TAMG: Topology-Aware Multi-GPU Allocation via Deep Reinforcement Learning.
Hajer AyadiJimmy X. HuangAijun AnYiming ShaoHao ZhouHossein PourmodhejiPublished in: CASCON (2023)
Keyphrases
- reinforcement learning
- function approximation
- learning algorithm
- state space
- real time
- resource allocation
- allocation strategy
- markov decision processes
- graphics hardware
- dynamic allocation
- gpu accelerated
- policy search
- implicit surfaces
- deep learning
- parallel computation
- reinforcement learning algorithms
- parallel implementation
- evaluation function
- general purpose
- learning process
- data sets