As a child, I was obsessed with video game maps. I would pore over them in old issues of Nintendo Power magazine, and even try to recreate the levels of my favorite games block by block on pieces of graph paper. I wasn’t alone. In the early days of the Super Mario franchise, that’s exactly how the folks at Nintendo were designing them.
 Quantized Neural Networks for FPGA InferenceSponsored by Intel Implementing a low precision network in FPGA hardware for efficient inferencing provides numerous advantages when it comes to meeting demanding specifications. The increased flexibility allows optimization of throughput, overall power consumption, resource usage, device size, TOPs/watt, and deterministic latency. These are important benefits where scaling and efficiency are inherent requirements of the application. |
