Impacts of floating-point non-associativity on reproducibility for HPC and deep learning applications

TL;DR

This paper investigates how floating-point non-associativity affects reproducibility in HPC and deep learning, analyzing its impacts, and evaluating deterministic approaches including hardware solutions to improve reliability.

Contribution

It provides a comprehensive analysis of floating-point non-associativity effects in modern parallel computing and assesses deterministic hardware as a solution for reproducibility issues.

Findings

Non-associativity causes significant run-to-run variability in HPC and deep learning.

Replacing atomic operations with deterministic alternatives impacts performance and productivity.

Hardware-based determinism, like on Groq accelerators, enhances reproducibility and correctness.

Abstract

Run to run variability in parallel programs caused by floating-point non-associativity has been known to significantly affect reproducibility in iterative algorithms, due to accumulating errors. Non-reproducibility can critically affect the efficiency and effectiveness of correctness testing for stochastic programs. Recently, the sensitivity of deep learning training and inference pipelines to floating-point non-associativity has been found to sometimes be extreme. It can prevent certification for commercial applications, accurate assessment of robustness and sensitivity, and bug detection. New approaches in scientific computing applications have coupled deep learning models with high-performance computing, leading to an aggravation of debugging and testing challenges. Here we perform an investigation of the statistical properties of floating-point non-associativity within modern parallel programming models, and analyze performance and productivity impacts of replacing atomic operations with deterministic alternatives on GPUs. We examine the recently-added deterministic options in PyTorch within the context of GPU deployment for deep learning, uncovering and quantifying the impacts of input parameters triggering run to run variability and reporting on the reliability and completeness of the documentation. Finally, we evaluate the strategy of exploiting automatic determinism that could be provided by deterministic hardware, using the Groq accelerator for inference portions of the deep learning pipeline. We demonstrate the benefits that a hardware-based strategy can provide within reproducibility and correctness efforts.