Specialists at Digital Foundry shared details about the capabilities of the PlayStation 5 Pro and the characteristics of the console's graphics chip, which they obtained from Sony documents intended for developers.
- The PS5 Pro will feature 30 workgroup processors (WGP), compared to the original console's 18 WGP. This will give the device a performance of 33.5 teraflops, a significant increase from the PS5's 10.23 teraflops.
- Despite the substantial difference in teraflops, the real-world performance boost in the PS5 Pro will be 45%, not the 227% that Sony noted in its documents. This discrepancy is attributed to the specifics of the RDNA 3 architecture and how instructions are processed.
- The standard clock speed of the PS5 Pro's GPU will be around 2.18GHz, slightly lower than the PS5's 2.23GHz. However, when overclocked, the new console's GPU can reach 2.35 GHz, theoretically providing a maximum performance of 36.1 teraflops. Digital Foundry noted that the GPU rarely reaches peak power, especially in demanding games.
- The PS5 Pro will feature 56 active processing units, an increase from the original console's 36. Digital Foundry believes this will result in a significant performance boost in existing games, despite the slight reduction in base clock speed.
- Additionally, the PS5 Pro GPU's L1 cache will double from 128KB to 256KB, while the L0 cache will increase from 16KB to 32KB. These changes accommodate the increased number of compute blocks and improve performance in ray tracing tasks.
- The PS5 Pro's video chip will include some features of DirectX 12 Ultimate that were not present in the original console. This includes hardware support for shading optimization technology, additional capabilities for hybrid MSAA anti-aliasing, and full compatibility with mesh shaders.
- Sony will also offer developers an easy way to integrate PSSR image scaling technology into existing games, using the same dataset as FSR 2. This allows developers to add PSSR support to their games at no cost, improving performance.
