AMD’S rDNA 2-based RX 6000 series graphics cards began to receive great attention as of their introduction, and the competition took on a new dimension. In the presentation of the red team, we saw both hardware and software innovations. In addition, three new models appeared: RX 6800, RX 6800 XT and RX 6900 XT.
One of the most notable among them was undoubtedly the Infinity Cache memory technology. Although GPUs, which are at the heart of video cards, are very important, memory (VM) is in second place during the importance.
All three graphics cards in the RX 6000 family have 16 GB of GDDR6 memory and a 256-bit memory interface. So it comes to mind that bandwidth may be very limited. At this point, Infinity Cache technology comes into play.
Although AMD does not go into much detail in its presentation, we can describe it as a large and very fast, on-chip cache structure, such as the L3 cache on the CPU. But conceptually, it is one-on-one for large operations that are used as a buffer for read and write operations from main memory and are much faster for data to be processed directly without having to go to VM. In addition, the GPU refers to the cache instead of waiting for data to come from the VM, improving the amount of work achieved on the GPU per clock cycle.
Internal caches for GPU use are not a new idea, especially for AMD. The company offered 32 MB eSRAM cache for Xbox One and Xbox One S SoCs. There was also an eDRAM on the Xbox 360 before it. But it was the first time a large cache appeared on a GPU used on the PC platform.
Having this kind of cache on the GPU also has significant advantages on the energy efficiency side. As for the performance value per Watt, this cache brings energy efficiency to a better point by reducing the amount of traffic to high-energy-consuming VM.
It also allows AMD to get rid of a small memory subsystem with fewer DRAM chips and fewer memory controllers. As a result, the power consumed decreases slightly.
It was said that memory bandwidth will increase by 117% with Infinity Cache Technology. As you know, when we switch to resolutions such as 2K, 4K and 8K, memory begins to become much more important.
The NVIDIA side solved this issue with GDDR6X memory technology. AMD, on the other hand, preferred a fast cache design, adopting a different path.
AMD says Infinity cache is based on the L3 cache design of the Zen CPU, meaning AMD is the fruit of Cross-working between CPU and GPU teams. As you know, HMB-based memory is quite expensive, although it is fast. The red team thinks Infinity Cache technology is a better engineering investment than going for this kind of expensive solution.
Finally, it was mentioned that this technology will also make a significant contribution to ray tracing performance. Because many of the processed data sets are stored closer to the compute units inside the GPU.