Baidu Research, a division of Baidu Inc. (NASDAQ: BIDU), today unveiled the next generation of DeepBench, the open source benchmark tool, which now includes the measurement of deep learning inference across different hardware platforms in addition to training.
Baidu Research released the initial version of DeepBench in September 2016, the first tool opened up to the wider deep learning community to evaluate how different processors perform when they are used to train deep neural networks. Since its initial release, several companies have used and contributed to the DeepBench platform, including Intel, NVIDIA, and AMD.
Following positive feedback from peers across the AI industry and academia, Baidu Research has now incorporated requests to include the measurement of deep learning inference in addition to training.
Inference involves using a previously trained model to make predictions on a new data set.
Speed is key to training neural networks, and the first step to improving speed is having an accurate measurement of performance,” said Sharan Narang, Systems Researcher at Baidu’s Silicon Valley AI Lab. “With the addition of the ability to measure inference, researchers will now have a more comprehensive benchmark for the performance of their AI hardware”.
Benchmarking inference is a challenging problem. Many applications that have been enabled by deep learning each have their own unique performance characteristics and requirements. In addition, there are several different deployment platforms. DeepBench attempts to solve this problem by benchmarking fundamental operations required for inference.
DeepBench originally gave researchers and developers insights into the best hardware solutions for their AI applications,” said Dr. Greg Diamos, Senior Researcher at Baidu Research’s Silicon Valley AI Lab. “Now, with the next generation of this platform, we will have even greater understanding to enable the design of AI hardware and applications that will continue to advance this technology.”
In addition to measuring inference performance, DeepBench also provides new kernels for training from several different deep learning models. DeepBench also sets new minimum precision requirements for training. Based on Baidu’s research, DeepBench establishes 16 bits for multiplication and 32 bits for addition for training operations.
DeepBench also provides results for training and inference across a variety of processors. We have collected results for server deployment as well as mobile deployment platforms such as the iPhone.
Sign up for the free insideAI News newsletter.