PlayStation 5 Could Potentially Yield More Raw Power for Developers Over the Xbox Series X?
written by Jih-Wei Peng
Recently Sony revealed the much-awaited details surrounding their next-generation PlayStation 5 gaming hardware. After the consoles’ lead architectural designer Mark Cerny broke down a thorough glimpse into the specifications of the system, there is more here to deduce than simply looking at the overall teraflop yields.
Aside from the obvious approach of seemingly being as cost-effective as possible, the PlayStation 5, due to its efficiently streamlined construction, associated with its CPU/GPU and audio dedicated SPU’s, among other beneficial features is truly something special. Based on how the development community decides to take advantage of Sony’s new hardware, it could potentially yield greater benefits over the competing next-gen hardware offered by Microsoft.
First, we must understand that a GPU teraflop rating is derived from a mathematical formula that no card can actually achieve. Sure, it can make some sense to discuss teraflops when talking about HPC computing, yet GPUs are not simply a collection of cores. Other critical aspects of design, including memory bandwidth, cache size and distribution, clock speed, bus width, and pixel/texel fill rates have huge impacts on how well a GPU can game. However, they don’t affect their theoretical FLOPS rating at all.
Many gaming enthusiasts have been giving the early power nod to Microsoft and their impressive spec detailed next-generation Xbox Series X, this largely due in part to the hardware’s 12.1 TFLOPS of computational power. Compared to the PlayStation 5 with its 10.28 TFLOPS of computational power, on the surface, those who bleed green might have much to celebrate with a 1.82 computational power advantage. However, as you peel away the layers of what is happening under the hood, Sony and Microsoft’s next-generation rigs might be much closer in raw power than you think.
Based on Mark Cerny’s overview of the PlayStation 5 engineering, Sony is going for a more overall feature-set. Interestingly, as with the PlayStation 3 Cell architecture, once again Sony is completely removing the audio workload off the main CPU/GPU. Sony has decided to create customized 3D software fully supported with audio SPU’s. As anyone in development understands, the audio takes up a significant amount of processing power. By housing the audio within a dedicated area, this frees up the entire GPU for what it is designed to do.
Today, depending on how developers utilize their 3D audio, this could be a substantial drain on the CPU/GPU. On average many games are utilizing upwards of 20% of the computational power and based on how the Xbox Series X is allocating its CPU/GPU for the next-gen, nothing here has changed.
On paper 12 TFLOPS is an awesome number, however, when you remove 20% of that computational power for audio purposes, you are left with about 9.6 TFLOPS for everything else necessary to bring your game to life. Still and impressive resource of power allocation. Yet, for the PlayStation 5, the GPU has no audio responsibility, allowing the creators to utilize as much of the systems raw 10.28 TFLOPS of power.
According to Digital Foundry Microsoft’s design decisions with Xbox Series X means developers must be more mindful of the potential power consumption that could impact clock speeds and lower performance. For the PlayStation 5, this means developers can hit GPU frequencies much higher.
Also, Sony is going with a whopping 448GB/s of RAM with a higher I/O throughput over the competition. From the looks of it, Sony is trying to eliminate potential bottlenecks in the main threads as much as possible. No audio computations, much higher RAM bandwidth and a much higher I/O. At a glance, both the next-generation consoles should be more than capable of performing at 4K resolution while running at a smooth 60 frames per second for the vast majority of games.
Of course, at the moment, everything on the specifications sheets are theoretical, with a wealth of overall computational power potential for both Sony and Microsoft’s next-generation hardware. As always is the case, the true determining factor of raw power output should be fully realized once the actual games become playable.
Until then, here’s to both Sony and Microsoft offering the best next-generation systems possible.