An FPGA-Based Hardware Accelerator for CNNs Using On-Chip Memories Only: Design and Benchmarking with Intel Movidius Neural Compute Stick

<table class="table-group" id="tab4"><tr><td><table class="table"><tr><td class="thead-hr" colspan="2"><hr/></td></tr><tr class="thead"><td class="align_left">Layer</td><td class="align_center"><i>N</i><sub>clk</sub></td></tr><tr><td class="thead-hr" colspan="2"><hr/></td></tr><tr><td class="align_left">Input memory</td><td class="align_center">819</td></tr><tr><td class="align_left">Time_0</td><td class="align_center">767</td></tr><tr><td class="align_left">Freq_0</td><td class="align_center">5192</td></tr><tr><td class="align_left">Time_1</td><td class="align_center">4840</td></tr><tr><td class="align_left">Freq_1</td><td class="align_center">7920</td></tr><tr><td class="align_left">Time_2</td><td class="align_center">6480</td></tr><tr><td class="align_left">Freq_2</td><td class="align_center">60480</td></tr><tr><td class="align_left">Final_conv</td><td class="align_center">3780</td></tr><tr><td class="align_left">Total</td><td class="align_center">90278</td></tr><tr class="table-tr"><td colspan="2"><hr class="tbody-hr"/></td></tr></table></td></tr></table>

<div><i>N</i><sub>clk</sub> values for the various layers.</div>

International Journal of Reconfigurable Computing

tab4

Table 4

Table 4: An FPGA-Based Hardware Accelerator for CNNs Using On-Chip Memories Only: Design and Benchmarking with Intel Movidius Neural Compute Stick 