“FIXAR: A Fixed-Point Deep Reinforcement Learning Platform with Quantization-Aware Training and Adaptive Parallelism,” ACM/IEEE Design Automation Conference (DAC), 2021