In this study, we characterize GPU failures in Delta, the current large-scale AI system with over 600 petaflops of peak compute throughput. The system comprises GPU and non-GPU nodes with modern AI accelerators, such as NVIDIA A40, A100, and H100 GPUs. The study uses two and a half years of data on GPU errors. We evaluate the resilience of GPU hardware components to determine the vulnerability of different GPU components to failure and their impact on the GPU and node availability. We measure the key propagation paths in GPU hardware, GPU interconnect (NVLink), and GPU memory. Finally, we evaluate the impact of the observed GPU errors on user jobs. Our key findings are: (i) Contrary to common beliefs, GPU memory is over 30x more reliable than GPU hardware in terms of MTBE (mean time between errors). (ii) The newly introduced GSP (GPU System Processor) is the most vulnerable GPU hardware component. (iii) NVLink errors did not always lead to user job failure, and we attribute it to the underlying error detection and retry mechanisms employed. (iv) We show multiple examples of hardware errors originating from one of the key GPU hardware components, leading to application failure. (v) We project the impact of GPU node availability on larger scales with emulation and find that significant overprovisioning between 5-20% would be necessary to handle GPU failures. If GPU availability were improved to 99.9%, the overprovisioning would be reduced by 4x.
View on arXiv@article{cui2025_2503.11901,
title={ Characterizing GPU Resilience and Impact on AI/HPC Systems },
author={ Shengkun Cui and Archit Patke and Ziheng Chen and Aditya Ranjan and Hung Nguyen and Phuong Cao and Saurabh Jha and Brett Bode and Gregory Bauer and Chandra Narayanaswami and Daby Sow and Catello Di Martino and Zbigniew T. Kalbarczyk and Ravishankar K. Iyer },
journal={arXiv preprint arXiv:2503.11901},
year={ 2025 }
}