Microsoft’s primary focus will reportedly be on cutting down latency.
With the upcoming Project xCloud, Microsoft have made it abundantly clear that cloud-focussed streaming solutions is something that they’re going to push for a great deal in the coming years. Earlier this year, we reported on leaks that detailed that Microsoft is planning on making not one, but two concurrent consoles for the next generation- collectively being called Xbox Scarlett for now, one would be a traditional home console, while the other would be a box entirely reliant on cloud-based streaming solutions, with the purpose of the box being to cut down on latency as much as possible.
Now, according to new information that is being reported by Wccftech, Microsoft might be considering using a semi-custom AMD Picasso APU to cut down on latency for the cloud-focussed streaming box mentioned above, with cutting down on latency being something they’re focussing on first and foremost for the hardware.
The reason behind Microsoft’s keenness on wanting to use the semi-custom APU is that reportedly, its power consumption-to-performance ratio is something that they feel will greatly benefit the hardware, and not only allow them to keep the price and production costs at a minimum, but also retain the smaller form factor they’re aiming for. Apparently, both the datacenters and the Picasso silicon in the hardware itself will have accelerated deep learning, which will, by definition, get better at predicting players’ actions over time, and thus cut down on latency continuously. While the hardware itself will locally work toward some latency solving solutions, the task of cutting down on latency will mostly be reliant on the deep learning tech.
These are far from the only things we’ve heard about the tentatively codenamed Xbox Scarlett in the last few days. Recently, we learned that the console might possibly make use of Zen 2 and next gen AMD GPU, while other reports have suggested that its dev kits are codenamed “Anaconda”. Read more about both stories through the links.