Nvidia Geforce Pascal can contain as many as 17 billion transistors

While Intel is in full swing with both Tri-gate transistors and 14 nanometer technology, AMD and Nvidia are still stomping on 28 nanometers. The reason is spelled TSMC, which has not yet succeeded in convincing with more advanced manufacturing processes for high-performance circuits such as graphics processors. Only at the beginning of 2016 can there be a change in that part.

Following the rollout of Geforce Maxwell, including the worsting GM200, Nvidia is looking forward to the next architecture. The name to add to the memory is Pascal, which will not only take the step to stacked memory circuits but is also said to go over to 16 nanometers with Finfet – TSMC’s equivalent of Intel Tri-gate.

Information is now coming in that indicates that Nvidia plans to take advantage of the new manufacturing technology to create more advanced graphics processors than ever before. The successor to the Geforce GTX Titan X could contain as many as 17 billion transistors.

That is more than double the current Nvidia GM200 of 8.0 billion transistors and also a substantial increase from AMD Fiji in the Radeon Fury X, which currently leads the league with 8.9 billion transistors. Despite this, the circuit to the surface is expected to be smaller than its predecessor Maxwell.

The new graphics processor is combined with the second generation of stacked memory chips from SK Hynix, known as HBM 2.0. It will allow up to 32 GB of memory to be squeezed and at the same time double the memory bandwidth from the first generation, which is used in the Radeon Fury series. In other words, the memory bandwidth will exceed one terabyte per second.

The first graphics processor with architecture Nvidia Pascal is expected to appear in early 2016.