1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Overclocking Overclocking questions

Discussion in 'Hardware' started by pimonserry, 2 Feb 2009.

  1. pimonserry

    pimonserry sounds like a party.

    Joined:
    20 Dec 2008
    Posts:
    2,113
    Likes Received:
    75
    So I have an Intel C2D 8500 running at 4.01Ghz, and IntelBurnTest says it's stable (testing with 1600MB RAM, 10 times). Interestingly, I can't test with more than 1600MB RAM or IntelBurnTest fails instantly. This may be due to Vista 32-bit.

    However, it may be running 'stable' at 4GHz, but temps have hit a high of 81degC (Hardware Monitor) or 84degC (Speedfan) during the BurnTest, which worries me a little. I have an OCZ Vendetta (92mm) with AS5, so I don't think it should be that high, but my voltages are on Auto in the Gigabyte BIOS.

    Voltages currently read: 1.536V Core Voltage (CPU-Z)
    RAM at 1.8V (speed is slighty overclocked so timings are 5-6-6-19, as it's running 1:1 with FSB at 422MHz)

    However, I am a complete noob at overclocking, and I want to drop the core voltage in order to drop the temperatures, but I don't want to fry or break anything. I know there's a slight risk involved, but has anyone got any idea what voltage I should aim for hitting? 'Cause I'm sure it's not supposed to be as high as 1.536V for stability.

    Also, should I be worried that my motherboard is 'rated' to 1600FSB, (4xFSB I assume) and I'm running it at 1688 (rated)?

    Thanks in advance :)
     
  2. DaveVader

    DaveVader Fast Action Response Team

    Joined:
    10 Aug 2007
    Posts:
    2,530
    Likes Received:
    78
    ouch, high voltage! Never leave them on auto when overclocking
    Intel burn test is the highest demanding stability test so it is likely to see ridiculously hight temperatures, your CPU will rarely (if ever) be put under as much stress as with IBT
     
  3. BlueTrin

    BlueTrin What's a Dremel?

    Joined:
    15 Dec 2007
    Posts:
    307
    Likes Received:
    1
    Someone correct me if I am wrong but if about your last point, I think it will put alot of strain on the motherboard chipset.
     
  4. pimonserry

    pimonserry sounds like a party.

    Joined:
    20 Dec 2008
    Posts:
    2,113
    Likes Received:
    75
    Would it be worth getting a better North Bridge cooler then? I currently have the stock one installed..
    Looking at the X48 specifications, it can take up to 1600MHz FSB, and I don't think any other Intel chipsets can take higher.. can they?


    Also: Success first try! I lowered (very tentatively) the voltage to 1.5V in the BIOS (CPU-Z reads it as 1.46 --> 1.49) and it booted first time, ran IBT with 1600MB RAM 7 times successfully, and core temps dropped by a massive 8degC with only a 0.05V drop!
    I'll keep going ;)
     
  5. JaredC01

    JaredC01 Hardware Nut

    Joined:
    24 Nov 2002
    Posts:
    1,259
    Likes Received:
    62
    Vader has it correct, NEVER and I mean NEVER EVER EVER IN A MILLION YEARS leave your voltage on auto when overclocking.

    With an 8500, you SHOULD be able to reach 4.01GHz without upping the voltage very much. My 8400 runs 100% stable at stock voltage up to 3.8GHz, though it takes a small bump to make it to 4GHz.

    Here's the same thing I posted to DorkSterr..

    Set the RAM at a 1:1 divider, and increase the FSB on the CPU by 10MHz using stock voltages until Prime95 fails in under 10 minutes.
    Once Prime95 fails, drop the FSB 2-3MHz at a time until Prime95 runs for over 10 minutes.
    Increase the voltage one notch in your BIOS, and move up 1-2MHz on the FSB until Prime95 fails in under 10 minutes.

    Repeat the process until you:
    1. Reach the maxiumum voltage recommended for air cooling on a E8500.
    2. Reach the maximum temperature recommended for a E8500.

    If you reach one of the above, drop the FSB down to the last working setup UNDER the required specs above. From there, your best bet is dropping the FSB about 5MHz, and running IntelBurnTest for somewhere between 5 and 15 cycles (the more the better).

    That's pretty much it in a nutshell. After you get the CPU where you can, then you can turn to the memory.

    I CANNOT stress enough though, YOU CAN'T RUSH THE PROCESS! If you're really wanting to get a good overclock, it WILL take a LOT of time. If you're not willing to put the time in, you're better off leaving your setup at stock speeds.



    Quick Edit:

    LOWER YOUR CPU VOLTAGE!!! The 45nm core chips shouldn't EVER run over 1.4v on air. Running at 1.5 can and probably will kill your CPU after a shortened lifespan. Drop it back down to stock, and follow my directions.
     
    pimonserry likes this.
  6. pimonserry

    pimonserry sounds like a party.

    Joined:
    20 Dec 2008
    Posts:
    2,113
    Likes Received:
    75
    JaredC01, thanks for the tips, and I know I don't really know what I'm doing, but I'd rather run it at 4GHz and just lower the voltage (and the temperature) rather than see how far I can push it temperature/voltage wise while keeping low voltages.
    It's alot easier to remember that I changed one value (CPU core voltage) one or two steps at a time than remember the FSB, the voltage, the Prime95 time (IBT simply says stable after ~10 minutes) etc.

