This site may generate affiliate commissions from the links on this page. Terms of Use .

HBM2 had a difficult period of 12 months. The standard of memory has not followed the path of its predecessors. If that was the case, we should have seen widespread deployments of AMD and Nvidia on mid-range and high-end products. Instead, the AMD Vega 56 and 64 are currently the only mainstream GPUs to support the memory standard.

One of the rumors of HBM2 last year was that the clock speeds were well below the AMD GPUs reach the expected bit rates of 2 Gbps per pin (Hynix adds speeds of 2.0 Gbps to its HBM2 offerings in August 2016, then deletes them in 2017). Samsung, however, may have found a solution to this problem, and plans to introduce a new HBM2 clocked at 2.4 Gbps per pin.

 SAMSUNG HBM2_B "width =" 640 "height =" 453 "/>  </p>
<p> The new HBM2, codenamed Aquabolt, would be clocked 1,2 times higher than the previous HBM2 and 1.5 times higher than AMD RX Vega 56 RAM. The other great benefit for higher pin bandwidth theoretically reduces the need for additional HBM2 batteries </p>
<p> This ability could be more important than you think: Rumors have suggested that the manufacturing difficulties of HBM2 have focused on the interposer – the difficulty of aligning and connecting the vias going through the silicon that pass through the stacks of memory and into the interposer itself before connecting with the GPU Some of you will remember the following document from the various discussions of HBM on AMM </p>
<p>  <img class=





Source link

LEAVE A REPLY

Please enter your comment!
Please enter your name here