1. Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Was thinking of upgrading my computer...

Discussion in 'Graphics Cards' started by BlackAzrael, Sep 7, 2012.

  1. xBruce88x

    xBruce88x

    Joined:
    Oct 29, 2009
    Messages:
    2,420 (1.33/day)
    Thanks Received:
    584
    Well actually... watts is technically a measure of both. More specifically its "watts" is the measure of work. Also... TDP is a pretty vague measurement as well, since some lines of cpu/gpus may have the same TDP for even a lower end model, though this is usually with CPUs and not so much GPUs. Not to mention the way TDP is measured/calculated varies from manufacturer to manufacturer. Yes its true that overclocking and increasing voltage will cause more heat and therefore cause the GPU to reach TDP sooner... IF the cooling solution can't keep up. This is why most people who plan to overclock typically buy a card with an aftermarket cooling solution... one that is capable of dissipating more heat. Or at the very least forcing their fans to maximum speed. TDP is the requirement of the cooling solution's ability to disapate heat in watts, or more specifically thermal watts, with the chip at its stock frequency and voltage at a giving usage of the gpu or cpu, not at any overclocked setting, in order to keep the chip below a certain temperature. Heck... running a cpu/gpu at full load can even exceed TDP since they figure at most times it would not be used at its max processing load. So therefore, naturally you would need a better cooling solution for overclocking. Of course manufacturers can get away with setting the TDP lower than the chip's actual max since most users rarely have their chip (gpu or cpu) running at max load all the time. The few times it does reach max load (loading a program or scene in a game) the cooling solution will have a bit of headroom left from the chip running below TDP for a few sec... or it may just start to be unable to handle the heat (and the chip rises a few deg) just before the task is finished but still staying below the max temp the chip can handle. And then the task finishes and the temps go back down since the chip is now back at a nominal load and therefore putting out less heat. They do this so they can skimp on their stock cooling solutions and save themselves money. Of course this is for your average user, i.e. the word processing, Facebooking, emailing crowed. For us gamers, video editors, and heavy multi-taskers we usually need a better cooling solution since we run our hardware harder and closer to "TDP".

    I'm hoping that helps a bit.. and that it came out the way I meant it to lol

    Actually heat IS what kills the cpu/gpu. Why do you think its called "frying" the cpu/gpu? Though suddenly increasing the voltage too much can cause it to fry since the temperature change due to the resistance would happen way too fast for the cooling solution to keep up. You wouldn't really feel the heat on the cpu (if you were to touch it after removing the cooler) since it would happen so fast it wouldn't really have time to heat up the cooling solution/entire cpu. It melts the one circuit, interrupting the flow of electricity rendering the chip useless, not the entire chip.

    @OP
    The only time you should really add voltage is when it starts to get unstable after a series of slight frequency bumps, unless you're very experienced with overclocking. The best way to go about it would be to see what the max voltage someone has manage... then stay slightly below that. Also, you'll want to go with the worst case scenario as far as the lowest "high" voltage for a successful overclock of your model cpu. this is a "on the safe side" precaution. of course you don't want to jump up to that amount from the start. you want to go up gradually. Basically how you want to do it...

    Frequency:
    Bump up (say 25mhz), apply, restart, load windows and test. Repeat until the system starts to get unstable.
    Voltage: only if the system is unstable... bump the voltage up the by the smallest amount you can. Apply, restart, test. If it stays stable then repeat the Frequency step until its no longer stable. Then repeat this step.
    Do this until you reach your desired performance in an app, such as a game or decoding video, or reach the clock speed you want. But be sure NOT to exceed voltages known to fry your model of cpu. You can find this out by searching for other overclocks. And of course you don't want to exceed the cpu's max operating temperature. Hell... you don't even want to be close to that.

    *a note... since you would most likely be overclocking via the Buss speed (since the muliplier is locked on your cpu) or core/fsb frequency, you may have to bump the voltage to the chipset (northbridge) as well but only if a few bumps of the voltage to the cpu after an unstable overclock doesn't work since too high a voltage can really mess up the chipset. Overclocking via the buss speed may also affect your ram's total speed as well. This may actually cause the ram to run SLOWER since the bios may change the ram:fsb ratio (or something similar) so you may have to adjust that accordingly, setting the ram ratio number higher than the fsb. It can also cause your ram to run too fast also, so a voltage bump to the ram may be needed, but follow the same rule with the chipset here as well. If you find that the overclock is making your ram unstable (use a memory test app to confirm) and you dont want to risk overvoltage, you can change the fsb/mem ratio also. you want the memory value lower than the fsb value lower in this case. If your motherboard's BIOS does not have settings for the ram and ratio, then stop the overclock if it makes the ram unstable and use the last overclock setting that worked.

    I know this is a lot to take in for a first overclock... so to keep it simple just focus on the CPU. Do small overclocks, and when the system starts to get unstable or wont boot (whichever is first) bump the voltage slightly. Then overclock a little more. keep repeating the process but you don't want too high a voltage. If it gets unstable and one voltage bump doesn't help... then call that a stopping point until you decide to try the more complex parts of overclocking or have learned more about overclocking and are confident in messing with other voltages and settings. I'd recommend having the motherboard manual with you while doing this. If you have a smartphone, you can usually download a .pdf of your manual (if you don't have the manual) to the phone and use a free .pdf reader (search your phone's app store) to view the manual. you may also want to save a copy of any notes like a text file to your phone as well so you can view them while messing with the bios.

    Hope this helps
     
    Last edited: Sep 11, 2012
    More than 25k PPD
  2. Black Mesa New Member

    Joined:
    Sep 3, 2012
    Messages:
    135 (0.17/day)
    Thanks Received:
    10
    Location:
    Canada
    Stopped reading after that ^^^^^^^^^^.
     
  3. xBruce88x

    xBruce88x

    Joined:
    Oct 29, 2009
    Messages:
    2,420 (1.33/day)
    Thanks Received:
    584
    Zubasa says thanks.
    More than 25k PPD
  4. Zubasa

    Zubasa

    Joined:
    Oct 1, 2006
    Messages:
    3,980 (1.35/day)
    Thanks Received:
    457
    Location:
    Hong Kong
    This is pretty much very basic middle-school/high-school physics. :toast:
    Watt is the SI unit for Power in physics period.
     
  5. xBruce88x

    xBruce88x

    Joined:
    Oct 29, 2009
    Messages:
    2,420 (1.33/day)
    Thanks Received:
    584
    ... that statement makes me want to cry. We touched on it a bit in high school, but I didn't really learn it well until a college course... My part of the country in the USA is pretty behind on education. its sad really. most of what i know computer, electronics, and automotive wise is self taught by just doing it or asking around
     
    More than 25k PPD

Currently Active Users Viewing This Thread: 1 (0 members and 1 guest)

Share This Page