At the 2017 Neural Information Processing Systems conference, Nvidia CEO Jen-Hsun Huang announced the Titan V, another high-end GPU to its line-up. This card is based on Nvidia’s Volta architecture, featuring 21.1 billion transistors on an 815 mm chip that can produce a full 110 teraflops of computing.
It brings tons of power for dealing with AI and scientific simulation processing which sells for $2,999. It is designed primarily for machine learning, big data analysis, and artificial intelligence development but not for gamers. Titan V is packed with 12GB of HBM2 memory, 5120 CUDA cores, and 640 tensor cores that offer up to 9 times the deep-learning performance of its predecessor. Nvidia calls its new compute units Tensor Cores that are designed specifically for deep learning, with independent integer and floating-point data paths that can operate in parallel.
Nvidia gave the Titan V a distinct black and gold color which looks pretty cool.
“Our vision for Volta was to push the outer limits of high-performance computing and AI. We broke new ground with its new processor architecture, instructions, numerical formats, memory architecture and processor links,” said Huang. “With TITAN V, we are putting Volta into the hands of researchers and scientists all over the world. I can’t wait to see their breakthrough discoveries.”
It seems that the Titan V’s specs are similar to Tesla V100’s because both cards have 5,120 compute cores and 640 “tensor cores” that are oriented for machine learning. Titan V can boost to 1.455GHz, slightly more than V100’s 1.370GHz. However, they both vary in the memory subsystem. While the Titan V has 12GB of HBM2 memory, with a 3,072-bit memory bus, Tesla V100 has 16GB of HBM2 and a 4,096-bit bus.
We have no clue on when Volta will make its way into more gaming focused GPU’s. While the Pascal architecture that powers Nvidia’s GeForce 10-series gaming GPUs introduced over a year and a half ago with the GTX 1080, still remains one of the best gaming GPUs available.
The Titan V is now available for data scientists on websites and is limited to two per customer.