Auviz image-processing and computer vision libraries deliver GPU-like performance for FPGA

Auviz libraries help SDAccel deliver GPU-like performance with significantly less power by harnessing FPGA hardware acceleration in a way that’s friendly to the software programming world.

Auviz image-processing and computer vision libraries deliver GPU-like performance for FPGAs with Xilinx SDAccel

by  sleibso ‎11-19-2014 05:39 PM - edited ‎11-19-2014 05:40 PM(7,088 Views)

Image-processing, computer vision, and linear algebra libraries from Auviz Systems are key elements in the Xilinx SDAccel Development Environment rolled out this week at SC14 (Supercomputing 14) in New Orleans. (See “CPU/GPU-like software development environment for OpenCL, C, C++ delivers FPGA-based app acceleration with 25x better performance/W.”) Nagesh Gupta, Founder and CEO of Auviz, gave me the following 2-minute tour of his company’s product:

And here are the functions in the AuvizLA linear-algebra (BLAS) library:

These libraries are used in their OpenCL form for SDAccel but they are also available as HLS modules and as modules for use with IP Integrator within the Xilinx Vivado Design Suite.

NB: The information provided in this press release is for general informational purposes and published on PRnob only.