4 ms·
I assume you mean that AMD's integrated graphics still do not compete with discrete GPUs.
by SubiculumCode 6y ago
I assume you mean that AMD's integrated graphics still do not compete with discrete GPUs.
- p1necone 6y agoWhich isn't surprising - integrated graphics are never going to beat discrete GPUs with their own power delivery, cooling and dedicated high speed memory.
- foota 6y agoThis is not intrinsically true. Cooling and power delivery certain have it beat, but it seems feasible to beat based on memory bandwidth. I bet a GPU uses roughly the same order of power/cooling that say half a threadripper does?
- eloff 6y agoNo, how do you beat the wide bus memory bandwidth of a discrete GPU with the standard dram bandwidth? Even if your processor was not competing with the GPU for bandwidth you'd still have a small fraction of the memory bandwidth available.
- p1necone 6y agoYou'd have to come up with a new architecture to make this work, but perhaps you could have CPU/GPU on the same chip with dedicated GDDR memory lanes. PS4/XBOne have an APU with powerful (ish) graphics - how does their memory work?
- eloff 6y agoThey work with graphics memory rather than regular memory if my memory serves. Hehe.
- foota 6y agoAw, I was thinking bw between the CPU and GPU directly, my bad.
- rubber_duck 6y agoAren't console GPU chips on the same die as the CPU ?
- wtallis 6y agoMight be more accurate to say that consoles put the CPU on the same die as the GPU. The GPU portion of the console SoC die is bigger than the CPU portion and the memory controller is very much a GPU memory controller not a CPU memory controller.
- speedgoose 6y agoYes sorry to not be explicit enough. It's still way too slow compared to a dedicated GPU.
- rurban 6y agoHW-wise the Vega is better, but the SW, OpenCL compared to Cuda at least on Linux, sucks. It rarely works, and you get the crashes at runtime, not compile-time.