    Although, thanks for the advice about never running >1.4v on air. Though TBH 80degC shouldn't do any permanent harm, and it's currently down to 1.475V (not much lower I know) that CPU-Z reads as 1.424 --> 1.440, stable with 7 runs through IBT, and only hitting 69degC.

    Though I'll drop it alot more now that I know. I'll get back to you tomorrow evening (hopefully without my CPU blown up ;)) as I have college all day tomorrow (unless it seriously snows).
     
  7. JaredC01

    JaredC01 Hardware Nut

    Joined:
    24 Nov 2002
    Posts:
    1,259
    Likes Received:
    62
    You can safely run up to 1.36 on air, all day long, without worrying about harming the CPU at all.

    If you overvolt the CPU, even if the temps are what you might consider reasonable, you risk burning the chip up.

    Quick Edit:

    The term for what I'm talking about is electro migration. When you put too much power to a chip, you risk the chance of electro migration. At a certain point, no matter what temp you're at, you will get electro migration.
     
    Last edited: 2 Feb 2009
  8. pimonserry

    pimonserry sounds like a party.

    Joined:
    20 Dec 2008
    Posts:
    2,113
    Likes Received:
    75
    Well, having read that, I just lowered voltages to 1.369 (CPU-Z says 1.33 --> 1.34) and to my surprise (me being dumb etc.) it POSTed and went straight into Windows (faster than before, I might add).
    Saved me alot of time, as otherwise I would have run an IBT on nearly every voltage step from 1.475 --> 1.369 (where I am now). +1 for you sir!
    Haven't got time to IBT this tonight, but it seems fine-ish so far.

    4GHz Cheesecake ;)
     
  9. JaredC01

    JaredC01 Hardware Nut

    Joined:
    24 Nov 2002
    Posts:
    1,259
    Likes Received:
    62
    Just be reasonable with what you're trying to do, and you should be able to keep your CPU alive and kicking until you're ready to upgrade, without any issues.

    :)
     
  10. pimonserry

    pimonserry sounds like a party.

    Joined:
    20 Dec 2008
    Posts:
    2,113
    Likes Received:
    75
    Update: college was off (snow) so I am burning through the IBTs again.. not good however.
    Last night it booted at 1.369V.
    Today it would boot at anything above 1.35V, however, at 1.35 it fails IBT instantly and BSODs.
    At 1.35625 it behaves a bit funny, and fails IBT after roughly 1min.
    At 1.3625, it fails IBT after about 1min.
    However, I thought I'd be clever here, and up the FSB voltage one notch (as it is 88Mhz over Rated FSB). So that's up 0.05V, and it boots much faster, and runs IBT successfully for 10 runs, and only hits 58C, but IBT gave it a 10% pass rate and the residuals are all over the place.

    Which would be the bit to change? Windows runs stable, but obviously underneath it's not quite right. Would it be worth going back up the CPU voltage ladder? It's getting perilously close to being back to 1.4V again. Or should I up the FSB or G(MCH) voltages?
     
  11. opal

    opal Eh?

    Joined:
    21 Jan 2009
    Posts:
    136
    Likes Received:
    2
    Your CPU will need a certain voltage to keep it running at that speed. If you lower it too much it will crash as you have witnessed. If you want to lower the voltage you're going to have to down clock it a little to get it stable.
     
  12. pimonserry

    pimonserry sounds like a party.

    Joined:
    20 Dec 2008
    Posts:
    2,113
    Likes Received:
    75
    Mmmm. I think I may have another problem now however: despite the BIOS setting now being at 1.375 (I'm going up, as IBT reports instability), CPU-Z has read the same voltages (1.328 switching occasionally to 1.344) consistently from BIOS values of 1.3625 to 1.375. Is this normal? Is the BIOS actually changing the voltage value? Or is something not quite right...

    Also, increasing voltages now has temps at 59C.. doesn't look like it's going to be stable under 60C :(
     
  13. JaredC01

    JaredC01 Hardware Nut

    Joined:
    24 Nov 2002
    Posts:
    1,259
    Likes Received:
    62
    Stable under 60*C full load for an E8500 @ 4+GHz is almost impossible without water-cooling. Mine currently runs about 65~66*C full load, but the thermal paste is also fresh (takes up to 200 hours of heat cycling to 'properly' break in), so final temps should be 2~5*C lower than they are now.

    Also, IBT gives me temps 10*C higher than ANY other benching program, so chances are if you're running at 60*C in IBT, you'll be somewhere closer to 50*C in any other bench program. You should be using RealTemp as well... CoreTemp and SpeedFan both report temps about 5*C higher than RealTemp, and RealTemp seems to be the accepted correct values.

    To add in a bit of useful info here... More voltage isn't always better. Most of the time chips will have a 'magic number' for their voltage to get the maximum overclock out of the chip. Some chips like lower voltage, some chips like higher voltage... The same can be said about heat... Some chips run warmer than others, and run perfectly fine. No two processors are exactly alike... That's where tweaking in overclocking comes in.
     

Share This Page