Q6600 overclock impass

[MattHB]

Hi folks..

Ive had my rig upto about 2.9ghz for a while now, and with idle core temps at around mid 40's I thought I would try to push it up a little more. Ive now reached 3.15ghz with a vcore of about 1.256v. Stesses fine on prime with cores upto about 60 (15 min blend mode).

If I try to go any further, the system will hang at idle, before I even get to stress test with prime. It doesnt seem to make any difference when I increase vcore, the system will still hang in either windows startup, or on desktop while loading background apps. The fact that it doesnt even get as far as stable idle suggests to me that it cant be over heating, as core0 (my hottest) is only sitting at 48C.

Im fairly new to this, and have only ever really tried nominal clocking, so my experience is very limited. I do understand the basics though, but Im aware that there are a lot of variables that can cause an overclock to fail.

ram is unlinked, sitting at 810 with a stock voltage of 1.8v. speedstep and all the other suggested intel 'features' are turned off as per the suggestions in GraySky's very good sticky.

Can anyone suggest anything that I could try to push things over this limit? It may be of course that Ive reached the limit for this cpu, but if you need any more info, I would be happy to provide.

thanks in advance

system details : antec 900, Q6600, P5N32-SLI, 8GB OCZ 6400, Zalman 9500, 2x 8800GTX, Corsair HX850w, ZM-RC1000

 

[jsc]

Practically every Q6600 should be able to reach 3.3 GHz in a decent case with a good cooler. The Zalman 9500 is not a particularly good cooler by today's standards. Even so, you should be able to reach 3.3 GHz.

Your system is hanging because of your memory settings.

HOWTO: Overclock C2Q (Quads) and C2D (Duals) - Guide v1.6.1
http://www.tomshardware.com/forum/240001-29-howto-overclock-quads-duals-guide

Next stop should be a guide for your particular motherboard. Google is your friend.

Go through the guides. Then go into the BIOS and change the System Memory Multiplier (or whatever your BIOS calls it) from AUTO to 2.00, 2.00B, or 2.00D - whichever you need to set the Memory Frequency to twice the FSB. Then when you increase the FSB, the memory clock will rise in step with it. If you are at 3.0 GHz (333 MHz X 9), your memory clock should be at 667 MHz.

Keep your core temperatures under 70C. and your core voltage under 1.45 volts - whichever you reach first.

 

[MattHB]

thanks for that jsc.

Ive reread the guide (as id been scouring it before anyway) and made the changes you suggested. I found a specialist guide for my mobo a while ago, which Ive been referring to, but its memory section was hazy to say the least, hence me making the unlinking mistake.

With those settings, I get to 3.17Ghz with some quite nice core temps, (60C in the first blend test), but anything further than that I get the same desktop hang.

One thing that I have noticed, although all my RAM is OCZ 6400, CPUZ reads the slightly older 2 sticks as 5300, even though is has a 6400 part number (confirmed by the OCZ website. any idea why this would be? and may be this is part of the issue?

To be honest, 3.17 is more than fine for me, its just out of curiosity that Id like to get it figured out. Thanks for your help.

edit: I just realised I have more to read on the guide. Sorry Im being quite blonde today!

 

[MattHB]

after a lot of reading I made a start at things again today. Firstly, I took off, cleaned up and reseated my zalman cooler.. I dont have the cash to swap it out at the mo, so itll have to do. but what a great difference. 13C lower core temps during smFTT tests in prime, so very happy with that. I also turned the cooler by 90 degrees so it pushes warm air directly at the rear exhaust fan, rather than upwards to the overhead exhaust, this means that hopefully Im drawing in cooler air from the front of the case, instead of warm air from above the 2 8800GTX's.

all things considered, Im still at 3.17Ghz now, any more than that and I get random desktop/boot crashes even though things test fine in prime (blend and small FTT tests for over 20 mins with no errors, but then the system crashes just after stopping the test), although core temps are much lower on the small FTT test now Ive sorted the heatsink.

I still dont know whats causing this, although I wonder if 2 of my 4 memory modules are faulty as they read in CPUz as 5300, not the 6400 that theyre supposed to be (confirmed by part number). Ram frequency is now in sync mode, which seems to be more stable. So by trying to maintain a 350 FSB its pushing the memory too far as it thinks its only rated to 333?? could this be right? Ive put a post up on the OCZ support forum to try to find out whats up with the modules, so will update if/when they answer.

anyway, would appreciate any feedback on this crashing error, it makes no difference if I push up the vcore or the NBcore values and it happens when cpu load is minimal, so cant be heat related.

 

[jsc]

I still dont know whats causing this, although I wonder if 2 of my 4 memory modules are faulty as they read in CPUz as 5300, not the 6400 that theyre supposed to be (confirmed by part number). Ram frequency is now in sync mode, which seems to be more stable. So by trying to maintain a 350 FSB its pushing the memory too far as it thinks its only rated to 333?? could this be right? Ive put a post up on the OCZ support forum to try to find out whats up with the modules, so will update if/when they answer.

