Microsoft has revealed some new data about how the Xbox Series X will utilize its advanced, high-speed storage capacity and some of the GPU efficiency enhancements we can expect to see. Some of these technological developments could have a direct bearing on the future of PC GPU technology, particularly if they were developed in partnership with AMD.
Microsoft’s latest blog post details what it calls the Velocity Architecture. The first section of the blog just steps over the data we already know, as far as speeds and feeds are concerned: 12TFLOPs, 16GB of RAM, >4x the CPU performance of the Xbox One, and so on. The Velocity Architecture is made of four components: A custom NVMe SSD, hardware-accelerated decompression blocks, a new DirectStorage API layer, and a brand-new technology dubbed Sampler Feedback Streaming. It’s this last that looks so interesting as a potential PC technology.
The SSD delivers 2.4GB/s of raw storage / 4.8GB/s of compressed data, which is 50 percent the rate Sony has specified for the PlayStation 5. Microsoft can’t resist a little FUD, writing: “Traditional SSDs used in PCs often reduce performance as thermals increase or while performing drive maintenance.”
The first means you have a terrible cooling system, while the second is not known to be an issue under most mainstream storage workloads.
Next up, there are the hardware-accelerated decompression blocks. Here’s where the new BCPack algorithm comes into play — but the XSX also supports the industry-standard LZ decompressor and can run both simultaneously. This is where Microsoft’s derived 4.8GB/s of bandwidth comes from. This a 100x uplift compared with current consoles and, Microsoft says, would require more than 4 Zen 2 cores to deliver in pure software.
The third piece of the puzzle is the new DirectStorage API. Here’s how Microsoft describes it: “We added a brand new DirectStorage API to the DirectX family, providing developers with fine grain control of their I/O operations empowering them to establish multiple I/O queues, prioritization and minimizing I/O latency.” Developers will have enormous flexibility to take advantage of the hardware’s underlying I/O performance.
These gains are important to the overall XSX system design. Unlike in previous consoles, RAM loadout in the Xbox Series X is only increasing by 2x compared with the Xbox One and by just 1.33x compared with the Xbox One X. That’s far less than in previous generations; the Xbox 360 had 512MB of RAM, while the Xbox One had 8GB. That’s a 15.625x increase from the 360 to the One, and a 2x increase from the Xbone to the XSX, best-case.
The only way to take another leap upwards in graphics quality without adding huge amounts of system memory is to use SSD storage very, very efficiently.
Sampler Feedback Streaming
GPUs today use a technique known as mipmapping to optimize game textures for different detail levels and resolutions. A player standing a long way away from an object will see one version of it, while moving closer will instruct the game to swap to a different mipmap. Texture resolution has to increase in order to provide higher-quality visuals at a closer range, but the way mipmaps are designed requires that a GPU load an entire mipmap into memory, even if it only needs a relatively small piece of it.
Determining how much RAM games actually use is a difficult endeavor, even today, but Microsoft was able to use specialized hardware inside the Xbox One X to measure just how the GPU used texture memory internally. What they found was startling: GPUs commonly access less than 1/3 of the texture data they require to be loaded into RAM. This has some significant implications for how more-efficient memory usage patterns could allow GPUs to consume less power or to spend more of what they do consume on improving performance. Microsoft writes, “A single scene often includes thousands of different textures resulting in a significant loss in effective memory and I/O bandwidth utilization due to inefficient usage.”
The XSX dodges this problem with new technology that allows it to load sub-portions of mipmaps rather than the entire texture. MS writes that this allows for 2.5x the effective I/O throughput and memory usage, above and beyond the raw hardware’s capabilities. This technology could help explain how AMD is claiming a 1.5x increase in performance/watt with RDNA2. I’m not assuming that there’s an automatic link between the two, because Microsoft might have an agreement with AMD that reserves certain tech specifically for the Xbox. But this could be part of the secret sauce that helps RDNA2 match or exceed Ampere’s performance.
Microsoft released the video above as part of the Velocity Architecture announcement, so check it out if you want more details. I’m now quite curious to see which features upcoming RDNA2 GPUs will and won’t support, relative to what we see the Xbox Series X deliver when it launches later this year.
- Microsoft Pushes Publishers to Offer Next-Gen Game Upgrades for Free
- Leak Suggests Second, Weaker Xbox Series X Console
- Microsoft’s New Xbox Series X Compression Tech Might Solve Next-Gen Storage Woes