Xbox Scorpio features 12 TFLOPS of 16-bit computational power according to Mark Cerny’s definition

If you’ve been keeping up by Sony and their rather mediocre PlayStation 4 Pro console, you’d know that Sony has been on the back foot recently mainly due to the fact that the Pro was powered by outdated hardware reaching 4.2tflops of compute. In stark contrast, Microsoft announced the Xbox One Scorpio that will be powered by next-gen technology reaching 6tflops of compute.

Since then, Sony has downplayed the overwhelming power deficiency the Pro had in comparison to the Scorpio. Mark Cerny was recently deployed by Sony to spin the lack of performance and ability to reach true 4K visuals. Mark Cerny didn’t disappoint after he announced that the PlayStation 4 Pro was capable of using ‘a new’ technology that allows the 32bit system to ‘switch’ to 16bit floating points–also known as half-floats–to give the system a boost in computational efficiency. Using some questionable math, Cerny concluded that the PlayStation 4 Pro was officially capable of 8.4tflops of computational performance.

However, this technology is nothing new. Microsoft and Nvidia have used this technology since 2002. The feature was used in the old Geforce FX series of cards. Also, the main advantage of the method is that it requires less bandwidth and storage; something the Pro obviously lacks. The problem is that all X86 processors and graphics units are capable of this. Hence making the Scorpio–by Mark Cerny’s own admission–12tflops in 16bit terms.

It’s great that game developers are looking at ways to make their games more efficient, but misleading comments like this have the foul stench of litigation to them. Mark Cerny probably needs a PR agent too.

Even a basic Google or Wikipedia search can reveal that Mark Cerny was lying. Read more about it here.

Leave a Comment