AMD's Radeon RX Vega GPU architecture brought many notable features to the table, but the one is very special is HBCC - or "high-bandwidth cache controller". In this article, we will be taking a quick look at what HBCC is and talk about what it could offer in the future.
At its Capsaicin & Cream event held alongside GDC in February, AMD has made a big fuss about RX Vega's HBCC (High-Bandwidth Cache Controller), promising huge benefits to the gamers. At its RX Vega editor's day a couple of weeks ago, Radeon chief Raja Koduri called it his favorite feature of the new Radeon RX Vega architecture, as reported by Techgage.
During his presentation, Raja has claimed the potential of seeing +50% gains to average framerates, and a staggering +100% boost to minimum framerates in our gaming. In the demo he gave, Deus Ex: Mankind Divided was run on a 4GB AMD Radeon RX Vega, giving a good example of how HBCC could help with big games on more modest hardware.
For the getting things clearer, HBCC allows your GPU to treat VRAM as a last-level cache, so that a request for data not currently located in VRAM can be pulled into Vega's HBC for immediate access while simultaneously flushing out data which is no longer needed. In addition to describing how the feature functions several reviewers also did a bit of testing to determine the real world effect of enabling HBCC in games and benchmarks.
In coming time, an application to utilize this mechanic and to maximize GPU efficiency in huge workloads could be seen. The pool of data that the Radeon Software will create is called the "HBCC Memory Segment" (HMS), and on our 32GB test PC, the minimum value that might be selected was ~11GB (max was ~24GB).
In the current AMD Vega driver (17.8.1), the function is disabled by default. This can be changed under » Games Global Settings « with one click under » HBCC Memory Segment «, as reported by Game Star.
In the future, this HBCC feature will enable to enjoy massive worlds without having to spend thousands of dollars on increasing the video memory. However, if the system RAM holds data that the GPU happens to need, it'll have to flow on down through the PCIe bus before the GPU can grab it up.
