OpenEvolve AI coding agent built a better algorithm

"Computer scientists at UC Berkeley say that AI models show promise as a way to discover and optimize algorithms. In a preprint paper titled "Barbarians at the Gate: How AI is Upending Systems Research," 17 UC Berkeley researchers describe how they employed OpenEvolve, an open source implementation of Google DeepMind's AlphaEvolve, to improve a load balancing algorithm so that it significantly outperforms prior human designs."

"Specifically, the authors claim to have used OpenEvolve to achieve a 5x speedup for an Expert Parallelism Load Balancer (EPLB) algorithm, which is used in large language models to route tokens to specialized expert modules - an efficiency mechanism that reduces the number of processed parameters. The authors say that AI-Driven Research for Systems (ADRS), through which an AI model iteratively generates, evaluates, and refines solutions, promises to transform systems research."

"Google in May talked up AlphaEvolve, an "evolutionary coding agent" that improved the efficiency of Google's data center orchestration, optimized matrix multiplication operations in its Tensor Processing Unit hardware, and optimized its FlashAttention kernel implementation in Transformer-based AI models. As if to further underscore the potential of machine learning as an algorithmic discovery mechanism, a paper published this week in Nature from Google DeepMind researchers describes "an autonomous method for discovering [reinforcement learning] rules solely through the experience of many generations of agents interacting with various environments.""

UC Berkeley computer scientists used OpenEvolve, an open-source implementation of AlphaEvolve, to improve a load-balancing algorithm and achieve substantial performance gains. OpenEvolve produced a reported 5x speedup for an Expert Parallelism Load Balancer (EPLB) that routes tokens to specialized expert modules in large language models. AI-Driven Research for Systems (ADRS) frames an iterative cycle where models generate, evaluate, and refine solutions, promising to change systems research workflows. AlphaEvolve demonstrated optimizations in data center orchestration, TPU matrix multiplication, and FlashAttention kernels. DeepMind researchers reported an autonomous method that discovers reinforcement-learning rules through many generations of agent interaction.

Read at Theregister