Overclocking q6600 lockup

[J9Darkwing]

Salute and howdy!
I'm having a bit of a problem. . .
Here are my system specs:
Q6600 G0 stepping
Asus P5N32 e sli (680i)
4gig Crucial Ballistix ddr2 ram
x2 BFG 8800GT in sli
850w Rosewill PSU
Vista 64bit

Ok, I've been running this system overclocked at 2.8ghz for about two months since the only cpu fan I had was stock. I had NO problems whatsoever. I never tried to run it higher than that but assumed it would clock higher from the posts I've seen in the forums. Friday I bought my second 8800GT and an Arctic Square cpu fan. I installed them and left my clock at 2.8 and started to play crysis and my puter hard locked. Then it wouldn't boot right until I dropped my clock back down to stock 2.4 ghz.
I've had no success getting it to overclock now at all. My coretemp shows me to be in the 30s and 40s running in the low 50's when loaded with prime95. These temps are at 2.4ghz.
Someone please help. I'd really like to get it to 3.2 or 3.4ghz stable but as it stands I can't even run it at 2.8. Best I can get without locking up is 2.6.
As a side note I've taken out the second video card and still can't OC it. Can't imagine that a new and better cpu cooler would be the problem. I have 7 120mm fans so there's plenty of air moving through the box.
Here are the stock voltages shown in the voltage monitor section of my bios. (I've tried increasing some of them with no success.)
vcore 1.2
mem 1.88
1.2V HT 1.26
NB 1.24
SB 1.53
CPU VTT 1.26
DDR2 Term .96
3.3v 3.31
5V 4.86
12V 11.84

Thanks for any help you can give. If someone with a p5n32 e sli and a q6600 that's oc'd to 3.2 or 3.4 has a saved bios profile I may have that would be most appreciated.

 

[BonzaiDuck]

You need to find a thread running last week or the week before started by "coolamasta" -- an Aussie member here, with contributions by me and another member "Idontcare."

It was about over-clocking a Q6600 on the ASUS Striker Extreme (680i).

Pay attention to what I discovered about BIOS revisions, do your homework in finding which BIOS your board shipped with, and whether you need to flash regressively to an earlier BIOS.

Not kidding here, either.

 

[krnmastersgt]

You should probably get new cooling, the added temps from the 8800 might have caused the air your stock cooler took in too warm, does your case have a lot of airflow? You should have at least 1 exhaust fan, preferably 120mm, mine moves a lot of air.

 

[BonzaiDuck]

It could be something else, but what he's saying describes a symptom which three of us had, and three of us solved.

 

[BonzaiDuck]

What's the max recommended voltage on your memory, Darkwing?

 

[BonzaiDuck]

Nnnn--Kay. 2.2V Crucial Ballistix. got it.

If you're running a 1:1 CPU to RAm ratio, what are your timings?

Listen. I'm tuckered out here, and starting to get careless. If you had the board for a couple months, when was the BIOS released? You have a G0 stepping -- granted. Did you tighten your Crucial latencies, or leave them stock? If they were tightened, and "1:1", then I'd suspect VDIMM should be about 2.125V to get to 3.0 Ghz.

My board is a different 680i board, but I think these boards are very similar:

3.0 Ghz:
1.319 to 1.32+ Vcore
1.30 to 1.35V 1.2VHT
1.35v NBCore
1.35V CPU_VTT
[1.30V might be sufficient for these latter, and many have good luck with "auto", but they don't even begin to increase at Auto 1.25 until the second notch, and I "fix" them.]

If Crucial Ballistix DDR2-800 or DDR2-1000, THEN:

3,3,3,8, 2T, tRC = 11 should work at 1,333 FSB and 1:1 with memory as "DDR2-667" with VDIMM/memory-voltage at 2.125V to 2.150V

Now -- with my Striker board, you cannot get beyond 2.8 Ghz this way with BIOS versions released after July 2008. With BIOS versions up to v.1303 in July, 2008, you can do this laying on your back, and 3.15 Ghz is also pretty easy. 3.2 isn't much harder. With a G0 stepping, they ALL should be easy.

 

[J9Darkwing]

I'll search for that thread you mention BonzaiDuck. My max voltage on my mem is 2.2v.
Krnmastersgt, I added an Asus Arctic Square CPU cooler plus I already have 5 120mm fans mounted on my case. (I said seven in first post but was miscount) Plenty of airflow. Plus the overclocking problem persist with only one card installed.
My current bios version is 1205. If I remember right it shipped with 11XX. 1104? Something like that.

 

[J9Darkwing]

I appreciate your help BD. I haven't messed with my timings at all but I do know that the timings are more lax than the recommended Max. Something like 4,4,4,4 2T at 12 and they're running at like 5,5,5,5,2 at 16. <this is just an example of how the timings are lax and not actual. Something I'll need to check to be certain.

