Jeff, what you’re doing kind of parallels turbocharging a car’s motor. Performance gain, heat increase, possible reliablility issues, etc.
Of course, much like what you are doing, turbocharging can be acheived with great success but there are certainly risks, even when the install is done by an experienced mechanic.
I think I pretty much agree with everything Mike Dimmick said in his post. So naturally the next turn my mind took was to wonder about what was not said. Perhaps what was omitted was a few words about tolerances.
This would be a good thing to leave out because it’s the tolerances we know the least about. And ultimately the component tolerances are what make overclocking possible.
While it’s true that Intel tests every CPU, do we know how they are tested? Within what range of environmental conditions does Intel try to guarantee a given CPU will work? And if you severely limit the range of one condition, does that affect (increase) the safe operating range of other variables? If you stick a heat sink the size of Cleveland on the processor chip, does that increase the range of clock rates where it can run without error?
I think the answer to that one is a solid “probably”. If the failure rate were as high as I think Matt’s post tries to imply it will be, OCing wouldn’t be as popular as it is.
OCing is just about statistics. About betting that component tolerances … tolerances which we don’t actually know … will allow the system to work rather than fail at extremes … as it appears to have done this time for Jeff.
A fun statistic that I never expect to actually see would be the percentage of OC-ers who are primarily software folk vs hardware. My gut feeling is that OCing is more appealing to software types. Why? Because programmer’s are usually optimists … like Finagle and Murphy. 
Hardware folk not so much. Which (ironically?) is probably a key reason why system components have those “big” tolerances which make OCing possible in the first place.
-john
Actually, from a friend who works at Intel, has this to say:
A single production line may churn out identical except for their speed. The chips differ in speed due to the complexity of the manufacturing process. As exact a science as producing chips is, it isn’t exact enough to remove all the variability in the process. Manufacturer will at times intentionally underrate chips in order to meet market demand and create differentiation between high-end and low-end product.
Therefore, if we all remember the overclocking of Celeron 333’s in the past, there were only a few chips (typically marked with SL2W8/ SL2YK, brings back quake, and voodoo cards!!) that could be overclocked reliably. The trick is just finding the chip ID that was planned for 3.4 and is now 2.4 due to marketing needs, or just overclock a 2.4 to a 2.8, and be happy with the increase in speed.
~loki
In case anyone is interested, I did blow up my original P5B motherboard (as noted by Frederik above, see that thread for more details). The replacement is working fine. I think my problem was increasing the 965 northbridge voltage to 1.55; it overclocks perfectly well at the default voltage, anyway. So increased NB voltage isn’t necessary.
Thus, I removed the recommendation to turn up northbridge voltages since I now think that is a dangerous recommendation. The NB gets up to 80c under full load, with no voltage boost at all. When mine stopped working, I recorded northbridge temps at the cap of 120c (!!). Regardless of what’s going on with the computer, it’s always the hottest thing on the motherboard according to my temperature gun…
I did remove the “cap” – really more of a metallic sticker-- that ASUS puts on the northbridge heatsink. This uncovers the heatsink fins and seems to help a bit with temperatures.
People that don’t like overclocking should go back to playing in their sandboxes and stay out of these overclocking posts! My E6600 is overclocked to 3.33 and runs at 50c. It’s steady as a rock, fast, and makes my homebuilt PC a pleasure to use, just as my homebuilt 130mph boat is a pleasure to drive. Some people just don’t understand what go fast is all about.
prime 4 linux is mprime
Great blog! I’ve added a link to your blog on Blog of the Day under the category of Computer. To view the feature of your blog, please visit http://blogoftheday.org/page/112699
You don’t need 2 prime 95 installs. Follow the directions listed here:
Thanks for an excellent how-to guide, better since you have chosen almost the same components that I decided upon. Now that I read this I want to pick up a Scythe Infinity to replace my sad stock heatsink/fan.
Just asking what programs could possibly need a 3.00 up dual core processor to run ,… are you like trying to run 100 different programs at once?
I just upgraded my system.
MOBO = ASRock 775Dual-VSTA ATX Intel Motherboard.
CPU = Intel Core 2 Duo E6600
RAM = 2 Ghz (sdramm ddr2 667)
After running CPU-Z I noticed the following
Only displaying 1 core insted of 2.
Only displaying 1 thread instead of 2
Please help. I’m getting the run around from the manufactures.
Hey Jeff, I would liketo recomend you to use Orthos instead of Prime95, this is because orthos is already a dual core cpu stresser. I would also recomend to use some artic silver thermal paste and a thermaltake or zalman cpu fan even if you do not have high temps on your cpu and motherboard. This will help your computer to prevent furher damage because of the heat. Thanks for your report and please tell us your new results with these new features.
Hi,
I would really really like to know how to overclock my Intal Core 2 Duo E6300 from 1.86Ghz to about 2.13GHz.
Can any1 plz tel me~?~?~?~~?
I would like to know which version of Vista you are running is it 32 bit or 64 bit? I haven’t been able to get CPUZ to show me all the information in the 64 bit version of windows, it’s possible I may need to run it in compatibility mode I suppose due to the fact that you don’t actually install it. I am at work now, I suppose I will try that when I get home.
