Understanding the Performance Behaviors of End-to-End Protein Design Pipelines on GPUs

TL;DR

This paper characterizes the computational performance of end-to-end protein design pipelines on GPUs, revealing low utilization and sensitivity to input parameters, and provides tools for future research.

Contribution

It offers a detailed system-level analysis of GPU-based protein design pipelines and releases open-source tools for further performance studies.

Findings

Low GPU utilization across the pipeline

High sensitivity to sequence length and sampling strategies

Open-source pipeline and profiling scripts released

Abstract

Recent computational advances enable protein design pipelines to run end-to-end on GPUs, yet their heterogeneous computational behaviors remain undercharacterized at the system level. We implement and profile a representative pipeline at both component and full-pipeline granularities across varying inputs and hyperparameters. Our characterization identifies generally low GPU utilization and high sensitivity to sequence length and sampling strategies. We outline future research directions based on these insights and release an open-source pipeline and profiling scripts to facilitate further studies.