Pinpointing the Memory Behaviors of DNN Training
This work addresses memory optimization for DNN training on accelerators, but it is incremental as it provides observations for future improvements.
The authors tackled the problem of memory-hungry deep neural network training by characterizing GPU device memory behaviors, finding that memory access patterns are stable and iterative.
The training of deep neural networks (DNNs) is usually memory-hungry due to the limited device memory capacity of DNN accelerators. Characterizing the memory behaviors of DNN training is critical to optimize the device memory pressures. In this work, we pinpoint the memory behaviors of each device memory block of GPU during training by instrumenting the memory allocators of the runtime system. Our results show that the memory access patterns of device memory blocks are stable and follow an iterative fashion. These observations are useful for the future optimization of memory-efficient training from the perspective of raw memory access patterns.