The Redmond-based company’s more powerful console (which it says should be considered a half-generation update, even though it’s been out for four years) has capabilities that Microsoft hasn’t really used much.
One of the new features of the PlayStation 5 Pro is machine learning. It’s just that the Xbox Series X also has such hardware, but it’s understandably inferior to Sony’s. That shouldn’t be a problem, but it’s interesting that Microsoft doesn’t seem very interested in using it. The editors of Digital Foundry discussed this in their weekly podcast. So in theory, Sony has caught up with Microsoft in machine learning, which wasn’t hard since Redmond hasn’t done much with it, but in practice, that doesn’t mean Sony has caught up…
On the Xbox Series X, 100 TOPS refers to INT4, which is not used for machine learning upscaling. That said, Microsoft could take a step to develop a machine learning model that could run on both the Xbox Series X and Xbox Series S. This could be applied to Auto SR (Automatic Super Resolution) for Xbox 360 supported titles. Microsoft has created a model that can run on MPUs capable of 40 INT8 TOPS, a level that is absolutely unmatched by the hardware of the current generation of consoles.
The Xbox Series X and Xbox Series S architecture was developed by Microsoft back in 2016. At that time, machine learning was still a bit far away, so it’s not a big surprise that it wasn’t used much. That could change with the next generation of consoles, and the PlayStation 5 Pro could be a precursor to that, as the PlayStation Spectral Super Resolution upscaler has performed well with it so far, which is a very good first attempt.
So machine learning is likely to be a key factor for the next-generation Xbox, as Microsoft will not let Sony’s achievements go unchallenged.
Source: WCCFTech
Leave a Reply