NVIDIA 12nm FinFET Volta GPU Architecture Reportedly Replacing Pascal In 2017

One of the biggest moments in the history of 3D graphics is when NVIDIA released its Pascal architecture last year. The generational leap in graphics performance lived up to the hype with higher end parts negating the need to run multiple graphics cards in SLI for consistently smooth gaming at 4K. But while Pascal has barely had time to stretch its legs, NVIDIA is already looking ahead to its next GPU architecture, Volta, which may release later this year.

Rumor has it that NVIDIA has tapped TSMC to build Volta GPUs on a 12-nanometer FinFET manufacturing process. It is a loose rumor—basically a dude on a Beyond3D's forum who is listed as "site staff" in his profile—though it is certainly plausible. News and rumor site Digitimes previously reported that TSMC planned to introduce an enhanced and smaller variant of its 16nm technology, which TSMC later confirmed, and the site took that to mean 12nm.

NVIDIA GeForce GTX 1080

If that is the case, a 2017 release is entirely possible, unlike some previous rumors suggesting Volta might be built on a 10nm manufacturing process. For NVIDIA to base Volta on 10nm technology, it would likely be looking at a 2018 release. In case anyone is wondering, TSMC builds most of NVIDIA's Pascal GPUs, from mid-range gaming chips on up to data center GPUs, on a 16nm FinFET+ process; Samsung produces some of the lower end Pascal GPUs on a 14nm low power plus process.

The other reason to believe that Volta could stick its head out in 2017 is because IBM references the GPU architecture when outlining Summit, its next generation high performance computing (HPC) system. Here is a description pulled straight from the Oak Ridge National Laboratory website:
Summit will deliver more than five times the computational performance of Titan’s 18,688 nodes, using only approximately 3,400 nodes when it arrives in 2017. Like Titan, Summit will have a hybrid architecture, and each node will contain multiple IBM POWER9 CPUs and NVIDIA Volta GPUs all connected together with NVIDIA’s high-speed NVLink. Each node will have over half a terabyte of coherent memory (high bandwidth memory + DDR4) addressable by all CPUs and GPUs plus 800GB of non-volatile RAM that can be used as a burst buffer or as extended memory. To provide a high rate of I/O throughput, the nodes will be connected in a non-blocking fat-tree using a dual-rail Mellanox EDR InfiniBand interconnect.
We've bolded the relevant bits, which point to Summit releasing in 2017 using a combination of IBM Power9 CPUs and NVIDIA Volta GPUs.

So at least in the supercomputing sector, it looks like Volta has a good chance of seeing daylight later this year. Whether or not it trickles into the consumer before 2018 is not known and may depend on well AMD's Vega architecture fares. If Vega fails to dethrone Pascal as the performance champion, NVIDIA may decide to give the architecture a longer shelf life before introducing Volta.

In any event, it looks to be another exciting year in graphics.