Q6600 OCing. am I doing ok?

[MjWoNeR]

I recently received my new rig and started overclocking with the help from the Overclocking Guide (thanks graysky!)
Specs:
■Q6600 G0
■Zalman CNPS9700LED 110mm
■Gigabyte P35-DS4 v1.1
■Corsair XMS2 Extreme 6400c4 800Mhz 2x1024MB
■Corsair 620W HX PSU
■BFG 8800GT OC 512MB
■Lian-Li PC7 Scandinavian Edition V3

But since I am a noob I would like some feedback from you guys, you know just in case I am doing something that I really shouldn't!!
My goal was a stable, permanent and cool 3Ghz clock so I am all set and starting prime95 for the 24hour marathon.

After two days of adjusting this is what my settings are:
■Multiplier and FSB 9x334
■Memory multiplier 2.5 that gives 835Mhz at the RAM and timings 4-4-4-12
■Upped +0.10v the RAM which solved the BSOD issue.
■CPU Voltage 1.3125v

When I had my voltage settings on automatic I was getting high CPU temps averaging at about 55c-58c and maximum 61c after 10+ hours of prime95, so moving to manual it seems to have dropped the temperatures quite enough averaging around 51c-53c and maximum 55c.


Questions:
■Although I have set the CPU voltage at 1.3125, CPU-Z and Everest shows 1.264v. Is this normal?
■Are the temperatures ok?
■And finally, am I doing everything right? Is there something I should have done differently?

 

[cnumartyr]

It sounds great man. Read the guide from Comp at the top of the screen for some info on temps.. however those look fine for a G0.

That is pretty normal. My board drops the vCore 111 mV from BIOS setting to Windows, it then drops another 8 mV under load.

Seems like you did fine, if you want to push it further you have some head room as well.. I would say 400x8 would be within your grasp with 1:1 memory if you wanted to.

 

[MjWoNeR]

Ok 3 out of 4 cores just failed in prime95 so I upped the voltage from 1.3125v to 1.31875v. This is what I should have done right?

And according to Everest the DIMM voltage is 1.94v.


Thanks for the help man, appreciated!
Where is that guide again you are referring? lol

 

[cnumartyr]

That's exactly what you do.

Mine failed 3.6 GHz after 2.5 hours at 1.390 vCore. They passed 10 hours (good enough for me) at 1.4 vCore. It's a finicky thing.

Comp's Guide

He'll probably be nice enough to copy and paste the section pertinent to your G0 Q6600, but I'm lazy..

I wouldn't bother with 24 hours. 8 hours of Small FFT with 10-12 hours of blend is good enough (imo) for a stable system.

 

[MjWoNeR]

No problem


The guide says:
Scale 2: Quad

Q6x00: Tcase Max 71c, G0 Stepping, Tjunction Max 100c, Vcore Default 1.372, TDP 95w, Delta 10c

-Tcase/Tjunction-
--70--/--80--80--80--80-- Hot
--65--/--75--75--75--75-- Warm
--60--/--70--70--70--70-- Safe
--25--/--35--35--35--35-- Cool


Does this mean that anything bellow 60 is considered safe?

 

[cnumartyr]

Well.. there are two different temperatures.

Under 60 CPU Temp (Tcase) normally reported as CPU in the BIOS or monitoring software is safe.

Under 70 Core Temp (Tjunction) is considered safe as reported by CoreTemp or the latest Beta of SpeedFan.

 

[MjWoNeR]

The maximum CoreTemp temp is 62 at the moment but its usually lower than that.

I had to raise the voltage to 1.38125v trying to get a stable blend test with prime95 for a few hours.


Godz, overclocking takes time..

 

[cnumartyr]

CPU Stability it tested with Small FFTs, blend checks overall system stability.

It might run the temps up another 3-4C, just so you know.

 

[MjWoNeR]

I see, so then I first test the CPU with Small FFTs, and after I get a nice and stable result I move on to RAM configurations and blend tests for overall stability?

 

[cnumartyr]

I see, so then I first test the CPU with Small FFTs, and after a get a nice and stable result I move on to RAM configurations and blend tests for overall stability?

Yep!

My rule of thumb is 8 hours Small FFT then 10 hours Blend.

In general though I can run 2 hours of Small FFT and get a rounding error and then run the OS for 3-4 days and not notice anything.. I just like knowing it's rock solid for my 24/7 stuff.

That being said, make sure you keep Round Off Checking enabled in the Advanced tab for Prime95v255.

 

[MjWoNeR]

Well your rule sounds quite reasonable, and I am going to follow it and extend the length of the tests a bit too.