 

[Mondoman]

J9, remember that your Arctic Square is not really "better" than the stock cooler, it's different. The stock cooler is a "downdraft" style, blowing air down and out over neighboring MB components. This secondary cooling of nearby components is an intentional part of the Intel design, but not all MBs really need it. In your case, your MB has two big radiators right next to the CPU, so it DOES use it. Note that your Arctic Square can at most cool only one radiator or the other (depending on how it is oriented), but not both.
Thus, your new cooler works better than the OEM Intel for its primary cooling function, but worse for its secondary cooling. Since your north bridge and/or south bridge and/or PWM components are now not being cooled as well, that could very well result in instability at higher clock speeds and thus inability to OC as high. My guess is that if you reinstall the stock cooler, you will "magically" be able to reach faster overclocks again. Another option is to add a small extra fan blowing on the radiator that doesn't get cooled by the Arctic Square.

 

[xj0hnx]

Have you cleared CMOS? Did you check that when your computer reset to defaults to check through all the settings again, and make sure they are right? Is it running linked maybe? RAM voltage/timings?

 

[J9Darkwing]

I understand what your saying Mondoman and it makes sense. I have another small fan designed for those radiators and will look into installing it.
XJohnx, I have SOO been over the whole clearing the cmos issue as well as the timings. I've been doing most of my OCing with the memory unlinked at 800mhz running at 5,5,5,5,16 2T with a 2.2v which is the manufactures max recommended voltage. 4,4,4,4,12 2T are the manufacturers timing settings so they are running lax.

Here's a follow up of what I did last night to get it to run at 3.37ghz but still unstable. Perhaps you guys can help me get it stable:

NosBoost300 said: lol these asus boards are bad at oc'ing quad cores, i've owned the p5n32-e sli plus, and the regular p5n32-e sli..

ii've taken mine up to 3.337 ghz... takes alot of juice on this mobo

my vcore was 1.45
mem 2.4
1.2ht was 1.45
nb 1.40
sb 1.65
cpu vtt 1.5

My response was: Didn't even change the fsb and it locked up. Put in the suggested voltages and left the clock the same. Rebooted and it hung on post.

NosBoost300 said: this board has alot of fsb holes, set it to 1500 and run the settings i gave.

My response was: I got it to boot up using the 1500. Tried to run superpi but it wouldn't calculate. Says something about it 'not in round' or something like that. Also, it hung on reboot. But whoaa, proves to me that it can do it. I was astounded 3.37ghz. Now if I can just get it to do that and be stable I'll be one happy duck.

That's as far as I got last night and haven't heard back from NosBoost300.
Going to look for that little radiator fan now. Any other help getting stable would be most appreciated.

 

[VirtualLarry]

You sure it's not the PSU? Adding a video card would put a greater load on the PSU, thus perhaps limiting the overclock.

 

[Markfw]

850 watt pcu, but its a rosewill.....I wouldn;t trust my system to that.

 

[J9Darkwing]

This is my second rosewill and they have both given top shelf performance as far as I could tell. Plus, I bought this PSU with the intent of buying to 8800GTXs so it has four 12 Sli power plugs. Since I bought 8800 GTs instead it's twice what I currently need.

 

[john3850]

I have evga 680i e6600 @3650 sli,4gig Crucial Ballistix pc up 14 months.
Crucial Ballistix @ VDIMM/memory-voltage of 2.2 and up fails mem test for me.
BonzaiDuck said it right on"DDR2-667" with VDIMM/memory-voltage at 2.125V to 2.150V .
Blew 2 sticks @2.3 rma me 8500
Some 680i doesnt always like 4 sticks in mem test.
I start with only 2 sticks till stable then add last 2.
No mater what timming i use my Fps stay the same.
pi 1m=high 13 when tweaked good.

 

[J9Darkwing]

I pulled two sticks and lowered the voltage down to 2.12 with no luck. Swapped the two I pulled with the two I left in with identical results. Considering most of my oc's are unlinked with mem at 800 with lax timings, confirmed by cpuz, I'm fairly certain the problem is not with the ram. Here's the mfg specs DDR2 PC2-6400 ? 4-4-4-12 but I've been running them lax at 5-5-5-16. Crucial says they are 'sli ready' but if I enable that function in the bios the system won't post.

 

[J9Darkwing]

By Darwin I swear (I'm an athiest) I'll never buy another Asus motherboard for as long as I live!

 

[Rubycon]

Originally posted by: J9Darkwing
By Darwin I swear (I'm an athiest) I'll never buy another Asus motherboard for as long as I live!

Don't swear off the game, blame it on the player. I've had great luck with Asus boards with Intel chipsets. NV OTOH, bleh!

 

[BonzaiDuck]

NV is tricky, given the "concept" of the wide-open PCI_E lanes and chipset not designed by Intel.

Darkwing, did you check to see when those BIOS versions were released? I'm imagining the possibility that the one's you're using are pre-July'07. If so, they should work, or you could upgrade to any BIOS version released prior to end-July.

Now I see -- I tHINK-- you're making some progress. But it's my opinion, unless the quad core is a G0, pushing beyond 3.3 is moving toward "extreme."

