[CommanderDJ]

Hey guys,
I recently got into overclocking, and am trying to overclock an Intel Core 2 Duo E8400 (normally 3.00GHz) with a Gigabyte EX38-DS4 mobo. I successfully overclocked it to 3.60GHz (400x9), and Windows started with no worries. I'm not getting any stability issues, but under full load my processor has (according to RealTemp 3.40) reached 82 degrees celsius!! To make sure that it's not the program stuffing up I downloaded two other temperature-measuring programs and they both show similar values. I've changed it back to stock settings and under full load i get no more than 67 degrees. I usually run a distributed.net client in the background while I use my computer (which uses my spare processing power), and so if I were to overclock it, it would be running at full load for a lot of the time, but I don't think it's good for my computer to be running at 80 degrees celsius. What can I do?
Thanks,
CommanderDJ

[Kosh]

If you're going to overclock it's better to start with cheap, budget stuff because it voids the warranty, this way if you screw up you wont be out too much $$$$$$$.

[Nuke]

theres lots of hardware out there thats designed for overclocking. most motherboards over $150 have bios level support for overclocking. you can get cpu and ram that is designed for overclocking as well. of course im not sure if the warranties still apply if you do. id assume they do because the hardware is designed to be overclocked. besides ive never trusted a company to honor a warranty. they make you jump through too many hoops. id rather salvage it for parts and/or throw it away and save more time than its worth.

[MP-Ryan]

I have an E8400 and a different Gigabyte motherboard.  I attempted to overclock it a while back, but I was also experiencing massive heat issues.  It bugged me, since I've seen reports of people hitting 3.6 Ghz on an E8400 with the stock cooling, so I tried a new HSF unit as well - still massive heat issues.  E8400's are supposed to be OC stars though, so I did a little research and had the shop I bought it from take a look too.

Turns out my Gigabyte motherboard has a faulty temperature sensor.  Works fine at lower temps, but it goes up exponentially at higher temperatures despite other readings remaining stable.  You might be having a similar problem, or you may need a higher-end HSF unit.

I won't be buying any more Gigabyte boards, though.  Haven't been really impressed with this one.

[Klaustrophobia]

the easiest solution to heat is add more cooling.  get a beefier heatsink and/or add case fans. 

[Nuke]

i personally think overclocking just makes your case loud.

[Klaustrophobia]

it doesn't make it louder at all if your fans don't spin any faster.  which mine don't.  if you don't mind voiding warrantees, you can just use up the rather substatial margin built into the chips and get a decent overclock for free essentially.

[Nuke]

the a380 resorts to manual control for the fans which i think is totally lame. they seem to use an analog control line instead of the 2 pulse per rotation system your usual computer fans use. then again i could probibly use a microcontroller (probibly one of the 8 pin attiny mcus) to take the tach signal and convert it to analog with a pwm output. then simply let the mobo decide how much cooling is required.

[Klaustrophobia]

or just get a different fan.  sounds easier. 

[FUBAR-BDHR]

Overclocking = Fastest way to turn your processor into a key chain. 

Don't do it.  Just put the money into better equipment in the first place.  Seriously if the processor could reliably run at a faster speed it would be marked with the higher speed and cost more. 

[Nuke]

or just get a different fan.  sounds easier.  

find me 25cm fans with digital speed control thats also cheap. to me the $2 mcu sounds like an easier solution.

[Klaustrophobia]

Overclocking = Fastest way to turn your processor into a key chain. 

Don't do it.  Just put the money into better equipment in the first place.  Seriously if the processor could reliably run at a faster speed it would be marked with the higher speed and cost more. 

false.  the reason overclocking works so well for the most part is because the exact opposite of this is true.  giant blocks of chips are mass produced at a time, all more or less identical.  locks and limiters can then be thrown on them to create "levels" and allow them to sell to say, 3 different markets of low, mid, and high end with one production run, cutting the prices slightly for the low/mid  (or probably more likely, jacking up the "better" high end one).  re-tooling a production process is insanely expensive, so the fewer different blocks of manufacturing you have to do, the better.

there's also the issue of what we call here at my school "engineering overhead".  if you market something as working at 3 ghz say, you better make damn sure it actually works at  3 ghz and not 2.99.  to that end, the rated performance is always lowered beneath what it is physically capable of as a safety margin. 

[Nuke]

i had always assumed they performance tested the cpus after a production run and sorted them by performance. ones with higher ratings were sold at the premium end, and once that scored lower were downclocked and sold as the economy end. i know nvidia did something like this with their g92 line of gpus.

[QuantumDelta]

Where are you from?
If you want a warranty AND overclock there are some vendors (at least in the UK OCuk is an awesome one for it) who sell overclocked goods under warranty.

[CommanderDJ]

QD, I'm from Australia.

Okay, update: I tried the overclock again, and after half an hour of Prime95, the core temperature (according to RealTemp) at full load was 77 degrees. However, I then got to bios as quickly as I could, and the CPU temp there was 50 degrees. Should I just ignore RealTemp and go by what bios says? I would tend to think that bios would be more reliable...
If bios is right, then my CPU temp isn't too bad, right?

I'll probably buy a few more case fans anyway.

[Nuke]

do memory tests. if the results ever indicate errors, tweak your memory settings until they go away. this is another reason i dont overclock. unstable memory is bad and will lead to crashes. there's bound to be timing errors when you change the clock rate, so adjust the memory to compensate. for me i could never get my expensive memory which was designed for overclocking (and this is covered in warranty to a point) to run stable at the rated settings, it always had to be downclocked to run error free.

[Klaustrophobia]

memory in general is notorious for not setting itself up properly in BIOS if left to automatic.  mine i had to manually up the voltage to the 2.1 it was supposed to be at, and set the timings.  i'm not sure how intel works with this, but with AMD setups, you always can use dividers to keep the memory clock down when overclocking the CPU.  i've personally never seen a difference in performance with lower timings.  i sacrifice them if i need to to get the rest of it stable. 

[Nuke]

i had configured the memory manually, but it always gave errors. i was able to implement low latency timings but i gave up making it clock att he rated speed. it was just too unstable. besides the mobo im currently using doesnt have the overclocking features of the last one.

[Klaustrophobia]

if it returns memtest errors at the rated speed and voltage, it's bad.  send it back if you still can.

[Nuke]

well its been running fine for a couple years now and i dont think the warranty is valid anymore. i think it was just newegg hiking the specs to make it look better than it was. the memory (ddr2) according to the manufacturer was 800mhz which could have been overclocked to 1066mhz without voiding the warranty. newegg claimed that it was 1066 and didnt say nothing about overclocking. at the time my mobo bios had a problem with its overclocking features. so i blamed the mobo and said **** it. i never did like the idea of overclocking things. parts are rated to a certain frequency for a reason (hopefully a technical one and not a marketing one), and i dont like runing them faster than that.