Nvidia and AMD recently released their newest generation of graphics cards, based on a 16nm FinFet and 14nm FinFet manufacturing process respectively.

The new Pascal and Polaris architectures bring with them a massive boost in power efficiency and performance.

However, consumers could see another GPU generation sooner than usual, with AMD set to launch Vega in early 2017.

AMD has stated that Vega will be using a new type of VRAM technology called HBM2.

HBM (High-Bandwidth Memory) provides massive improvements in VRAM performance and has already been implemented on a number of Radeon cards, such as the R9 Nano and Fury X.

HBM2 is a subsequent improvement over HBM, offering more VRAM and better bandwidth, and is used in Nvidia’s Tesla P100 Pascal GPUs.

HBM3

Both SK Hynix and Samsung are working on HBM3, which is set to offer twice the bandwidth of HBM2 at a lower cost than current VRAM technology.

While HBM offers 256GB/s of bandwidth per layer, HBM3 doubles that amount of bandwidth-per-layer to 512GB/s.

This means that when graphics cards begin implementing HBM3, they could boast up to 64GB of VRAM in a standard form factor.

While it may be a while until we see HBM3 used in consumer GPUs, specialised graphics cards such as Nvidia’s Tesla lineup could make use of this technology in the near future.

Low-cost HBM

In addition to this new technology and its increased performance, SK Hynix and Samsung are also working on low-cost HBM technology.

This VRAM technology is specifically aimed at consumers and dramatically lowers the cost-per-GB of HBM while maintaining performance.

Low-cost High-Bandwidth Memory will feature increased pin speeds and similar bandwidth at a much lower cost than HBM2.

Consumers can expect this type of HBM to be implemented in GPUs soon after it is completed.

More gaming news

Here is why the prices of games and graphics cards in South Africa are insanely high

What the NSA being hacked means for cybersecurity