AMD has published benchmarks for the Radeon RX 6000, which show the performance gains by using Smart Access Memory in conjunction with the latest Ryzen 5000 series processors. But NVIDIA already has something to do with this advantage. According to a statement from the company to Gamer’s Nexus, it will introduce a similar feature that will improve the efficiency of transferring data between the GPU and the processor. In fact, in the NVIDIA lab, this functionality is already working on Ampere graphics cards.

In addition, NVIDIA claims that its technology will be universal and will be able to work equally well with Intel and AMD processors, including through the PCIe 3.0 bus, while AMD’s solution requires an AMD Ryzen 5000 series processor, a motherboard on the X570 chipset and the Radeon RX 6000 graphics accelerator.

The description says that Smart Access Memory allows the CPU and GPU to exchange information over a wider PCIe channel. The technical details are unknown, and AMD simply says that the CPU and GPU are usually limited to a 256MB “aperture” for data transfer, which makes game developers’ hands tied, and also requires frequent switching between the processor and main memory if the data exceeds this limit. As a result, efficiency drops, and productivity decreases. Smart Access Memory removes that barrier, AMD says.

This description is like changing the BAR size in PCIe – a standard feature described in the PCIe specifications, according to NVIDIA. And if the graphics processor supports it, then you can increase performance by simply adjusting the appropriate parameter in the BIOS of the video card and motherboard. NVIDIA said its GPUs support this feature, although it requires activation. As for processors, any Intel or AMD chip that works with the PCIe bus should be able to enable this technology.