OC Advice on my e6850 @ 3.75

[morikal]

Hi guys,

I'm trying to get my 6850 to 3.75 GHz (I have 1333 memory--the OCZ ReaperX HPC... I initially wanted a 3.6 GHz overclock, but figured why not go for the extra .15 so I can have my memory up at full speed with a 5:4 ratio)

Anyways, my system is:
Case: Coolermaster Stacker 830 Evolution
PSU: Realpower Pro 1000W
MB: Asus Striker II Extreme (refurbed... had a striker extreme and it broke, they didn't stock it anymore, so they sent me a refurbed striker II... its been fine so far)
CPU: Core 2 Duo e6850
CPU Heatsink/Fan: Thermalright Ultra 120 Extreme with a Scythe S-FLEX SFF21F 120mm fanRAM: 2 x 1GB OCZ ReaperX 1333 @ 6-5-5-20-1T (stock)
GPU: 2 x 8800GTX @ 600/900 (stock, BFG pre-overclocked)
HDs: 4x Seagate Barracuda 250 GB in RAID 0+1 (using the onboard controller)
Sound: Using onboard--couldn't get my SB fatality series whatnot card to work in here (wouldn't really fit all the way into its slot properly due to the case's back; was having weird issues with it before I figured that out)


I fiddled for a bit, and got stable with the following for 30 minutes in OCCT (I just started messing with it today... will prime 95 it overnight sometime this week; wanted to check that my voltages aren't too crazy first)

Bios Settings:
FSB @ 1666 (416.7 x 4)
RAM is at stock (using 5:4 ratio)
VCore is 1.475, shows up at 1.49 in CPUZ
PLL Voltage @ 1.6 (stock is 1.5)
VTT (FSB Termination) @ 1.36 (stock is 1.2)
NB Core @ 1.52 (stock is 1.3)
SB @ 1.5 (stock)

During the 30 minute run (which seems to run at 100% CPU usage for about 45 seconds, then dip down to 99.X (fluctuating) for about 5 seconds, then back up to 100%), my temps according to core temp maxed at 61C for both cores. I think I read that up to mid-60s is ok for completely maxed CPU? I'll have to see what prime 95 does to it... and keep the room temp pretty cool (I live in the northeast US, so thats not too much of a problem most of the year)

I am wondering if the above voltages would be ok for regular use? I do sometimes leave my machine on overnight (but do not usually need to). Its main use is gaming. I'm not too concerned about longevity--I've had this machine for about a year and a half (but it was out of commission for a good 4 or 5 months due to the motherboard breaking and me not having time to swap it out). I'm looking to get maybe another year of use out of it (more is just gravy). So, I know that running at higher voltages will wear stuff out faster, but am I likely to blow stuff out before a year is up?

My case currently has 3 exhaust fans (including the PSU) and 2 intakes for a negative pressure air cooled interior. I'm conflicted about whether to utilize the fan cage on the side of the case to add more intake fans (which would make it a positive pressure system, which isn't good for anything that isn't getting direct air...). My MB has that heatpipe passive cooling thing going on (with the CPU fan exhaust blowing over some fins). Is my NB going to be getting too hot with 1.52V running through it?

All that said, I am planning on lowering voltages once I verify that I'm prime95 stable overnight. (I've been going up in non-incremental jumps to find my 30 minute stable settings... I did google around first and saw other people with even higher voltages than I'm using, so didn't take the time to step up slowly...). Hopefully I'll be able to get some of those voltages down. Anyone have any advice on which to try lowering first?

Sorry for the long post. Summary:
- Are my temps likely ok? (ie: if prime95 gets it up to mid 60s?).
- Are my voltages too high for any use at all? (ie: is it probably ok to do an overnight, 12+ hour prime95 run at those voltages, on an air cooled system?)
- If so, are my voltages likely to kill any of my hardware in less than a year?
- Given that I upped a lot of my MB voltages, should I add 2 more side intake fans (which will make the whole case a postive pressure sytem, which I was trying to avoid... but I want to make sure the MB is sufficiently cooled)
- When doing stability tests, any advice on which of the listed voltages I should try lowering first, assuming that I am 12 hours prime95 stable at my current settings?


