Microsoft is building fast, low-power neural networks with FPGAs

Microsoft on Monday released a white paper explaining a current effort to run convolutional neural networks — the deep learning technique responsible for record-setting computer vision algorithms — on FPGAs rather than GPUs.

Microsoft claims that new FPGA designs provide greatly improved processing speed over earlier versions while consuming a fraction of the power of GPUs. This type of work could represent a big shift in deep learning if it catches on, because for the past few years the field has been largely centered around GPUs as the computing architecture of choice.

If there’s a major caveat to Microsoft’s efforts, it might have to do with performance. While Microsoft’s research shows FPGAs consuming about one-tenth the power of high-end GPUs (25W compared with 235W), GPUs still process images at a much higher rate. Nvidia’s Tesla K40 GPU can do between 500 and 824 images per second on one popular benchmark dataset, the white paper claims, while Microsoft predicts its preferred FPGA chip — the Altera Arria 10 — will be able to process about 233 images per second on the same dataset.

However, the paper’s authors note that performance per processor is relative because a multi-FPGA cluster could match a single GPU while still consuming much less power: “In the future, we anticipate further significant gains when mapping our design to newer FPGAs . . . and when combining a large number of FPGAs together to parallelize both evaluation and training.”

In a Microsoft Research blog post, processor architect Doug Burger wrote, “We expect great performance and efficiency gains from scaling our [convolutional neural network] engine to Arria 10, conservatively estimated at a throughput increase of 70% with comparable energy used.”

fpgacnn

This is not Microsoft’s first rodeo when it comes deploying FPGAs within its data centers, and in fact is a corollary of an earlier project. Last summer, the company detailed a research project called Catapult in which it was able to improve the speed and performance of Bing’s search-ranking algorithms by adding FPGA co-processors to each server in a rack. The company intends to port production Bing workloads onto the Catapult architecture later this year.

There have also been other attempts to port deep learning algorithms onto FPGAs, including one by State University of New York at Stony Brook professors and another by Chinese search giant Baidu. Ironically, Baidu Chief Scientist, and deep learning expert, Andrew Ng is big proponent of GPUs, and the company claims a massive GPU-based deep learning system as well as a GPU-based supercomputer designed for computer vision. But this needn’t be and either/or situation: companies could still use GPUs to maximize performance while training their models, and then port them to FPGAs for production workloads.

Expect to hear more about the future of deep learning architectures and applications at Gigaom’s Structure Data conference March 18 and 19 in New York, which features experts from Facebook, Microsoft and elsewhere. Our Structure Intelligence conference, September 22-23 in San Francisco, will dive even deeper into deep learnings, as well as the broader field of artificial intelligence algorithms and applications.

Facebook open sources data center efficiency code

Facebook has open sourced the code it uses to measure the energy and water consumption of its data centers, as well as the code for the dashboard that visualizes those readings in real time. Facebook first publicly shared the dashboard in April 2013 for its Prineville, Ore., and Forest City, N.C., data centers and it has been available online since. Facebook says cloud provider Rackspace helped get the code ready for open source and is considering implementing it for its data centers.

fbpue

Google to acquire connected home device maker Nest for $3.2B

Google just made another major acquisition — clean tech and home device star Nest Labs. The $3.2 billion purchase of former Apple executive Tony Fadell’s new company comes at a time the search giant is expanding its activities in science, robotics and energy.

New algorithms reduce the carbon cost of cloud computing

http://spectrum.ieee.org/computing/networks/new-algorithms-reduce-the-carbon-cost-of-cloud-computing

This article in IEEE Spectrum highlights some interesting research into making cloud computing more efficient by balancing the carbon footprints of global data centers and the latency of serving requests from those data centers. It’s an admittedly incomplete study, and one that’s probably more important to large web companies (e.g., Facebook) and cloud providers than to normal businesses just consuming cloud resources. Smarter load balancing could help providers cut operational costs and pass the savings onto users. Of course, large cloud users such as Netflix might want to research and develop their own systems, as well.