At what point does the CPU bottleneck the GPU?

Hey all,

Building a rig with my son for homework gaming. ;) We live about 20 miles from a Microcenter, so keep that in mind when pricing. Anyways, here's what we are thinking. Already have a monitor, a 22" 1680 x 1050 and no SLI or Crossfire.

Antec 750 Earthwatts $80 @ Newegg

Biostar A770E3 AM3 $39.99 @ MC

X3 450 $80

G.Skill 4GB 1600 PC 12800 CL9 $50 Newegg

Antec 300 $55 @ MC

Seagate 7200.12 500GB $38 Newegg

Hyper 212+ $30

So we're at about ~$375 right now. He has just under $600 to work with, but I'm willing to help if need be. So with about $225 for a GPU left, my main question is, at what point is does the GPU become irrelevant due to the limitations of the CPU? We were thinking GTX 460 1GB, but a Zotac 470 is $200 after rebate at MC. Not to mention the recent availability of the 5870 @ $200. What about the 560 Ti for $250? Will the X3 cut it?

5 answers Last reply Best Answer
More about point bottleneck
  1. anyone?
  2. **crickets** Ok then just rate this build:

    Antec 750 Earthwatts 80 plus -- $80 @ Newegg

    Biostar A770E3 AM3 $39.99 @ MC [...] id=0349768

    X3 450 $80

    G.Skill 4GB 1600 PC 12800 CL9 $50 Newegg

    Antec 300 $55 @ MC

    Seagate 7200.12 500GB $38 Newegg

    Hyper TX3 $18 @ MC

    Sapphire 5870 - $200 AR

    Total out the door -- $561

    I decided to go with the 5870 after reading Tom's GTX 560 review. Can't beat the price...
  3. Best answer
    Re: question in the title

    There are many factors that determine this point. So I'll divide it into CPu and GPU

    CPU half of the explanation-
    Short answer. Depending upon resolution and type of game. IE how well threaded it is and whether it has physics and AI, or just AI. Also if it's AI intensive like RTS or not, like FPS.

    Long explanation
    Games are very poorly threaded. The part of gaming that is well
    threaded, the graphics portions, is handled by the GPU, which already
    have hundreds of cores.

    In gaming, the most CPU intensive task is AI. AI, by definition is not
    a parallel process. It is extremely difficult to thread AI. Most games
    that are "multi threaded" actually keep AI on 1 thread and throw the
    rest (minor far less intensive stuff) on the other.

    Can you design a game to utilize 4 or more cores? Sure, you can throw
    all the CPU non intensive calculations onto their own threads, but
    until someone figures out a good way to thread nonparallel
    computations, the performance increase will be minimal, as the hard
    work is still restricted to 1 thread.

    This issue has been stumping programmers for decades. There are ways
    to do this in specific situations, but no general solution yet. A
    general solution allowing infinite threading of nonparallel
    calculations would be the programming equivalent of finding the cure
    for cancer, noble prize stuff for sure.

    Basically think of it this way. On a math exam you have a 3 part
    question in which the answer to part each part depends on previous
    answers. IE

    A. Add up 3 and 5.
    B. Use the answer from part A and divide by 2
    C. Use the answer from part B and triple it.

    what is the final answer?

    This is the type of thinking AI requires. Threading this is the
    equivalent of calculating the answer to A, B and C simultaneously.
    It's not impossible like the mathematical equivalent is, but it's not

    For this reason, more than 3 threads has very little benefit. So if a game is well threaded, it'll utilize CPU up to 3 threads.

    On the other hand, if it's not well threaded you're better off with high speed single or dual core.

    GPU half of the answer:
    Short- depends on resolution, eye candy and drivers.

    Long Answer
    The higher the resolution, the more the GPU needs to calculate. Hence, the higher the resolution, the more the the bottleneck shifts towards GPU.

    At the same time, enabling AF and AA also require further GPu calculations. So if you turn up the eye candy, GPU has to work harder and further pushes bottleneck to GPU.

    Additional features such as tessellation also are GPU calculated. So these will also shift bottleneck to GPU.

    At the same time, even if the hardware is capable, if drivers aren't then you got issues. If drivers are extremely poor (take TERA currently for ex) both ATI and Nvidia GPU's are getting unplayable FPS when theoretically they should be fine. So in this case, the bottleneck is with the drivers, not the CPU or GPU technically.

    Bottleneck moves toward CPU if:
    Lots of Physics
    Lots of AI
    Poor threading
    Low resolution

    Bottleneck moves to GPU if:
    High Resolution
    Eye candy enabled
    Tessellation is used

    Bottleneck moves to drivers if drivers just fail.

    In general, since people want to game at the highest res and with as much eye candy as they can, you'll find GPu bottleneck.

    The one exception to this is poor software coding. If a game isn't coded to use CPU /GPU properly all bets are off and it'll depend upon game where bottleneck goes.
  4. Best answer selected by goatsword.
Ask a new question

Read More

New Build Bottleneck Systems Product