Approximate Computing Multiprocessing, multithreading and vectorization GPU architecture and optimizations Floating-point, fixed-point and integer formats Roofline model and arithmetic intensity Common optimizations Cache Blocking Four convolution strategies: Direct, GEMM, FFT and Winograd Model-size aware and system-aware pruning of CNNs Deep Learning on Systolic Array and TPU Deep Learning on FPGA Deep Learning on Mobile GPU

AI