Thanks for all the help man, you solved some very important questions I had and helped understand how things work more properly.

 

[cnumartyr]

No problem, if you have any more questions don't be afraid to ask!

 

[MjWoNeR]

Now I am running prime95 again but this time with small FFT's, 9x333 and 1.2625v
It's been a few hours and I'll probably let it work overnight since it seems quite stable and the temps have never exceeded 55c. If everything goes well tomorrow I will start playing with the memory.

CPU-Z in the "Timings Table" shows that the memory (Corsair XMS2Extreme 6400c4 800Mhz) can do 400Mhz at 4-4-4-12 and shows voltage 2.100v.
■Does this mean that in order to achieve that 400mhz at 4-4-4-12 I must set the voltage 2.100v?

Is there a chance that I have to up the CPU voltage more when I try to adjust the memory even though the CPU has been solid stable in the Small FFT testing with the 1.2625v for 12h+ ?

 

[cnumartyr]

Now I am running prime95 again but this time with small FFT's, 9x333 and 1.2625v
It's been a few hours and I'll probably let it work overnight since it seems quite stable and the temps have never exceeded 55c. If everything goes well tomorrow I will start playing with the memory.

CPU-Z in the "Timings Table" shows that the memory (Corsair XMS2Extreme 6400c4 800Mhz) can do 400Mhz at 4-4-4-12 and shows voltage 2.100v.
■Does this mean that in order to achieve that 400mhz at 4-4-4-12 I must set the voltage 2.100v?

Is there a chance that I have to up the CPU voltage more when I try to adjust the memory even though the CPU has been solid stable in the Small FFT testing with the 1.2625v for 12h+ ?

Once the CPU is solid stable at a given clock, it shouldn't change (for a couple years atleast).

Also yes, odds are the Corsair Memory is rated for 4-4-4-12 at 2.1v DDR2-800. If it's in the SPD, it should run at those settings (or it's faulty).

 

[MjWoNeR]

In the Q6600 overclocking guide, graysky says that he run memory tests with memory set to 667Mhz and 1.111Mhz and noticed a very tiny difference between them as well as higher temps in the second.

Does this mean that I am better sticking with synchronous 667Mhz rather than asynchronous 835Mhz?

If yes, can I set the timings 4-4-4-12 with 333Mhz(x2) even though CPU-Z in SPD says 4-4-4-12 with 400Mhz(x2) ?

 

[cnumartyr]

In the Q6600 overclocking guide, graysky says that he run memory tests with memory set to 667Mhz and 1.111Mhz and noticed a very tiny difference between them as well as higher temps in the second.

Does this mean that I am better sticking with synchronous 667Mhz rather than asynchronous 835Mhz?

If yes, can I set the timings 4-4-4-12 with 333Mhz(x2) even though CPU-Z in SPD says 4-4-4-12 with 400Mhz(x2) ?

To be totally honest, it's not going to matter either way. The benches in real world will be about the same.

You might even be able to run them 1:1 and drop the timings to 3-3-3-9.. even that won't show a big difference.. Honestly.. for you I'd just run them 4-4-4-12 at 1:1 and drop the voltage a bit (less heat).

Only reason to get high bandwidth RAM is for synthetic benchmarks.

 

[MjWoNeR]

To be totally honest, it's not going to matter either way. The benches in real world will be about the same.

You might even be able to run them 1:1 and drop the timings to 3-3-3-9.. even that won't show a big difference.. Honestly.. for you I'd just run them 4-4-4-12 at 1:1 and drop the voltage a bit (less heat).

Only reason to get high bandwidth RAM is for synthetic benchmarks.

Yes, this is what I understood as well from reading the guide.

So your suggestion would be, stock 4-4-4-12 with 1:1 and maybe lower voltage for lower temps?
(By 1:1 you mean 333Mhz(x2) right? Same as FSB)

 

[cnumartyr]

Yep, that would be it!

 

[MjWoNeR]

Yep, that would be it!

Good. Hopefully settings will work out nicely as planned in theory.

Thanks again for the reply man, you are an ace. I so much owe you. lol




Edit :
I finished the 8hours+ Small FFT testing successfully, and now I moved on to blend tests and adjusted the memory at 1:1 and 4-4-4-12.
I wonder though if it will make a noticeable difference if I lower the timings a bit..

Edit 2:
I edit because I hate double posting.
Well everything turned out well. Rock solid at 3Ghz by 9x333 and temperatures less than 55c under load and synchronous memory at 4-4-4-12.
Thanks again