AMD, Nvidia GPUs Using HBM2 Won’t Launch Anytime Soon Thanks to High Costs
Rumor has it that the next gen High Bandwidth Memory (HBM2.0) based GPUs won’t ship anytime soon. Both AMD and Nvidia will stick with GDDR5(X) memory for their upcoming Polaris and Pascal based graphics cards, respectively, at least for the time being.
The HBM2 for high end GPUs is expected to happen in late Q4 2016, but it won’t ship until 2017 in any volume. The word on the street is that HBM2 is currently in a very limited supply, which is the reason behind making things expensive.
The first graphics chip to support this next iteration of memory will be AMD Vega 10 GPU, which has been known by the codename Greenland for some time. Even according to AMD’s official GPU roadmap, Vega graphics architecture will hit the shelves next year.
Not only does HBM2 operate at twice the speed of its predecessor for double the memory bandwidth, but it also scales to capacities 8 times larger than the first generation.
Recently, SK Hynix released a side by side comparison of GDDR5, HBM and HBM2, revealing the latter of three has a lot of capacity for its size when compared to the old memory standards.
The photo compares the size of 8GB GDDR5 memory against 1 GB of HBM memory and 4GB of HBM2 memory. As it’s obvious, HBM has huge capacity for its size compared to GDDR5 memory, while HBM2 has 4x more capacity compared to the first generation.
Have anything interesting to add to the story? Let us know in the comments section below.
Gohar is the lead editor at TechFrag. He has a wide range of interests when it comes to tech but he's currently spending a big chunk of his time writing about privacy, cyber security, and anything policy related.