A hardware matchup to bridge the gap between Quantum computing and quantum computers. Though a true quantum computer has yet to hit the public marketplace, IBM and Nvidia have teamed up to make one of the most powerful computing devices available.
Specifically catering towards machine-learning and Artificial Intelligence, the new system is designed for high performance applications. Combining the capacity of (two) IBM’s Power8 CPUs and (four) Nvidia’s Tesla P100 GPUs, it is capable of “21 teraflops of half-precision floating point performance”. After hitting Google to find out what a teraflop actually is, it is a trillion floating point operations within a second, and to put it simply that’s a lot of calculations. Recently GPUs have been recognized as the best systems for simulating quantum computations, where the current use for visual construction has many similarities. It also surpasses the limits of traditional data buses by using an integrated NVLink Interface which speeds up transmission between processors by up to 2x.
It’s called the IBM Power Systems S822LC for High Performance Computing and as this platform rolls out it will be great to see the many new and innovative applications that it finds in industry.
To learn more about the IBM and NVIDIA collaboration, click here – https://www.techcrunch.com/2016/09/08/ibm-and-nvidia-team-up-to-launch-a-new-server-purpose-built-for-machine-learning/
The post Machine Learning At The Forefront appeared first on Partstat Blog.