Someone else you quoted to speak of going to 3.34 noted a VCORE of 1.45V. This seems more like B3-stepping behavior. I was pretty sure -- through posts I've read, that you can get farther on less with the G0 stepping. If not, you're doing something wrong with it.

I stopped pushing when my VCORE was at 1.42V or just shy of it -- 3.2 Ghz. I was almost planning to graph the VCORE versus FSB/CPU-speed settings and load temperatures for stable over-clocks, but I thought I saw an exponential function or parabola in there.

With the E6600, I got to nearly 3.4Ghz, 375 Mhz CPU-host-frequency 1:1 on my Striker with voltage around 1.44V. In fact, I think I got to 3.3 at 1.4125V. Need to check my notes.

I'd be antsier about voltage increases on the quad-cores, because you'll get disproportionately more thermal energy released. You've doubled the cores, and the voltage increase generates an increase of temperature proportional to the square of the voltage.

That's why I'm keeping my overclock setting limit at 3.15Ghz until I lap the processor. Or lap a replacement G0, or whatever I use to upgrade this puppy. At 3.2 and room-ambient 79F, I was able to keep the PRIME load temperature below 65C on a B3 stepping. But I like the more modest setting, even though there was no instabiility at 3.2. Or -- look at it this way. I can go to 3.0Ghz with 1.32V = VCORE < 1.35V INtel max spec. I can go to 3.13V with VCORE (idle) at 1.378V. Getting to 3.15, I need VCORE between 1.41625 and 1.4188V. I could get rock-solid stability for 14 hours terminated with 0 err, 0 warns in PRIME95 at 3.2 Ghz and 1.4188V (go figure . . )

You'll all tell me I'm nuts, but I'm going to watch the prices on the Penryn dual-cores to put in a Striker board. After I flash in v.1305 or higher, that is. I THINK . . . . that it will probably work fine. We'll see.

[Shee-ut. Evul Kneevul did more insane things than that . . .]

 

[J9Darkwing]

I'm sooo sick of seeing superpi tell me 'Not in round' or 'Not convergent in SQR'. Grrrrrrr. I'm overclocked to 3.37 and the first thing I do is run a 1meg superpi and no matter what I do I get the same result. Here's my setup now
FSB 1500
Mem in synch mode at 750mhz with loose timings
vcore 1.45v
mem 2.2v
1.2HT at 1.45
NB at 1.4
SB at 1.65
CPU VTT at 1.5

My highest core temp is 48 and my lowest is 44. Any thoughts to help me get stable?

 

[J9Darkwing]

I flashed my bios last night from 1205 to 1303. . .
Flashing my bios fried my bios. Called Asus and after literally 45 minutes waiting for tech support I got a nice lady who is sending me a new bios chip. Should be here tomorrow.
I was using bios version 1205 and tried to flash to 1303. When I talked to her on the phone she said both versions are betas and asked which one I wanted. I said, uhhhh 1104? lol
Now I'm surfing the web on a screaming 433mhz celeron with 128meg of ram. Bet y'all are jealous of my 'rig' now. lol
It's an odd sensation to just be typing and have to wait for the computer to catch up.

 

[VirtualLarry]

Originally posted by: J9Darkwing
Here's the mfg specs DDR2 PC2-6400 ? 4-4-4-12 but I've been running them lax at 5-5-5-16. Crucial says they are 'sli ready' but if I enable that function in the bios the system won't post.

There's a specific setting for "SLI memory" in the BIOS??? I wonder what that does, since normally dual-channel memory is essentially the same thing as SLI (two running in parallel).

 

[krnmastersgt]

Originally posted by: VirtualLarry

Originally posted by: J9Darkwing
Here's the mfg specs DDR2 PC2-6400 ? 4-4-4-12 but I've been running them lax at 5-5-5-16. Crucial says they are 'sli ready' but if I enable that function in the bios the system won't post.

There's a specific setting for "SLI memory" in the BIOS??? I wonder what that does, since normally dual-channel memory is essentially the same thing as SLI (two running in parallel).

It isn't SLI memory, it's rated as SLI Ready which basically gives you easier functionality on nV chipset mobos, I believe it OCs the memory on it's own when it's enabled to what it thinks is a good setting but I always just put it on 'Expert' and manual clock it from there. Still haven't found the time to properly OC my system with all the stress testing I need to since I have classes + tests this week.
Btw the EPP memory in my sig is better known as SLI Ready memory o_o

 

[J9Darkwing]

Got my new bios chip in this evening, got it installed and overclocked my system to 3.01ghz by doing nothing more than changing the fsb and leaving the memory linked. Didn't have much luck at any fsb over that but haven't felt like messing with it much this evening.
As my original post stated I couldn't get it to overclock to 2.8 without some kind of problems. With the new 1103 bios chip I've been at 3.01ghz running prime95 for three hours and the highest my temps have gotten were 66c. Most of the time the temp stays between 58 and 63c. Also, I have my case fans on Qfan which keeps them nice and quiet. I'm confident that if I get the computer up to 3.2 or 3.4ghz and turn my fans up to 60 or 70% I'll be able to keep my temps down.