I’ve doing research on the modern WDDM and how modern graphics cards with huge amounts of GPU RAM map to 32 bit Windows (e.g., Windows 7) systems.
I find a lot of documentation about how the entire GPU memory space must be mapped to the CPU address space, but this information appears outdated, and contradicts observations on modern systems.
I’ve found some fairly general wording in the Windows WDDM documentation that says stuff like:
http://download.microsoft.com/download/9/c/5/9c5b2167-8017-4bae-9fde-d599bac8184a/graphicsmemory.doc
Discrete graphics adapters generally share a portion of system memory with the CPU. Typically, these adapters do not ask for dedicated use of system memory for graphics, thus leaving more resources available for the rest of the system.
It seems clear the modern WDDM has “changed the rules” and made things more complicated, so statements like “installing a 1 GB video card will reduce your available CPU RAM by 1 GB” are now just wrong - they’re just too simplistic.
But I’d like to see it written down, from an authoritative source.
Can you point me to documentation on this? Something that specifically describes the strategy for mapping address range windows to GPU memory in modern 32 bit operating systems?
Thanks!
-Noel