3 ms·
The little I’ve heard about the M1 makes it sound radically different in one respect: the system RAM is unified with the video RAM and shared, rather than parti
by chongli 6y ago
The little I’ve heard about the M1 makes it sound radically different in one respect: the system RAM is unified with the video RAM and shared, rather than partitioned the way it usually is on integrated video.
Now I don’t know anything about GPU stacks, but I could imagine that one change essentially requiring a rewrite of the entire stack if certain assumptions were baked into the old one.
The new model would seem to allow the developer to allocate a buffer full of stuff and rapidly alternate between CPU and GPU instructions on that buffer, without copying. If so, that’s a pretty huge win for performance.
- my123 6y agoThat's present on basically every Arm SoC, with the CPU and GPU on the same die. (and Intel & AMD APUs too, the difference is that Apple is scaling UMA beyond what anyone did for sold systems in x86 land that aren't consoles) Currently, Apple and NVIDIA have the beefiest UMA w/ the CPU GPUs available, and those are Arm solutions.