memory intensive task vs. processor intensive task

Discussion in 'Mac Basics and Help' started by alien, Oct 25, 2005.

  1. alien macrumors member

    Jul 17, 2002
    London, ON
    I'm hoping that someone can help me understand the difference between "Memory Intensive Tasks" vs. "Processor Intensive Tasks". I'm wondering about the difference between these as I've read comparisons between Dual Core vs. Dual Processor, and it seems as though everyone is saying that the new Dual Core Powermacs are better at memory intensive tasks, while the old Dual Processor Powermacs are better at processor intensive tasks. If someone could explain to me the difference it would be greatly appreciated. Thanks.
  2. stoid macrumors 601


    Feb 17, 2002
    So long, and thanks for all the fish!
    The difference is subtle. A processor intensive task is any task that is speed limited by how fast the processor can compute the data. Example: encoding video.

    A memory intensive task is any task that is speed limited by how fast the memory can feed data to the processor. Can't think of a good example though off hand.
  3. robbieduncan Moderator emeritus


    Jul 24, 2002
    Database servers can often end up memory constrained (you basically want the whole database in RAM as disks are slow).
  4. alien thread starter macrumors member

    Jul 17, 2002
    London, ON
    Ok... so say that I'm doing some sort of rendering on an old Dual processor 2 vs. the new Dual core 2... would that be more memory or processor intense. I'm thinking processor, but wasn't too sure if there would be memory intensive things going on as well that would even things out in a situation like that.
  5. Coheebuzz macrumors 6502

    Oct 10, 2005
    Nicosia, Cyprus
    As you said the difference is subtle, you cannot really tell which processor is generally faster because all tasks have to read/process/write, so what you gain in the dualcore you loose with the dual single and vice verca.

    Encoding video is a memory intensive task as well, because a large stream of audio/video data has to be fed to the cpu for processing and a faster memory bandwidth could feed the processor with bigger chunks for processing.
    For rendering speed, you can go to the Luxology forums as they have some render times using different G5 configs plus the new Quad G5, they say that is 225% faster than the Dual 2.5 GHz G5!!! So it seems that the dualcores perform a bit better on rendering.
  6. mrichmon macrumors 6502a

    Jun 17, 2003
    As other posters have pointed out it isn't an either or issue. These are more labels to characterize programs and help performance specialists describe programs.

    Having said that, in the general case rendering is CPU constrained since the scene descriptions are relatively small in comparison to a final bitmap. This assumes that the rendering is done by the CPU rather than GPU.

    As I understand it, when using CoreImage a program will be constrained by the interconnect bus since the GPU is optimized for the operations used by CoreImage. The bottleneck with CI is getting the data from memory to the GPU and from the GPU back to memory.
  7. wordmunger macrumors 603


    Sep 3, 2003
    North Carolina
    Here's my guess:

    CPU intensive -- Video games
    Memory intensive -- nearly everything else
  8. alien thread starter macrumors member

    Jul 17, 2002
    London, ON
    So if the new dual core computers are more "memory intense", then will adding more RAM to these computers help speed them up more than the previous generation, or does it explicitly have to do with the fact that the chips have more cache and that they can 'cache snoop'?
  9. Makosuke macrumors 603

    Aug 15, 2001
    The Cool Part of CA, USA
    I'm not an expert, but I'd say not necessarily; their memory advantage comes from having *faster* main RAM, and more and faster cache, than from the volume of it.

    So, for example, if you have a task that requires moving a lot of data into and out of memory but only needs, say, 500MB of memory, then that task will not go any faster if you already have 1GB of RAM in your computer, be it old or new architecture, but it should go faster on the new architecture compared to the old, since it can move that data back and forth faster.

    This is among the reasons the dualcore chips will generally be a little faster than an otherwise identical dual-singlecore PM in most tasks at the same clock speed. How big the improvement is will depend on the type of task.

    Now, if you have a task that requires a large volume of RAM (say, huge scientific simulations or Photoshop), then the 16GB upper limit on the new PMs will be an advantage, since all else aside more RAM will go faster. These sorts of tasks aren't common, though--I have 2.5GB RAM, and there have only been a handful of times I've even used all of it (mostly opening *giant* satelite image files).

Share This Page