10-05-2018 04:31 AM
After succesfully running the perpetual demo using GoogleNet (following the example and runnning "./run.sh aws batch_classify large 8" i achieve almost 670 Im/sec (with 256 batch size on googlenet 8b). I measured this with "FPGA xdnn execute" time showing on terminal because the web GUI doesn't seem to work. Is there a problem with the web GUI? Also I wonder if the results are optimal as there is a screenshot of the web GUI in the ML-Suite repo showing higher speeds.
Thanks in advance.