Thanks for any help and advice you guys can give me!

 

[cjl]

Temps seem fine, and the voltages look OK. I wouldn't worry overall. Make sure the NB doesn't overheat, and you should be set.

 

[morikal]

Was it necessary to raise the voltage or did you just do it for the heck of it?

Those processor should be able to hit 3.6 on stock voltage.

It was definitely necessary. Here is what I did (I wasn't the most patient I could be, but I'm not crazy )

1) Try OCing to 1400 FSB with stock everything. RAM wouldn't go the extra 66 MHz without crashing (I assume it was the RAM).

2) Try OCing to 1600 FSB with stock voltages, underclocked the RAM. Trouble booting.

3) Bumped up to 1666 FSB and started tweaking voltages. Upped VCore and NB a bit, tried again. Another fail to boot. Tried bumping voltages a bit more, including VTT. Got it into windows, but immediate failure in OCCT

4) Repeated voltage bumps for a bit, but at some point raised the bar to 1666 FSB.

5) Was failing after 3 minutes or so in OCCT with 1.45 VCore, 1.5 NB, 1.26 VTT, 1.5 PLL.

6) Bumped up to current settings, was stable for 30 minutes.


So, I'm not sure why I can't hit 3.6 with stock settings... and even if I'm stable at these voltages, as I said, I don't think they'll drop much.

To keep the NB cool, would you just toss a bunch of fans at it? Does speedfan's MB temp monitor the NB? I assume so... whats a safe temp for it to get up to?

 

[morikal]

Also, coretemp and speedfan are reporting different temps... coretemp is 5 degrees cooler on both cores than speedfan is... which one should I be keeping the temps @ or under 65 on? (Assuming 65 is a safish max temp for a prime 95 run)

And, how can I tell which temp is my motherboard? I see Temp1 in Speedfan as chip IT8718F, BUS ISA, Address $290. It is getting up to about 55 degrees during the first part of a prime 95 run (keeps going between 50 and 55 degrees). I couldn't find a good guide for matching speedfan temp labels to actual components.... any help would be greatly appreciated

EDIT: About 25 minutes into the Prime 95 run... that temp1 setting is now fluctuating between 54 and 61 degrees...

 

[morikal]

Hmm, turns out I'm not stable. I changed the ambient temp to 70F (not sure what that is in celcius) from about 62F. OCCT reports an error within about 20 minutes.

In the BIOS, my NB idle temp was 63C (but my MB idle temp was 35). I looked around for a little bit, but couldn't seem to find a guide for safe NB temps...

My next step is to run OCCT with no overclocking, just to make sure I don't get errors in it normally.

 

[morikal]

Ok, I'm stable for an hour in OCCT after bumping the NB to 1.54. I was stable in Prime 95 for an hour with the NB at 1.52

My temps do get kinda high... I don't know what the TJmax is for the e6850 (I've found many conflicting things, ranging from 80-100 TJmax depending on the stepping), but using realtemp I get up to around 30 less than TJmax (whatever it is).

So, if TJmax is 95, that would be getting up around 65C.

If it turns out that the NB voltage was limiting my OC, then I should be able to get the temps down more by dropping the CPU Vcore.

Once I get this all stabalized, I'm going to start in on my memory, then my video cards. I'm going for being able to play Fallout 3 with this OC (It runs almost completely smooth for me at stock, but I want it ALL smooth )

 

[jay2tall]

My buddy had an e6850 and then passed it to me. He has water cooling and I had a thermalright Ultra-120. Neither of us could get this past 3.6 completely stable. We had to increase the voltage WAY to high. He had a 780I mobo and I had a 975x mobo. He was able to get it to 3.8 I believe but was NOT totally stable. We ran some benches and game benches and honestly the difference is SO small it wasn't worth the risk of running the higher voltages.

We both since ditched that CPU and got an i7. WAY faster. It got to the point where the OC wasn't much of a benefit, but the switch to a totally different architecture was the way to go. (just a thought)

 

[jay2tall]

And a not. we needed about 1.45v to hit 3.6V no way did we get that on stock voltages. We needed 1.4 to hit 3.4GHz. There just wasn't alot of OC headroom.