Nvidia’s Volta architecture is now rumoured to release in Q3 2017

Nvidia's Volta architecture is now rumoured to release in Q3 2017

Nvidia’s Volta architecture is now rumoured to release in Q3 2017

 
Nvidia’s Volta architecture is now rumoured to release in Q3 2017, moving the launch forward from their previously road mapped launch window of “early 2018”.  
 
Current reports state that Volta will release as the Nvidia GTX 20 series of GPUs, with Nvidia marketing this GPU series heavily for its use of stacked DRAM memory in the past, a memory type that is better known as HBM2. 
 
At this time it is unknown if these rumours are true, as Nvidia has not ever hinted at a pre-2018 release for Volta recently, with the usual hype mongers stating that “Nvidia is scared of Vega” and that they are hastening the GPU’s launch in order to try and take sales from AMD. 

  

Nvidia's Volta architecture is now rumoured to release in Q3 2017

 

Previous rumours have suggested that Volta will be based on TSMC’s 12nm manufacturing process, a process that is, in fact, an improved version of their 16nm process. This should make Volta more power efficient than Pascal, even when discounting other improvements that Nvidia has made to Volta’s design. 

If Volta is releasing in 2017, we can assume that Nvidia will not be using GDDR6 memory, since the earliest estimates for the release of that memory type is in early 2018. 

One other possibility is that Nvidia will announce Volta in Q3 2017 for HPC/server/AI compute market, releasing a high-end product in limited quantities to select partners. This is pretty much what Nvidia did with the release of their Pascal P100 server GPU, releasing “Big Pascal” to the compute market long before it ever reaches consumers.

At this time it is hard to tell what exactly Nvidia is planning with Volta, as it is far too early to tell if these rumours or true and if AMD’s Vega architecture is even worthy of such a large reaction from Nvidia. 

 

You can join the discussion on Nvidia’s Volta architecture on the OC3D Forums. 

Â