非专家 说点皮毛 TPU 是google 自己customized的东东 优化算法(trade精度with speed/power), 用 在TENSOR FLOW APPLICATION上 是CPU 和GPU 之外的硬件辅助,对NVDA/INTC 影响不大 TechRadar: What is the chip exactly? Norm Jouppi: [The] Tensor Processing Unit (TPU) is our first custom accelerator ASIC [application-specific integrated circuit] for machine learning [ML], and it fits in the same footprint as a hard drive. It is customized to give high performance and power efficiency when running TensorFlow. What makes the TPU different from your standard processor specifically? TPUs are customized for machine learning applications using TensorFlow. Note that we continue to use CPUs [central processing units] and GPUs [graphics processing units] for ML.
【在 s*********r 的大作中提到】 : 非专家 说点皮毛 : TPU 是google 自己customized的东东 优化算法(trade精度with speed/power), 用 : 在TENSOR FLOW APPLICATION上 : 是CPU 和GPU 之外的硬件辅助,对NVDA/INTC 影响不大 : TechRadar: What is the chip exactly? : Norm Jouppi: [The] Tensor Processing Unit (TPU) is our first custom : accelerator ASIC [application-specific integrated circuit] for machine : learning [ML], and it fits in the same footprint as a hard drive. It is : customized to give high performance and power efficiency when running : TensorFlow.