While AMD might have a priority deal with SK Hynix for High Bandwidth Memory 2 (HBM2), it appears that Nvidia will have a deal with both SK Hynix and Samsung for HBM2 on the upcoming Pascal GPU.

According to a new report coming from Business Korea, both Samsung and SK Hynix will mass produce second-gen High Bandwidth Memory for Nvidia Pascal GPU. The mass production is scheduled for Q1 2016 with samples and reliability tests scheduled for later this year.

With AMD securing SK Hynix as its primary source for High Bandwidth Memory, it is good to know that Nvidia will have a second source which will certainly help with its allocation.

The new second-generation High Bandwidth Memory will bring significant improvements including higher density, higher performance and lower power. The most important update is the higher density as the first-generation HBM was limited to 8Gb, leaving AMD's Fury lineup with 4GB of HBM memory, while the maximum chip capacity for HBM2 is at 32Gb, making 16GB graphics cards possible.

In addition to the HBM2, Nvidia will also use NVLink interconnect technology on Pascal GPUs, which will, according to Nvidia, bring improvements of 5 to 12 times compared to standard PCIe gen3.

According to the latest reports, Nvidia's Pascal GPU is scheduled for release sometime in mid-2016.





Source: Hexus.net.