Retail Q9650 Overclocking Thread

[nkresho]

Nope, ga-ep45-ud3r, just a misread in cpu-z.

[radaja]

nkresho,Hoss331 you guys got some nice chips there.

phsinc1,thats a very nice Q9550 you got too.

great clocks guys

[nkresho]

Hey thanks radaja!

I watched your posts very closely in the process of oc'ing my 8400. You helped me get to 4400mhz with it.

I'm finally replacing it with a q9550 when it comes in this week.



I just have to get control of my temps onthis q9650.
I might need a better rad or something.
I thought the hk3.0 would bring temps down more.
I'm getting over 65c in prime blend over 4500 mhz...

Maybe xspc 360 or the new fan spacing thermochill...

[andressergio]

nice to have you back nkresho 10C damn thats a nice temp for OC

hoss31 congrats thats a hella good OC

phsinc1 good chip there

lets keep the thread alive :O)

Cheers all
Sergio

[andressergio]

Hey thanks radaja!

I watched your posts very closely in the process of oc'ing my 8400. You helped me get to 4400mhz with it.

I'm finally replacing it with a q9550 when it comes in this week.



I just have to get control of my temps onthis q9650.
I might need a better rad or something.
I thought the hk3.0 would bring temps down more.
I'm getting over 65c in prime blend over 4500 mhz...

Maybe xspc 360 or the new fan spacing thermochill...

nkresho the HK3.0 didnt work as you expected ? i was going to swap my dtek fuzion for it tough it was better

Cheers
Sergio

[nkresho]

It is better by 1-2c but nothing too drastic. Worth the upgrade, IMO, especially since you're coming into summer temps down there.
It will let you clock a few mhz higher, which is why I got it. I might need to drop another few degrees (radiator upgrade) though.

Around the time of my cpu block upgrade (from fuzion v2) I also added a separate, second, loop for my northbridge/gpu, so my cpu had it's own dedicated loop on the mcr320.
Still, it isn't enough to keep these temps down while over 4500mhz. I'm spiking a bit over 65c at comfortable ambients (~22c).
And it's a little loud for my tastes at that speed/temp too with the panaflo fans up all the way.

[Gautam]

Hey all you guys with the insane Q9650 overclocks might want to check this out.

[Leeghoofd]

Don't do it Gautam I only have a 9550 !!!! there's no way I can beat these guys at or around 5Ghz

[Linuxfan]

Don't do it Gautam I only have a 9550 !!!! there's no way I can beat these guys at or around 5Ghz

There will definitely be some killer runs posted on the Q9550 and Q9650s....
If the my cascade system comes in soon enough, I'll have to see if I can get my Q9650 past 4.85ghz Superpi 1M...
Maybe it'll help over my watercooling...

[Hoss331]

Thanks guys. Temps for that run maxed at about 53c close to the 4 hour mark when the ambient temp started to warm up, average outside ambient was about 15-16c. Since im starting to see some cooler weather I have both my rads mounted side by side in my window with it closed down on top of them. For clock controls I use 800, 900, 0, 50 for up to 4.55 and 900, 900, 50, 50 for the 4.6, both my 3380 and 9650 have needed the 50 cpu skew to prime at 4.6. I also ran a quick prime at 506 to other day to help keep track of vcore scaling. Past 4.55 its definitely starting to not scale as well, either that or I just need to work on some other voltages, ill just keep playing with it.

[Ket]

Hey all you guys with the insane Q9650 overclocks might want to check this out.

We aren't insane just mis-represented

[nkresho]

Hoss, I agree on the scaling. I got stable at 4554 but it keeps freezing on me over that. No failed workers, just a screen freeze. I've tried adding more of everything vcore, vdimm, mch, fsb, etc. Nothing is working yet.

temps are geting a little high for me too. I have to keep the window open to keep the cores under 65. It's freezing in here.

[Hoss331]

Thats what was happening to me, id keep getting random lock ups and it was wanting abnormal references to run the times it didnt. Bump the cpu skew up to 50 and give it a shot, thats what solved my problem.

[In-Fluence]

Finally got my hands on my Q9650 today after 2 months of abysmal sevice and communication from overclockers. Ended up going to scan, got it all in 3 days. Along with this I got:
An akasa apache - will get the chance to see how it fairs against the noctua NF-P12
Inline water temp sensor
2x2Gb 8500 Corsair Dominator GT (not looking so promising)
A tube of MX-2 and MX-3
Another 120MM radiator

Anyway, currently testing at 490x6 (just FSB testing for now). Could post at 500-510 but I'm not anywhere near 'tweaked' enough yet, but I have a nice feeling about it.
After nudging the NB GTL to +30 it has given me 1024, 8 & 10k stability in p95 but has just failed 768k. Prime fails often on the first 2 cores, so they may be the weakest. Vid is 1.2 which I believe isn't bad? Temps are just skimming 60°C at 1.35v (with 1.35vtt), but once again, no voltages tweaked. Once the weekend comes, I'll be flushing the loop and doing a rebuild with the extra Rad which should help NB & CPU temp a little more

Just having a happy rant for a change

[nkresho]

Got my q9550 today to replace my e8400. VID is 1.3000. Batch in my sig. Kinda sucks. I might sell it on ebay and drive 2 hours to the nearest microcenter to get another, newer one tomorrow.

