GameSpot may receive revenue from affiliate and advertising partnerships for sharing this content and from purchases through links.

Here's How Many TFLOPs PS5 Has--And Why They're Misleading, According To Sony's Mark Cerny

During a deep-dive into the PS5's architecture, Sony technical designer Mark Cerny took some time to set the record straight on how teraflops actually work.


During Sony's Road to PS5 livestream, Mark Cerny laid out the core technical specs of the upcoming console, and why fans should be excited for the future of PlayStation. In addition to explaining the PS5's specs, and how the console's unique 825 GB SSD works, Sony's technical designer took some time to explain why the general focus on teraflops, a common topic over the previous years for the PS5 and Xbox Series X, is a bit misleading for general consumers.

In this incredibly detailed dive into the PS5's architecture, Cerny explained that a teraflop is defined as "the computational capability of the vector ALU." This aspect of the system mainly comes down to one facet of the processing speed of the system's GPU (graphics processing unit), specifically for the time it takes to for the console to generate high-quality images. While you can focus on teraflops as a defining indicator of console performance, it's just one of many different factors at work for the platform's capabilities. Prior to the reveal, PS5's specs were detailed, revealing that the console's CPU possesses 8 Zen 2 cores, and the custom GPU with an RDNA 2 architecture can generate up to 10.28 TFLOPs of power. For comparison, the Xbox Series X's GPU can generate up to 12 teraflops--but Cerny made a point to clear the air on what teraflops can actually do for a console.

Please use a html5 video capable browser to watch videos.
This video has an invalid file format.
Sorry, but you can't access this content!
Please enter your date of birth to view this video

By clicking 'enter', you agree to GameSpot's
Terms of Use and Privacy Policy

Now Playing: 7 Important PS5 Features, Explained

During the presentation, Cerny took some time to stress that the focus on teraflops can be misleading, as the overall processing capabilities of the system take into account many other factors in addition to teraflops. Cerny made a point to show that 36 CU running at 1 GHz and 48 CU running 0.75 GHz both equate to 4.6 teraflops in power. This explanation helped serve the point that he preferred to look at the GPU's frequency rate when measuring the system's capabilities.

"That's just one part of the GPU, there are a lot of other units--and those other units all run faster when the GPU frequency is higher," Cerny said during the presentation. "At 33 percent higher frequency, rasterization goes 33 percent faster, processing the command buffer goes that much faster, the L1 and L2 caches have that much higher bandwidth, and so on. The only downside is that system memory 33% further away in terms of cycles, but the large number of benefits more than counterbalance that. As a friend of mine said, 'a rising tide lifts all boats.'"

In addition to the detailed explanation of teraflops, audio, and design, we also got a more detailed explanation on how backwards compatibility will work, which includes internal legacy modes for PS4 and PS4 Pro within the PS5's system. For more on our breakdown of Sony's Road to PS5 livestream, along with other updates on the Xbox Series X, be sure to check back with GameSpot for more updates.

Got a news tip or want to contact us directly? Email

Join the conversation
There are 11 comments about this story