HOWTO: Overclock C2D Quads and C2D Duals - A Guide v1.0

I just overclocked my E660 to 333*9=3.0 GHz on a Gigabyte 965P-DS3 board. I was looking at cpuz after I did that and it now says that my pci express link width is at x1 instead of x16. Any ideas why and how I can fix that?
 
@mickster - I can fit my TRUE on my DFI LT P35-T2R just fine as long as it's oriented E/W so I would think you should be able to fit most HSs on your board as well. Starting points? If you have a quad, try my settings as a starting point and work up/down from there.
 
@boisebartian - dunno about that board. Did you move the video board to another PCIe slot? Many times, the first slot (the one physically closest to the CPU) is the only 16x PCIe slot. Simply overclocking shouldn't affect the link width unless you board has a known issue in which case you should check to see if there is a newer BIOS update that addresses it.
 
@boisebartian - that's odd... from what I've read you don't want to overclock the pci-e bus. What happens at 101 MHz?

@enlightened viet - how's your o/c coming along?
 
@bb - hmm... did you check in the Gigabyte forums about that. 105 isn't healthy for an expensive video card. I gotta think this is a known issue.
 
OK...made some progress today with a few voltage nudges. Got my Q6600/LT P35 T2R up to 3.3 Ghz and temps are looking very nice on a fairly warm day.
I'll be trying for more later...after a few hours to see how temps do.
 
Had a windows error message this morning related to a hardware issue so I nudged the voltages a bit and it's been stable for 12 hours now.
Makin' progress so I'm happy.
 
hi, does anyone can tell me how far can i go with vcore on a QX6850 watercooled? is 1,6V safe? I tryed to ocing him, but on 1,5 my limit is either 333x11, cant push more then it, or 430x9, is it better to rise multiplyer or fsb ?
mobo is a watercoooled Asus Striker II Forlmula (EK S-MAX, EK Mosfet 3a x2, EK SB Striker II), 2x2GB Mushkin XP2-8500, powered by P190 (650 and 550 W PSU's make a total of 1200W)
 
Edited on 08-Jun-2008: Guide is now version 1.7 – added a 2nd example minimizing the vcores on my system.
 
trying this out with my E4300 now. Was very disappointed at original E4300 results. Maybe this will help get them higher.
 
What should my max temps be for a Q6600? Here's my setup:

Q6600 SLACR
ASUS P5Q Pro
Hyper TX2
Centurion 590
ATI 4850
Corsair 650 TX
CORSAIR DOMINATOR 4GB (2 x 2GB) DDR2 1066 (PC2 8500)
Western Digital Caviar SE16 WD6400AAKS 640GB 7200 RPM SATA
LITE-ON 20X DVD±R DVD Burner SATA
Windows Vista Ultimate 64 Bit

I OC'd to 3.2 GHz and my highest temp at idle is about 43*C. Running Prime95 I've seen temps as high as 75*C. I've had it run stable under Prime95 with my core voltage at 1.35V in the BIOS and 1.312V at idle and 1.280V at load in CPU-Z for 15 hours. Everything else is set to auto except for my DRAM voltage which is set to 2.1V.

Is 75*C too hot?
 
Yeah, 75 °C is pretty hot. Try reading the temp with realtemp. It's tough to know exactly what the accurate temp of your chip really is due to Intel NOT publishing the Tjmax for each stepping/substepping.
 
I OC'd to 3.2 GHz and my highest temp at idle is about 43*C. Running Prime95 I've seen temps as high as 75*C. I've had it run stable under Prime95 with my core voltage at 1.35V in the BIOS and 1.312V at idle and 1.280V at load in CPU-Z for 15 hours. Everything else is set to auto except for my DRAM voltage which is set to 2.1V.

I am experiencing si ilar situation.
At stock speed my Q6600 G0 SLACR can run OCCT from 36c to 57c (21c range)

BUT

If I run my Q6600 G0 SLACR at 3.0Ghz runs OCCT from 43c to 70c+ (27c range+)

All stock voltages , common sense stuff turned off in BIOS
Asus P5Q -E , Scythe Ninja Rev. A , 4 Gigs Patriot 1066Mhz ram
used the "applied horizontal line of artic silver thermal paste" method etc

Should the temp range remain similar or vary when overclocking ?
Should I lower the cpu voltage ? reseat the heatsink?

