The OG Steam Deck (7nm, comparable to the Series S) has a die size of ~162mm2. In there, it packs an 4 core, 8 threads CPU and a 8 CU GPU.
On the other hand, the Xbox Series S packs an 8 core 16 threads CPU with 20CU GPU of the same architecture in ~197mm2 die.
This is a technical question, how come the Series S packs much more in just 25% more size? I’m not saying the Steam Deck should be as powerful as a Series S (that’d never happen, the power constraints would not make it possible), but I wonder if the CPU in the Series S is cut-back or if there’s anything in the Steam Deck’s SoC that could have been removed to get a lower cost.
Because van gogh ain’t designed for valve’s gaming device. It contained blocks that supported computer vision. Microsoft and sony also customized their console socs by removing minor features they didn’t need and decreased area by a lil more
I understand, but I wasn’t aware of any computer-vision accelerating block in there.
Is there really any? Might be used for a future Steam Deck-based VR headset as it’s rumored?
No. It was there because Microsoft asked AMD to add them, before canceling the Van Gogh SoC order.
Van Gogh has (suppsedly) 2.4 bn transistors while Series S has 8 bn.
Completely false. 2.4b transistors would mean it’s less than 1/3 as dense on the same node while being lower performant (so HP cells are not a factor). Why on earth would you even use 7nm when 20nm (and even 28nm) can reach that density easily.
Use you brain. Just because TPU listed wrong data, doesn’t mean you have to parrot it.
It has 40% the CU, 50% the CPU, 80% memory bus, nearly 100% front end (geometry, ACE etc), 100% display, video, I/O etc.
At bare minimum you are looking at 70% transistor count.
Bingo. The Steam Deck OLED’s SoC seems to have cut out the Cadence DSPs, which we can see with the significantly smaller SoC (especially when combined with the new process node).
Pretty sure VGH also has more PCIe lanes on die as well, alongside full size Zen 2 cores (all console variants have a cut FPU).
So does this mean the steam deck SoC has particularly good performance in machine vision applications? This is actually relevant to my work lol