Gp100: nvidia announces its new generation of graphics chips

Taking advantage of its 2016 GTC event, Nvidia has finally unveiled its graphics core (GPU) called GP100, which would mean the top of the range for the Pascal generation, an architecture that Nvidia has been working on for 2 years for its new graphics cards.

The new Pascal GP100 core will benefit from the innovative 16nm manufacturing process with technical characteristics that slightly differ from the rumors that had emerged from the hypothetical GTX 1080. The new GP100 chip will eventually possess some 3, 840 shaders along with 240 texturing units and an impressive 4, 096-bit bus.

Probably the most surprising thing about this new Nvidia graphics chip is that it will use the new high-performance HBM2 memories with a maximum configuration of 16GB, this amount of memory is surely reserved only for the TITAN version, while for the GTX 1080 TI it is used about 8GB of HBM2 memory, more than enough for all current and future video games.

Below these lines we can see a graph detailing more about the characteristics of the new Pascal GP100 core, among them the NVLink technology is named that seeks to enhance the scalability of the GPU, and unified memory, which allows simplifying the programming processes.

GP100 the most powerful graphics chip today

If we go into comparisons with the previous Nvidia (Maxwell) architecture, we see that the number of transistors almost doubles, from 8 billion Maxwell to 15.3 billion Pascal transistors in the same package size but also requiring more power than the computer, with a TDP of 300W compared to the 250W required by Maxwell chips.

The Pascal GP100 chip in detail

With the data on paper, the Nvidia GP100 chip will easily be the most powerful GPU in the world, although it will still have to wait a few more weeks to know the launch of the first Nvidia graphics cards with GP100 and especially the price.