Graphics Memory

This topic contains 2 replies, has 2 voices, and was last updated by  AndyM 6 years, 9 months ago.

Viewing 3 posts - 1 through 3 (of 3 total)
  • Author
    Posts
  • #30418

    AndyM
    Member

    Hi, I was wondering if anybody could explain how memory is typically structured in smartphones, or even just those featuring PowerVR graphics chips. If you are aware of some resource that I could read then that’s great too, as I spent a while googling and couldn’t find any information on the subject.

    I only want to understand in simple terms. I know that on desktop hardware, graphics chips typically have dedicated, low-latency memory (and game engines are designed with this in mind), and I believe that integrated chips use a portion of the desktop’s main memory instead, as they do not have their own dedicated memory (or perhaps just not enough). If any of this is wrong, please do correct me.

    Some comments that I have seen in reference to smartphone graphics chips seem to indicate that they do not have dedicated memory, and instead use a portion of the CPU’s memory, but I can’t find any information that explicitly confirms this. If it is true, would this mean that the structure is:

    “storage -> main memory -> graphics chip”

    Rather than:

    “storage -> main memory -> graphics memory -> graphics chip”?

    That’s quite an informal way of writing what I mean, but I’m just trying to express the path that the data takes, and where it is cached along the way.

    Thanks for reading,

    Andy

    #34703

    Joe Davis
    Member

    Hi Andy,

    The memory that is available to POWERVR graphics chips is very device specific. It is possible for a device to have a discrete POWERVR graphics core with dedicated graphics memory but, in the mobile realm where physical space used for chip silicon and memory needs to be kept as small as possible (and manufacturing costs need to be kept low), a much more common approach is to integrate the GPU on the same chip as the CPU using a design such as System on Chip (SoC). SoC is a design where the GPU and CPU (and other processors) are integrated onto the same chip. It is usually the case that devices using SoC utilise a Unified Memory Architecture (UMA). This means the memory assigned to the SoC is utilised by all of the processors on the chip. The way that memory is assigned to areas of the chip is completely device dependent and may or may not be user configurable.

    Hope this helps 🙂

    JoeJoe2011-01-17 10:12:59

    #34704

    AndyM
    Member

    Thanks, that covers pretty much everything that I wanted to know, and I have useful search terms now 🙂

Viewing 3 posts - 1 through 3 (of 3 total)
You must be logged in to reply to this topic.