COCO
 
Just overclocked my 9800GTX a little, from gpuclock 675/ memclock 1100/ shader 1688 to 740/1200/1850.
Now I'm looking at my 3.0Ghz CPU and I'm displeased. :)

The thing is when I raise my FSB from 333 x 9 @ 3.0Ghz to 350 x 9 @ 3.17Ghz and I Exit & Save Changes, the computer shuts down for a few seconds and then boots back up by itself and loads normally. Is this because of low vcore voltages? They were 1.216V. 48C temperature at idle isn't too high either, right? So what is it?
 
E8400 should be good at stock voltage up to 3.6 I think.

Also the picture you had up earlier showed 2.0ghz at 6x multiplier. Are these just idle clocks?
 
what prog are you using to monitor temps? Use realtemp. If thats what your using then yeah 48c is high for a tiny OC. On my old board I'd get 4.0 and idle would be 42c.
 
I raised the vcore voltage a little and I got rid of the annoying booting problem, now I'm running stable at 3.3Ghz with temps at 40C on idle according to realtemp. Hardware Monitor at BIOS isnt trustworthy it seems.

@Armoured, yeah, those are idle clocks. Well, have to be. :D real multiplier is 9 of course.

I now have FSB : DRAM ratio set to 1:1 when I have DDR2-PC6400, and the FSB is set to only 365, does that affect memory performance in a bad way?
I think I'll try to go for 3.4Ghz next.

Also how high can I go with temps at idle? 50C is too much?
 
I idle at around 44C with 3.8. 3.2 should be next to nothing for these e8400's. That's a high jump for such a tiny overclock as said earlier.

Are you using the stock cooler? Hum even using the stock cooler is sufficient for these guys.

Realtemp is .....very real. I thought those were other programs you were getting readings from but.....maybe some one with further knowledge can help you with this.
 
I started using realtemp when rydawg suggested it, before that I just checked the Hardware Monitor in BIOS, which obviously shows the wrong temperature?

I have now clocked my E8400 to 3.6GHz, I see no need to go higher than that. Well ok, I do.
Armoured, are you using the stock cooler on that 3.8GHz overclock?

EDIT: Hmm, seems to crash on Crysis and Flatout Ultimate Carnage, may rollback to 3.5GHz. Temperatures aren't the problem, only 50C on load. Vcore 1.300V

EDIT: I think it's stable at 3.5GHz now, played some Crysis - Reckoning.


By the way my GPU is now clocked in at 800/1900/1100. Can't touch the memory clock or I get some hardcore artifacting .
 
Very nice guide graysky, it certainly de-mystifies a lot of the process.

I do have a couple of questions though, to see if I fully understand it.

Say one has some DDR3-2000 memory. Is it correct that a 500mhz FSB with an 8x CPU multiplier with the ram in 1:2 divider mode sets up a 4ghz CPU with 2000mhz memory?

Are there any detrimental effects to toning down a CPU's multiplier to keep its clockspeed down under a higher FSB? In this particular scenario, I wouldn't want to push the CPU to 4.75ghz it would get with its standard 9.5x multiplier, but I do want the full 2000mhz from the memory; is there anything dangerous or otherwise ill-advised to running such a high FSB with a lower multiplier?
 
Does the Asus P5Q Deluxe have a bios setting for something like "PCI Clock Synchronization" to set to 33.33mhz? I don't want to forget to lock down any settings as I'm increasing my FSB.
 
I can pass prime stress tests at 333x9 with a 1.28v vcore in bios(1.26v in cpu-z) my hottest core reporting 45C, I can't however get my Q6600 to pass prime at 355x9 with a bios vcore of 1.32v now(1.304v in cpu-z)

My load temps under large FFT in prime hit 48C, do you think i should continue increasing the vcore or do something else? My ram is set to 1:1 and my motherboard is an Asus P5Q pro(also same question as above, I don't have a PCI Clock Synchronization)

edit: bumped my vcore to 1.34v(confirmed by cpu-z in windows) and prime still failed, load temps on the hottest core was 49C.
 
http://i37.tinypic.com/e6op6v.jpg
http://i33.tinypic.com/25a7os3.jpg
Realtemp calibrated at 264x6 and a .984v vcore to 6C above ambient(17.7C) at idle.

