QX9650 on Water

[Getttosmurf]

hm is it really not possible to run say 1.55-1.6v for 24/7 on these 45nm ? i mean
my e6600 has been running 1.7v (bios setting) 24/7 on watercooling for almost a year now. I know this is 45nm but stock voltage is almost the same right ?

I understand that the chip might not last for 5 years @ 1.6v but all i need is 1-2 years , by then i will have some other poor chip on way to high vcore and il be happy if i can run the qx9650 @ stock in some linuxbox or smth.

[Zeus]

Dissapointingly the highest FSB I attained on this processor under high cpu clocking was 445, anything higher will FREEZE at post. So, I kept on upping the multi. Please note also: My X38-DQ6T was good to 535 FSB on my old E6850 cpu. So, this FSB wall is definitely CPU limited.

Try raising PLLvoltage when you're aiming for high FSB, this might help to some extend.

Be carefull not to overdo it in combination with high Vcore.

[Yoxxy]

So Dominick,

after your findings, what do you think is the sweetspot on this chip?

I've found the safest and sweetest overclocking @ 4.00Ghz that is:

10x400 @ 1.4Vcore. I'm thinking of a 24/7 setup.

But I've seen in one of your pictures that you were able to have 10x400 and running at a stunning 1.365? Did that worked well? Did you set that stock volts manually or you set your mobo to AUTO on the Vcore?

Also I want to see where the whitepaper is (anyone has a link). I just went to the intel processor finder, and it tells me that the voltages on this chip range between 0.85V-1.365V.

http://download.intel.com/design/pro...s/31872601.pdf

Page 17.

[RPGWiZaRD]

I would use up to 1.55v on water, up to 1.45~1.5v on highend aircooling if temps are good. Seems like this one didn't scale well with voltage, a bit like my E6750 that does 3.36GHz stock voltage and 1.41v at 3.6GHz and 1.52v at 3.75GHz.

[CrashOv3r1De]

Very nice results, looks like you've got a great cheap. I seriosuly suggest sending this over to someone who has phase cooling to see how high you can scale it.

However with current setup dont go over 1.45v for 24/7 use. See if you can get 4.2 for 24/7

[xgman]

Intel whitepaper says 1.45v, said 1.55v on Conroe before they started to degrade.

I trust the whitepaper I guess.

I wonder how this really translates to heat vs chips cooled on phase with minus temps?

[xgman]

Very nice results, looks like you've got a great cheap. I seriosuly suggest sending this over to someone who has phase cooling to see how high you can scale it.

However with current setup dont go over 1.45v for 24/7 use. See if you can get 4.2 for 24/7

My unofficial results and some of Jinu's own testing seems to indicate that the average 9650 will on single stage phase, run stable in the neighborhood of 4600-4900 mhz with "up to" 1.6v. That tells me that the chips run fairly cool to begin with and scale fairly aggressively with air and then level off pretty fast even under phase. dual stage phase and more extreme methods will obviosuly get you somewhat higher, but may need more voltage. The gains over air on the qx6850 were slightly higher with regard to scalling. I think the fsb limits on the qx9650 may be due in great part to the extra cache, but that's just a guess.

[Yoxxy]

Should degrade just as much on phase as it does on air.

Voltage doesn't matter based on temperature.

[xgman]

Should degrade just as much on phase as it does on air.

Voltage doesn't matter based on temperature.

I think it does at least to the extent that over volting on air can produce higher temps that could damage the cpu in some instances which would be a direct result of the over volting, no?

[Yoxxy]

Temperature can kill the cpu, but then it is the temperature not the voltage.

[dominick32]

I think it does at least to the extent that over volting on air can produce higher temps that could damage the cpu in some instances which would be a direct result of the over volting, no?

Yes, that has always been the main correlation. Higher Volts = More Heat = Sudden Death

But, on the topic of adequately cooled QX9650 processors, (maximum 65*C load temp) and high volts. Is there any more data on 1.60 to 1.70 vcore for short term benchmarking??

[xgman]

Yes, that has always been the main correlation. Higher Volts = More Heat = Sudden Death

But, on the topic of adequately cooled QX9650 processors, (maximum 65*C load temp) and high volts. Is there any more data on 1.60 to 1.70 vcore for short term benchmarking??

I think it's too early to say one way or the other. I ran my qx6850 at 1.61v on phase since prior to the release date and it is still running strong. Only time will tell I guess on the 45's.

[dominick32]

Here is some interesting data:

It actually posts to the bios at 0.95vcore...
But it takes 0.99vcore to actually boot windows without a BSOD.
Check out those idle temps. 3.00 GHz @ 0.99 vcore

The only thing it cant do at .99vcore is quad prime. It errors out immediately but Vista is fully functional.

[jabski]

now thats energy saving for you

[oscar3d]

Here is some interesting data:

It actually posts to the bios at 0.95vcore...
But it takes 0.99vcore to actually boot windows without a BSOD.
Check out those idle temps. 3.00 GHz @ 0.99 vcore

The only thing it cant do at .99vcore is quad prime. It errors out immediately but Vista is fully functional.

Excellent findings!!!

So, can you have 4Ghz stable on this thing at stock volts (1.3625V)?

[dominick32]

now thats energy saving for you

Yes, but the scary part is that I just juiced an additional 60% more voltage last night 1.60vcore compared to what this puppy can theoretically handle at stock core speeds.

[dominick32]

Excellent findings!!!

So, can you have 4Ghz stable on this thing at stock volts (1.3625V)?

Patience is a virtue my friend. Thanks for the heads up and responses throughout my thread.

Give me a few more days, but I am really trying for a 24/7 stable @ 4.3 GHz 1.45vcore

[dominick32]

One more update: Crysis is playable at this setting: 3.00 GHz @ 1.03 bios selected vcore. Well, I played Crysis for 10 minutes or so and started quad priming.

[oscar3d]

LMAO!!! HAHAHA!!! THAT"S F>>>> AWESOME!!!

Power savings here I go... that's going to save me green from PG&E.

Thanks man! Though I didn't appreciate when you called me "ridiculous" at first... in this thread.... :-)

[dominick32]

LMAO!!! HAHAHA!!! THAT"S F>>>> AWESOME!!!

Power savings here I go... that's going to save me green from PG&E.

Thanks man! Though I didn't appreciate when you called me "ridiculous" at first... in this thread.... :-)

I still love you man!!
Just thought you were trying to call me out.. I already forgot about that.

[oscar3d]

No man! You got all my respect... And I can say you have big balls to be testing your Yorkfield like you've shown. Kudos!

[dominick32]

HOLY **** GUYS!!! SORRY FOR MY LANGUAGE

4.70 GHz on Water:

[dominick32]

I think I am finally happy with a 24/7 spot. Last night late I managed let let my rig sit @ 4.2 GHz 1.47vcore quad priming for 1 hour during dinner and shower. Tonight for 15 minutes or so, just to verify everything is OK and to get some temperature monitoring I ran quad prime again at my 24/7 speed here.

Temps are usually around 66*C to 70*C loaded, but I think I am OK 30 degrees away from tjunction.

[dominick32]

Got it completely 3d/benchmarking stable @ 4.56 GHz 1.60vcore.
New 3dmark05 record for me.

[vmsein]

For comparison and just to add more data to the knowledge base, here are my results on water. This is with vcore set to 1.53750 in the BIOS of my ASUS P5E3 Deluxe WiFi: