Swiftrl: Towards Efficient Reinforcement Learning On Real Processing-in-memory Systems

Abstract

Reinforcement Learning (RL) trains agents to learn optimal behavior by maximizing reward signals from experience datasets. However, RL training often faces memory limitations, leading to execution latencies and prolonged training times. To overcome this, SwiftRL explores Processing-In-Memory (PIM) architectures to accelerate RL workloads. We achieve near-linear performance scaling by implementing RL algorithms like Tabular Q-learning and SARSA on UPMEM PIM systems and optimizing for hardware. Our experiments on OpenAI GYM environments using UPMEM hardware demonstrate superior performance compared to CPU and GPU implementations.

Stats

• citations9
• S2 citations—
• github stars0
• HF likes0
• heat score7.50
• arxiv keygogineni2024swiftrl

Related papers

• A Dual-memory Architecture For Reinforcement Learning On Neuromorphic Platforms (2021)6.34
• Accelerated Methods For Deep Reinforcement Learning (2018)0.00
• Efficient Deep Reinforcement Learning With Predictive Processing Proximal Policy Optimization (2022)0.00
• Efficient Off-policy Reinforcement Learning Via Brain-inspired Computing (2022)8.35
• Human-inspired Framework To Accelerate Reinforcement Learning (2023)0.00
• Associative Memory Based Experience Replay For Deep Reinforcement Learning (2022)6.34
• Reservoir Computing For Fast, Simplified Reinforcement Learning On Memory Tasks (2024)0.00
• Reincarnating Reinforcement Learning: Reusing Prior Computation To Accelerate Progress (2022)0.00