I am actually pondering getting another q9650 instead. LOL because this is my secondary rig. My "good" 9650 has me spoiled.
I can't seem to get this 9550 to 4000mhz with any less voltage than my 9650 made it to 4500.

Any input guys? Temp me...

And my temp sensors are off... by a lot... 46c 43c 38c 41c all prime loaded 99%, and the third mount???

[MrDucky]

Wow 1.3 is realy high vid i tought mine was high with my first Q9550 it was 1.275 it was the same batch well nearly... L846B770. i traded with my brother since he's not a real Overclocker, my vid now is 1.1375 4GHz stable @ 1.2v but i want to get more out of it.

[tool_462]

Got my q9550 today to replace my e8400. VID is 1.3000. Batch in my sig. Kinda sucks. I might sell it on ebay and drive 2 hours to the nearest microcenter to get another, newer one tomorrow.

I am actually pondering getting another q9650 instead. LOL because this is my secondary rig. My "good" 9650 has me spoiled.
I can't seem to get this 9550 to 4000mhz with any less voltage than my 9650 made it to 4500.

Any input guys? Temp me...

And my temp sensors are off... by a lot... 46c 43c 38c 41c all prime loaded 99%, and the third mount???

I'll sell you my Q9550 L847C219, ran 4.25 24/7 with 1.32v, 4.1 with 1.297v on the UD3P with 8GB installed. But I'd be asking eBay value, not MC value Could run higher I'm sure, my RAM was holding me back.

[nkresho]

Thanks tool, but I just ordered a new q9650 to replace it. I'm putting that q9550 on ebay tonight. Ocing the 9550 wasn't as fun as the 9650.

This one will be on air so we'll see what I can get out of it.

[In-Fluence]

A couple of tasters of mine so far:

Small FFTs @ 1.25vcore & 1.35vtt:

Now at 56Kfft & still going

Highest FSB so far:

[1nvisible]

Pardon me but why use Prime for testing? isn't OCCT Linkpack or LinX the best choice to check rock solid stability, haven't had much time testing.

[unclewebb]

Some computers can pass Prime but can't pass LinX. Some can pass LinX but fail at Prime. Some pass Prime Small FFTs but fail the Blend test. Some pass the 32 bit tests but fail when running the same test when using a 64 bit OS.

And a really good computer can pass any damn test you throw at it even when grossly overclocked.

Here's the latest version of RealTemp for you.
http://www.fileden.com/files/2008/3/...alTempBeta.zip

[In-Fluence]

Thanks for that Unclewebb, you're quite right. I always found I needed more vcore to pass P95 than linx on my Q6600, might be different on 45nm but I don't wanna run temps too high atm anyway.
Another good plus point about p95 is you can determine what voltages you need to increase if a certain fft size fails. Following that, I normally do a quick 1 hour test on OCCT to see if it's stable.
It can often give an error whilst doing this, but from that point it doesn't take too long to fine-tune the system, then I'll do a few hours in OCCT to finish. If I feel comfortable doing it, I'll give it a pew passes of linx too, but I'm gonna be very wary of it as I know 45nm are a little more fragile and I certainly dont want to fry or degrade the chip.
Currently at 494x8.5 to give 4.2GHz on 1.25vcore. This chip is chugging along nicely, will post pics when done

[1nvisible]

Some computers can pass Prime but can't pass LinX. Some can pass LinX but fail at Prime. Some pass Prime Small FFTs but fail the Blend test. Some pass the 32 bit tests but fail when running the same test when using a 64 bit OS.

And a really good computer can pass any damn test you throw at it even when grossly overclocked.

Here's the latest version of RealTemp for you.
http://www.fileden.com/files/2008/3/...alTempBeta.zip

Thanks for the tip.

Only have 2 fans enabled on MCR320, that's why temps are up in the 60's, going to run some more test tonight.

[1nvisible]

Still a learning curve for me, I see people mentioning VID, please tell me what does my VID mean, estimate how much voltage I'll need for the Vcore, currently its set to auto in the bios, thanks.

[In-Fluence]

1.1625 VID is good, the lower the better (mine is 1.2). Although this may not precisely determine how much vcore you'll need for a given overclock it is likely it'll be nice and low

If you're current setting is stable, try lowering the vcore a bit more to see if it can do the same with less.