The new architecture underlying Nvidia's Titan V GPU was designed to "push the outer limits of high performance computing and AI," according to co-founder, president, and CEO Jensen Huang. Huang unveiled his company's latest GPU, the $2,999 Titan V, yesterday at the Neural Information Processing Systems (NIPS) conference running Dec. 4-9 in Long Beach, Calif.
Nvidia described the Titan V's GPU architecture as "the world's most advanced." The company said the processor features 21.1 billion transistors that support 110 teraflops of deep learning computing power, more than nine times that of Nvidia's previous GPU for artificial intelligence (AI).
Available now, with a limit of two per customer, the Titan V (pictured above) is designed for developers who "want to use their PCs to do work in AI, deep learning and high performance computing," Nvidia said. Buyers also get access to software optimized for AI, deep learning, and high-performance computing through Nvidia's GPU cloud.
Designed for 'Groundbreaking' Computing
Aimed at scientists and researchers working on the cutting edge of AI, the Titan V is the latest example of Nvidia's mission to help users "do things that would otherwise be impossible," Huang told NIPS attendees yesterday. "Our ultimate purpose is to build computing platforms that allow you to do groundbreaking work."
After 20 randomly selected attendees were presented with a Titan V during his presentation, Huang introduced an on-stage orchestra that performed a Star Wars-inspired piece of music. The piece was composed by AI from Aiva Technologies, a Luxembourg-based "virtual artist" startup that also created a musical composition for Huang's keynote at the Nvidia GPU Technology Conference earlier this year.
Aiva used multiple GPUs, including Nvidia's Titan Xp, to compose its latest piece. After the NIPS performance, Aiva co-founder and CEO Pierre Barreau said that he planned to add a Titan V to his company's tool chest as well.
A Sign of Titan's 'Market Shift'
The new Titan V uses Nvidia's Volta GPU architecture, which also supports predecessor GPUs, including the Nvidia Tesla V100. In October, Amazon Web Services revealed that it is using multiple Tesla V100s to support its new EC2 P3 instances for compute-intensive AI, scientific research, finance, and other applications.
"Titan V's Volta architecture features a major redesign of the streaming multiprocessor that is at the center of the GPU," Nvidia said yesterday. "It doubles the energy efficiency of the previous generation Pascal design, enabling dramatic boosts in performance in the same power envelope."
Unlike many of Nvidia's other GPUs, the Titan V is not aimed at gamers but at researchers and scientists with high-performance computing requirements. It's a sign that the company's target market for the Titan series "has shifted and will continue to evolve," Patrick Moorhead of Moor Insights & Strategy wrote yesterday in Forbes.
"Titan V is targeted at machine learning scientists who want to conveniently buy the card and install it into their desktop PC," Moorhead said. "This means the researcher doesn't need a special server, storage or networking."
Moorhead called Nvidia's Titan V announcement a "bombshell," adding, that the GPU "appears to be a powerful, capable new offering from Nvidia, great for simple entry into desktop-based machine learning training and inference."
Image credit: Product shots from Nvidia.