It has been a while since we’ve heard any major developments regarding HBM2 (High Bandwidth Memory 2). According to SK Hynix, HBM2 will be made available to its customers by Q3 2016; which spans from July to September. Currently, the only GPU to come with the HBM2 memory standard would be Nvidia’s Tesla P100, which unfortunately, isn’t meant for general consumers.
SK Hynix’s announcement means that PC gamers and enthusiasts should expect to see HBM2-equipped graphics cards being announced by both AMD and Nvidia. Interestingly, the memory capacity of the HBM2 memory chips that’ll be supplied by SK Hynix is 4GB (4 Hi-stack form-factor); and it’ll come in two speed configurations: 1.6GBps and 2GBps. A combination of four HBM2 stacks would result in a 4096-bit graphics card with 16GB VRAM.
Despite the encouraging developments on the highly anticipated HBM2 memory standard, one would most likely only see a consumer-grade HBM2-equipped graphics cards around 2017 – at least according to AMD (Vega). Nvidia on the other hand, could release a HBM2 graphics card (GTX 1080 Ti?) earlier, but it seems unlikely.
Regardless, it would be interesting to see just how good HBM2 really is, especially when one considers the fact that HBM2 is significantly faster compared to Micron’s GDDR5X memory standard that’s used in Nvidia’s GTX 1080 graphics card.