INFRA
INFRA
INFRA
IBM and Nvidia launch PowerAI, a hardware-optimized deep learning toolkit
IBM Corp. and Nvidia Corp. today revealed the next step in their plan to capitalize on the rapid adoption of artificial intelligence.
At the annual SC16 supercomputing conference today, the companies are showcasing a jointly developed toolkit dubbed PowerAI that contains custom versions of Torch, Theano and Caffe, three of the most popular open-source deep learning frameworks in the industry. The distributions have been optimized to run on the much-touted S822LC for HPC server that IBM unveiled in September. According to Big Blue, the machine is specifically geared towards running artificial intelligence and scientific workloads that rely on graphics processing unit chips to carry out their calculations.
To support this requirement, the S822LC for HPC combines two of IBM’s homegrown POWER8 processors with four accelerators from Nvidia’s Tesla P100 series. The chips are connected by an interconnect known as NVLink that the semiconductor maker developed to speed up the flow of between its GPUs and third-party processors. According to the company, the technology provides up to 12 times faster transfer times than the PCIe interfaces normally used for the task.
The result is that the server can run artificial intelligence workloads more efficiency than competing GPU-optimized servers such as Hewlett-Packard Enterprise Co.’s recently unveiled Apollo 6500, according to IBM and Nvidia, and it works even faster when paired with today’s newly introduced PowerAI toolkit. In an internal test carried out by IBM, the machine provided twice as much performance as the regular configuration of the S822LC while running the AlexNet image recognition system on Caffe.
PowerAI is available for free to organizations that deploy S822LC for HPC. Early adopters include cloud provider Nimbix and the Human Brain Project, a research initiative funded European Commission to advance the study of neuroscience. IBM will likely add support for more artificial intelligence frameworks to the toolkit over time as adoption of the server widens and customer requirements become more diverse. The first technology on the roadmap will likely be Google’s TensorFlow engine, which is rapidly gaining ground on traditional alternatives such as Caffe.
Image via Pixabay
Since you’re here …
… We’d like to tell you about our mission and how you can help us fulfill it. SiliconANGLE Media Inc.’s business model is based on the intrinsic value of the content, not advertising. Unlike many online publications, we don’t have a paywall or run banner advertising, because we want to keep our journalism open, without influence or the need to chase traffic.The journalism, reporting and commentary on SiliconANGLE — along with live, unscripted video from our Silicon Valley studio and globe-trotting video teams at theCUBE — take a lot of hard work, time and money. Keeping the quality high requires the support of sponsors who are aligned with our vision of ad-free journalism content.
If you like the reporting, video interviews and other ad-free content here, please take a moment to check out a sample of the video content supported by our sponsors, tweet your support, and keep coming back to SiliconANGLE.