Bios Settings:
vcore: 1.52v
VTT: 1.38v
NBv: 1.40v
FSB:RAM 1:1
Everything else was on auto

Any suggestions? I think its obvious I shouldn't set the vcore any higher, I'm already knocking on the door of 65C under load and that is the hottest I will ever let my Q6600 get.

For reference my case has 3x120mm intake fans, 3x140mm+1x120mm+HD4870 exhaust fan, temperature of the room the PC is in is 64F. Xigmatek hdt-s1283
 
I OC'ed by Q6600 to 3.0 and i got the TRUE cooling. my Coretemps are 55,55,51,51 on idle. Is this too hot?
 
Hi there,

This is my first OC.

I have:
- Q6600 GO
- XfX 780i Mobo
- Quatro 3700
- OCZ Raptor 6400/800

I have read through the guide a few times, but am having issues finding out how to adjust the FSB in the Phoenix Bios. I found the multiplier and the mem stuff but see no options for FSB.

Is there any way the anyone can throw in a quick enable/disable checklist for phoenix and also explain where to adjust the FSB. I am very gun shy but just want to get 3gz out of this proc for now.

Thank you
 
I just read the whole guide (very well done, btw!) and I have a question about Bus Speed and Vcore. In much of the guide, you were lowering the Vcore. Do you *first* find the max Bus Speed and *then* lower the Vcore? When you find the max Bus Speed with a lower multiplier, how do you know what Bus Speed will be stable with the stock multiplier and how do you know what Vcore to start that at?

Using an e8500, should I start by raising the Vcore to a little less than 1.4V then find my max Bus Speed with the stock multiplier that way? Then, I'll gradually lower the Vcore till just-stable....
 
@Kyaydarin - Pick a target FSB speed and then see if you can run stable for an hour or so. If you can, you can either push it faster or begin minimizing the vcores. That's one approach. I'm not an extreme overclocker looking to squeeze every last MHz out of the machine. For my PC, I know 8.5x400 is very reasonable and I haven't pushed it for any extended period of time. I dunno about the e8500, but look around in the forums to see what others are hitting w/ that chip. 3.6 GHz? 3.8 GHz? You'll have to read up on it.

I don't put too much into lowering the multiplier and upping the FSB. You paid all that extra mash for the 9.5x multiplier didn't you? Yes, it's true that you can run faster memory benchmarks at higher FSB speeds, but unless you're folding or whatever 24/7 I don't see that the higher FSB with a lower multiplier will translate into anything beyond bragging rights and increase power consumption. That's just my opinion and interpretation of data I've seen. What I always end-up telling people is: it's your machine, push it as hard as you want. You'll probably replace it with something faster in a year or two anyway :p
 
graysky, could I also pick a target Vcore and then see how I get can up my Bus Speed (while leaving the multiplier at 9.5)? I'd up the Bus Speed in small increments, obviously, then back it down once it wasn't stable anymore.

Or would you stick to your method of finding the "target speed" first then adjusting the voltage from there?
 
That's perfectly acceptable as well... if you want to limit it to a particular vcore max and adjust the FSB around that, go for it. It might be a little more complex when dealing with your memory speeds depending on the divider you select though... let us know how it goes.
 
My new desktop is currently running through Prime 95 (Q6600 G0 8x400). I've had the same core fail over and over almost instantly, but now it took 30 minutes, so I guess that's progress. Just a few questions.

Is there a temperature or voltage difference between running the same fsb speed with a different multiplier? I'm running at 8x400. Would bumping that to 9 and reducing the FSB help my voltage?

I'm just bumping the volt core up a couple notches after a fail. After I get it stable in Prime, and I want to start tweaking the temps, should I move the NB or CPU Termination up, in order to move the CPU core down? I'm not sure how these relate to each other I guess.

I haven't OCed in so long! I'm glad the pencil trick is gone, but I think I'm a little in over my head.
 
awesome guide, I will give some of these things a shot tonight, I am having issues overclocking my Q9550
 
Back
Top