NVIDIA Goes For AI With Latest Volta Architecture
NVIDIA CEO Jensen Huang's keynote at the annual GPU Technology Conference (GTC), included mini-tutorials on various types of machine learning, and an endless number of plugs for AI-based applications hosted on Nvidia GPUs. However, the highlight of the event was the announcement of Nvidia's new Volta architecture and V100 chip.
In raw specs, the launched Nvidia V100 is impressive. With 21 billion transistors crammed into its 815 square millimeters die, Huang claims it is the largest and most complex chip which can be created with current semiconductor physics. At a cost of $3 billion in R&D, the final chip is made-up using a 12nm process by TSMC and runs on the highest-speed RAM available from Samsung. After the keynote, Nvidia has also explained that it used 12nm and such a large die size since it deliberately wanted to create the most stylish chip possible, as reported by Extreme Tech.
Latest NVIDIA Volta also features over 5,000 GPU CUDA cores, 300 GB of system communications bandwidth through six high-speed NV Link interconnects. It also boasts 16 GB of the second generation HBM2 on TSMC's new 12FFN manufacturing process technology.
NVIDIA has pioneered the development of artificial neural networks through deep learning with the company's GPUs and CUDA software platform. NVIDIA is engaged with over 40,000 companies and over a half a million developers for neural network applications. However, while the GPU has proven very effective in parallel processing as the addition of shader cores, TIRIAS Research has maintained that even NVIDIA would need to eventually migrate to architectures dedicated to AI while preserving its tools and ecosystem to advance platforms further, as noted by Forbes.
The NVIDIA Volta announcement is important because of this mark a change for the entire industry. The competition will also have to follow suit and innovate in this new processing vector in the near future