Engineering samples of Arria 10 FPGAs are said to offer up to 40 GFLOPS/W; more than three times that offered by GPUs. This performance is achieved through using OpenCL to code the Arria 10 FPGA and its IEEE754 hard floating point DSP blocks.
“We are seeing a significant leap forward in CNN performance and power efficiency with Arria 10 engineering samples and the silicon's precision hard floating point in the DSP blocks is part of the reason we are seeing compelling results in our research," said Doug Burger, director, Client and Cloud Apps, Microsoft Research.
Altera FPGAs power Microsoft neural network development project
Microsoft is using Altera Arria 10 FPGAs to boost performance/Watt in data centre acceleration applications based on convolutional neural network (CNN) algorithms, used for image classification, image recognition, and natural language processing.
