them maybe 2 seconds after touching them..

If I'm thinking this right, heatsink getting hotter just means that it's working to relieve the ic of heat, right? I just need more airflow in that area now..
If that's the case, I think I'll re-sit the mosfet and southbridge as well...
Unless.. heatsink getting hotter = bad?
 
The heatsink getting hotter means that it is taking more heat from the northbridge which isn't a bad thing, a fan would be a nice addition to that heatsink
 
ymic said:
I run 475 * 8, but I'm on water and everything is overvolted. I can run 500 x 8 but its not stable unless the fans are on full blast which gets annoying. Before I put the water in I was running 425 stable with the stock intel cooler. Mobo is an Asus IP35.
Click to expand...

At least your E6750 can hit 475. Mine + IP35-E is being stubborn, won't go beyond 3.4 Ghz. :mad:

Awesome recommendations here, too. Let's keep this thread going.
 
You must log in or register to reply here.
Back
Top