I just built myself a new home server/desktop machine, which is very quiet and fast.
Antec P150 case (comes with NeoHE PS, and 120mm rear exhaust fan)
Intel DG965WH mobo (onboard video, lan, audio, firewire, 6 hotplug-capable SATA)
Core 2 Duo E6600
Zalman 7000B Al-Cu (with ZM-CS1 mounting bracket)
Zalman 92mm intake fan to cool hard drives.
1GB Kingston DDR2-800 DIMM
WD 500GB RE2 hard drive (WD5000YS) - very quiet, esp. when suspension-mounted in the P150. supports NCQ.
AMD64 Ubuntu 6.10 (Edgy Etch).
Linux driver support for the G965 graphics are still a little rough. starting a second X server (e.g. with gnome’s switch user) corrupts the vid mode.
Support for the onboard IDE isn’t in any official stable kernel yet; I’m going to try Alan Cox’s patch for the Marvell IDE. I installed from CD using a USB-IDE adapter. (Vantec’s USB to IDE/2.5" IDE/SATA works well, and the power brick with a 4 pin molex output is handy for testing fans, and other messing around kind of things even when you’re not using the adapter!)
The onboard gigE doesn’t support jumbo frames; only the “server” e1000 boards do jumbo frames :(.
Intel’s mobo fan headers do PWM fan speed control in response to its temp sensors, so the front fan runs quiet at ~1600RPM. At 12V it runs fast and loud, 2800RPM. (I’m not using the inline resistor it came with). If I boot the board with a fan connected to the mobo’s rear fan header, and the disconnect it, it will spin up the front fan to compensate. (still PWM with a period of ~1 sec, so it’s a series of whooshes…) The BIOS has an option to “detect fans” at next boot, or at every boot.
The Zalman 7000 heatsink is just the right size for this board in this case. There’s maybe 1cm of clearance between the heatsink and the power supply. The 7700 heatsink would probably be too big, hanging further over the edge of the mobo, so I’m glad I got the 7000 + the mounting bracket. It has a 3 pin fan connector, so the mobo can’t control the fan speed, but the variable resistor it comes with works well. I leave it on near-minimum, and CPU temps stay under 50C, I think. (I can’t monitor from within Linux, only from the BIOS. lm-sensors doesn’t support the ich8 yet, and I haven’t found anything to read the CPU’s die temp sensor.)
The Zalman “flower” heatsink design has the advantage of moving air around over the voltage regulators and northbridge heatsink, which is why I went for it instead of a tower-style cooler like the Arctic Cooling Freezer 7 Pro.
The Intel board has absolutely no overclocking options. I can set the RAM timings manually, but the voltage is fixed at 1.8V. Make sure you get RAM that runs at its rated timings (as programmed in the SPD eprom) at 1.8V. Kingston ValueRAM is on Intel’s list of tested and recommended memory, so that’s what I got. I am sick of problems when using multiple sticks of RAM… (DDR2 is expensive right now; It was cheaper half a year ago, and I saw some industry analysts saying it should be cheaper in the future, too, so I’m going to add more RAM later.)
I actually work at Intel in the chip production area. (Etch). I recently built (one of my first builds) a Core 2 Duo system. I plan to overclock it (just because… ) However, I am unable to get the CPU temp at idle under the high 40’s to low 50’s. I haven’t even tried loading it up.
My build: Thermaltake Tsunami VA3000BWA case
XFX PVT71PUDP3 GeForce 7900GS 256MB vid card
Raidmax RX630-A Volcano power supply
G Skill DDR2-800 memory
Asus P5B-E mobo
Seagate Barracuda 7200.10 HD
SAMSUNG 18X DVDR DVD Burner
Silent 775 Thermaltake CPU cooler.w/artic silver5
I am trying a similar motherboard Core 2 Duo upgrade but having problems. Can someone tell me whether the Core 2 Duo requires 64-bit drivers for Windows XP?
You only need 64-bit drivers if you are using an x64 OS…
Hi there nice article one thing i am curious about i have an e6600 with 3 gb of ddr 533 pny and an asus p5n deluxe and i am having major problems getting the clock speeds over 2.7??? did you up the memory voltage as i have left mine as is so far?
Slowly increase the FSB speed in the BIOS. I have an E6600, which is a 2.4 GHz chip with a 9x multiplier. That means the FSB speed is 2400 / 9 or 266 MHz. As I increase the FSB speed, the CPU speed also increases. I first tried 333 MHz, which results in 333 * 9 or 3.0 GHz. As you can see in the screenshot, I’ve currently gone a bit further for 3.15 GHz. Remember, make small changes and test as you go. Don’t immediately go for the highest possible overclock. Be conservative initially; you can adjust upward more later after you develop confidence.
I CANT DO IT , MY CPU FREQ STAYS AT 266, AND WHEN I PRESS IT NOTHING HAPPENS NO OPTIOBS TO INCREASE IT, I GOT SAME SISTEM US U BTW 