CPUz reads the contents of the SPD chip that was factory programmed. As far as the factory - meaning assembly line - is concerned, those are DDR2-667 RAM sticks. They may have tested out at DDR2-800, but the SPD chips do not lie. OCZ has a pretty good reputation. I'm surprised they did something like this.

You can try two things. You can relax the timings from 4-4-4-12 to 5-5-5-15 or 5's to 6's and see if it's stable. You can also try increasing the RAM voltage to 2.2 volts if it's not there already.

Personally, I'd try the RAM voltage first. Then if it's stable, decrease it until you find the point where it's still stable.

 

[MattHB]

thanks jsc, Ill try that. timings are at 5, 5, 5, 18 which are suggested stock at the moment. voltage is currently 1.8, which is suggested stock. Ill try upping it and coming down in steps like you suggest.

So basically, although I have 2 sticks sold to me as ddr2-6400, and the part number confirms that, they are actually only 5300? they have a life time warranty, so maybe I can swap them with OCZ.

Do you agree that my issues are more than likely the ram issue causing the desktop crashes? its the only thing I can come up with.

thanks for all your help.

 

[MattHB]

Ive just spotted something really odd in CPUz while taking a screenshot. Although my timings are set at 5, 5, 5, 18 in the BIOS, and on slots 1 and 2 it confirms this, on the 'faulty' 2 slots with the older part number, it reads it as 5, 5, 5, 15. How is this possible? surely when you set the ram timings in the BIOS it would be for all slots?

heres a screenshot of both the newer module and the older module in case there's something that I've missed.

old module - reporting not as expected


new module - reporting as per bios


many thanks.

 

[lothdk]

Ive just spotted something really odd in CPUz while taking a screenshot. Although my timings are set at 5, 5, 5, 18 in the BIOS, and on slots 1 and 2 it confirms this, on the 'faulty' 2 slots with the older part number, it reads it as 5, 5, 5, 15. How is this possible? surely when you set the ram timings in the BIOS it would be for all slots?

heres a screenshot of both the newer module and the older module in case there's something that I've missed.

What you are listing there is just the SPD timings, the actual settings appear on the Memory pane.

You might need to bump the northbridge voltage a tiny bit, but reread the sticky, and play around with the settings, as you said yourself, no two CPUs are the same.

 

[MattHB]

What you are listing there is just the SPD timings, the actual settings appear on the Memory pane.

You might need to bump the northbridge voltage a tiny bit, but reread the sticky, and play around with the settings, as you said yourself, no two CPUs are the same.

ah ok, I see what you mean.

so are the 2 older modules running at the right speed then?

 

[lothdk]

ah ok, I see what you mean.

so are the 2 older modules running at the right speed then?

All of your memory sticks should be running at the settings you have entered in BIOS, and those are the settings that appear in the Memory pane (just left of the SPD pane).

All you see on the SPD pane is the JEDEC standard coded into the ram that the BIOS will use to configure it, if you do not enter settings manually.

 

[MattHB]

thanks for all your help guys.

I understand the fact that CPUz is reading the chip info and that the bios will be controlling all 4 modules regardless of what the SPD info says, and controlling the memory modules the same across all 4 based on the settings Ive manually given them.

But what Im worried about is that the 2 modules that read wrong are ACTUALLY 5300, and therefore causing me issues when I clock up the FSB to over 333mhz. Its only when I go over this figure that I have the issue.. surely if the memory was rated at 400mhz Id not have these memory instability problems. therefore to remain stable at the 350mhz that Im aiming for I wouldnt be overclocking the memory based on a 1:1 ratio and I wouldnt need to increase its voltage as it would be running well within its limits.

Sorry if Im being a bit dim (no pun intended!), may be Ive understood the process wrongly.

 

[lothdk]

thanks for all your help guys.

I understand the fact that CPUz is reading the chip info and that the bios will be controlling all 4 modules regardless of what the SPD info says, and controlling the memory modules the same across all 4 based on the settings Ive manually given them.

But what Im worried about is that the 2 modules that read wrong are ACTUALLY 5300, and therefore causing me issues when I clock up the FSB to over 333mhz. Its only when I go over this figure that I have the issue.. surely if the memory was rated at 400mhz Id not have these memory instability problems. therefore to remain stable at the 350mhz that Im aiming for I wouldnt be overclocking the memory based on a 1:1 ratio and I wouldnt need to increase its voltage as it would be running well within its limits.

Sorry if Im being a bit dim (no pun intended!), may be Ive understood the process wrongly.

No indeed you are right, it does look like they are 667 MHz and not 800 MHz, if they are under warranty, I would try and get a new pair, this could very well be why you are having issues.

 

[MattHB]

No indeed you are right, it does look like they are 667 MHz and not 800 MHz, if they are under warranty, I would try and get a new pair, this could very well be why you are having issues.

cool, thanks for the confirmation. Ive raised a ticket with OCZ, so will await the reply and see what they say. Im more glad that I understand everything enough to come to that assumption!

many thanks.