Retail Q9650 Overclocking Thread

[PortCitySlim]

Is "low vid CPUs are not extreme overclockers" and "CPUs with stuck sensors are not good ones" myth true?

Does CoreTemp shows real vid?

#1. High VID chips tend to be harder to overclock up to a certain speed and then everything seems to even out, you either get a chip that needs tons of vcore to push 4.2ghz or a get a chip that hits 4.2ghz with low volts but at a cost of higher heat(Generally). As far as trying to overclock on air I would much rather have a lower VID chip.

#2. Stuck sensors do not have anything to do with overclocking, but it is very annoying not being able to read idle-mid load temps. Of course having a stuck sensor is a bad thing but not in the sense of overclockability.

#3. I would hope the latest version of Coretemp(0.99.4-64bit) would display the correct VID. When I get a chip even if it is a high VID I still try to overclock it before branding it a "lemon" there is always the chance a high VID chip will turn out to be a bad a$$.


Edit: I want to point out that the chip I had to RMA not only had stuck sensors but would not boot at > 450FSB no matter the volts. My motherboard/bios/ram should not be the problem because as we speak my Q6700 is running rock stable at 460FSB.

[TheGanG]

I want to point out that the chip I had to RMA not only had stuck sensors but would not boot at > 450FSB no matter the volts. My motherboard/bios/ram should not be the problem because as we speak my Q6700 is running rock stable at 460FSB.

Got a 6* max FSB run?

Thanks, got new CoreTemp now, trying at the moment.

[mreld3r]

My chip read the vid as 1.1 in coretemp on a p35 mobo, AND it defaulted to 1.1 Vcore. When I changed mobo's the vid and default Vcore went up to 1.225.
RealTemp reads mine as low vid 1.1 and high vid 1.225. Everest also reads my vid now @ 1.225. I'm running mine now @ 4.000ghz 9x multi. prime and intelburntest completely stable @ 1.3 Vcore bios and 1.264 Vcpu-z. Im sure that I can lower that a little, but i haven't tried yet. My problem is that one of my sensors (cpu0) reads almost 10c higher than the rest at full throttle.
That is kinda alarming really. Especially with intelBurnTest. That program scares the hell outta me. It will really heat up the cpu fast.
Even with the factory HSF on another mobo cpu0 was still a lot higher. So its not an issue with the seating of the HSF. I reseated it several times with both HSF's. Now I'm running a Big Typhoon and the temps across MOST of my sensors are about 60c at full throttle. I have a LC setup in the works. Im poor and have to wait till payday to get the last piece (cpu block). Right now the choice is a GTZ. Still kinda looking at a DTEK Fusion a little though.
All of my sensors read 37 degrees @ idle right now. Im not sure if they are stuck or not, i dont usually look at my idle temps. BUT they are the same @ idle. How can it be possible for one cpu to be 10c higher than the others when they are RIGHT next to each other. That seems like a really big difference really. im gonna play with RealTemp sensor test now. I dont have a clue what it is testing, but im gonna find out.

[mreld3r]

Apparently the difference is not so bad at lower tempuratures. How does this look to everyone else. Does yours look the same? My last chip (dual core) was almost the same temp across both cores and they "tracked" the same across the temp changes.



Have a look at the last 2 cores,,they are perfect. The EXACT same.

[Zucker2k]

Edit: I want to point out that the chip I had to RMA not only had stuck sensors but would not boot at > 450FSB no matter the volts. My motherboard/bios/ram should not be the problem because as we speak my Q6700 is running rock stable at 460FSB.

This is from one whose 9650 has stuck temps at 32/32/37/37. I think this rumor about overclockability is nothing but a rumor. My chip, which has a vid. of 1.250v overclocks like any other; on x48 mind you. There are too many variables when it comes to people's results that it is a mixed bag of chips when user tries to duplicate the results of another user; eg. user, cpu+board combo, ambient, rams, etc.

About your result, I believe you gave that pc too much volts; at least you may have gotten the voltages messed up somewhat. These chips without exception will do 450 fsb on the right board and with the right combination of volts. The stuck temp sucks, but the load temp is what matters. I never thought I'd see the day when a 4Ghz easy chip will get RMA'ed for being a poor clocker.

[Jor3lBR]

This is from one whose 9650 has stuck temps at 32/32/37/37. I think this rumor about overclockability is nothing but a rumor. My chip, which has a vid. of 1.250v overclocks like any other; on x48 mind you. There are too many variables when it comes to people's results that it is a mixed bag of chips when user tries to duplicate the results of another user; eg. user, cpu+board combo, ambient, rams, etc.

About your result, I believe you gave that pc too much volts; at least you may have gotten the voltages messed up somewhat. These chips without exception will do 450 fsb on the right board and with the right combination of volts. The stuck temp sucks, but the load temp is what matters. I never thought I'd see the day when a 4Ghz easy chip will get RMA'ed for being a poor clocker.

This is not the first time Intel does that, wouldn't be surprised (Q6600, Q6700 etc).

[PortCitySlim]

About your result, I believe you gave that pc too much volts; at least you may have gotten the voltages messed up somewhat. These chips without exception will do 450 fsb on the right board and with the right combination of volts. The stuck temp sucks, but the load temp is what matters. I never thought I'd see the day when a 4Ghz easy chip will get RMA'ed for being a poor clocker.

I might not have the experience you have with overclocking, but when I do test a new chip I keep it simple. I stated the bios setting I used before but here is the run down. I backed down my ram to PL 10 and auto settings for mostly everything(On my old quad I can run PL8 with 450FSB just fine), I did this to make sure that RAM was not an issue if I had any stability problems. I set the Vcore to auto, when trying to see where to begin with vcore I always start about 0.04v less then what CPUZ reports on auto. I set my NB to 1.38 and VTT to 1.40(Highest safe setting I know of for VTT with the intension of lowering later if I could), Ram was stock at 1.9v. I set CPU PLL to 1.54 although I could go up to 1.58-1.60 safe on 45nm according to Leeghoofd. I have never messed with GTL Ref volts, I always leave them on auto. I also turn all the Speedsteping/VT/Disable bit stuff off in the CPU tab.So given the above info, I was very surprised when I booted to 9x450 and got instant BSOD's, then when I tried 8x475 I could not boot period but 8x425 ran just fine.

I see it this way, there are plenty of people here who have Q9650's without stuck sensors, and when I pay the same price for my chip as they did why should I expect anything less? I did not RMA my chip because it was a poor overclocker, I RMA'd it because it had stuck sensors and wouldn't really overclock at all. Like you said these things are easy 4.0ghz+ overclockers and to suggest that it is impossible that I got one that was jacked up and would not hit 4.0ghz on a P5Q-D P45 board with the latest bios and the setting I mentioned above is kind of far fetched. Now, your comments are reasonable though, I did not give this chip a deep look through because my return/replacement warranty was running out and I had to make the choice of spending a lot of time trying to figure out the problem and running the risk of not being able to return it if it was a dud or just returning it and hoping for the best with a replacement.

BTW, username Makubex_GB has the same CPU/Mobo/Ram as I do and used one of KET's modded bios's and got awesome results with a 1.25VID chip. When I get my replacement I will be posting results and asking for help here in this thread or the P5Q-D thread.