3DMark05 crashing at stock speed

[Cyberparasite]

warranty is at least a year. sounds like a decent oc.

[naddie]

Thanks. That voltage drop saw a 40*C under load according to MBM5 - it says 23*C idle. This is out of the case, and I'm dubious of the temperature, seeing how others report that their Expert is up to 10*C lower than what they recorded the temps are at with a probe. But the area the base of the HS is barely warm to the touch.

 

The NB is the only real issue I have. Idles at 43*C. Around 46*C under load (an all outside of case).

 

The temps may what cause the stability issuses once inside a Super Lanboy case.

[naddie]

BUPKUS! NOTHING! NADDA! Prime errored within 5 hours on the 2nd core. The other one is stll going (it's about 9 horus now).

 

Looks like it really DOES need the voltage increase afterall.

[ЪngêR&#33]

mine is priming on 1.55 volts for 2.5hz right now.

 

also, what, of yours, needed more volts? the chipset?

[naddie]

The CPU it seems. I am afraid to bump the voltage up on the NB because it's so hot already. Not sure if it's suppose to get around 50*C under load. I did the TIM upgrade as well as took off that silver ring to hope to keep the stock HSF (it's so quiet!).

 

The video card is the only loud part in my setup (eVGA GT series fan and GPU freq is stuck on 100% even in 2D it seems).

 

I also added a lil more voltage to the CPU. It was at 1.6v before but did beautifully. I tried 1.55v and it crashed, so I tried "the middle" - 1.5v x 104.8% = 1.572v.

 

I am SOOO close to keeping the entire thing stable! I can taste it! Prime, then 3Dmark05 is my regimine for now, since it seems that these are the two that will fail first in my gammet of tests.

 

Thanks everyone for reading all this crap, lol. My sanity hopefully will return, along with my COD2 and Q4 skills.

[naddie]

Nope - had to bump it back to 1.6v to keep Prime stable.

 

I think I solved the crashing of 3DMark05 - I not only removed the silver right around the NB HSF, but also re-TIMed it with Artic Silver Ceramic. I also made sure the NB Fan runs at 100% at 48*C instead of the stock setting of 55*C. I can hear the fan spin up and down sometimes, but it keeps the NB from going over 48*C.

 

Currently 6 hrs into 3DMark05 without issues. It's still going, and hoopefully tomorrow morning I still see it running.

 

Also, I just bought one of those PCI card slot fans that sucks air out and blows it out of a PCI slot. I hope that will add additional air flow in that lower area.

 

Right now, I have negative pressure in the case (more air getting sucked out than air coming in). I can tell because when I removed the Super Lanboy's screw compartment, I put a piece of paper near the resulting hole - the paper wanted to go IN the hole.

 

Is negative pressure better than possitive? I can equalize it better by putting a slightly faster (more cfm) fan in the front. This will also keep me happy becaus I won't have to see the stock blue LED fan that came with it.

[wevsspot]

Assuming the ambient air temperature going into your case is nice and cool, the more the better. Your paper test is a neat trick, and if it's trying to suck the paper into the case, you've probably got plenty of intake air to work with.

[naddie]

Nah, it means that I have plenty of EXHAUST air going out of the PC. This is why air is trying to suck in an open hole without any fan assistance. Not sure if adding the PCI slot is going to help because of this newly found fact.

 

I will try to upgrade the front Antec blue LED fan with a Vantec Stealth fan. They seem to run at 1500 RPM, sucks air at 43 cfm, and is about 23db. And best of all, NO LEDS!

 

Of course my local Frys just ran out of the Vantecs i was looking for.

[naddie]

Okay, weird - 05 works now, but 03 crashes. >_<

 

Chipset temps are 48*C as far as I can tell (MBM5), but after a while, it crashes. I got the BIOS to turn on the fan at 100% all the time (noisey 8k RPM motor!), but it still gets that hot after a while.

 

How hot can the chipset get while being in the safe zone?

 

I am at the end of my ropes here. Its been 3 weeks or longer and not having a working PC. I wonder if I can RMA it for a non-Expert, non-UT (so the price is the same) SLI-DR. Not sure if I want to deal with the Expert anymore (yes, I am taking my word back, lol).

 

I don't think there is anything wrong with the board technically - just the placement of the NF4 chipset during the design stage.

 

What can I do in this situation, DFI techs?

[naddie]

Well, okay - after an extensive seach, I see Angry saying it is okay to go up to 60*C with chipset. SO, maybe it's not getting enough voltage? It is currently stock 1.52v.

 

New question (did a quick search found nothing, have to go to work now): what is the safe over voltage for the chipset?

 

sorry so many Qs. this is drving me ntus.

[wevsspot]

You are speaking of the NF4 chipset right Naddie?

 

If so, 1.6v is very safe, and I've ran two different NF4 chipsets at voltages of 1.7v for six months on one rig and three months on another. No problems so far.

 

Please, remember, overclocking results your mileage may vary. Your rig just might not want to o/c as far as some others have pushed theirs. I always try to determine four things before I even start overclocking;

 

1. What do I gain by overclocking?

2. What is the maximum voltage I'm willing to put through the components?

3. What is the maximum temperatures I'm willing to let those components run at day in and day out?

4. What is my maximum financial risk tolerance for the component I'm overclocking.

 

My whole point being, find a spot thats toaster stable based on your stress testing, benchmarks and maybe some of the considerations I noted above.

[naddie]

Thanks for the safe voltage. I'll give it a shot.

 

I plan on raising it slowly until I hit 1.6 or even 1.7 (if need be), and of course, I will also be monitoring temps.

 

How much did the temperature rised with your voltage increase?

 

Since according to A_G, the chipset can go as high as 60*C safely, I will keep that in mind while I am trying this voltage experiment.

 

Does the LDT voltage also affect stability? If so, I may try to play with that setting after seeing if the chipset voltage helps.

 

Thanks.