论文信息 - Neutron sensitivity and hardening strategies for Fast Fourier Transform on GPUs

Neutron sensitivity and hardening strategies for Fast Fourier Transform on GPUs

In this paper we analyze the neutron sensitivity of GPU devices when executing a Fast Fourier Transform algorithm. The provided experimental results demonstrate that in the majority of cases the output is affected by multiple errors, caused by thread and data dependencies. ECC is experimentally proved not to be sufficient to provide high reliability. Experimental data and analytical studies are employed to design specific software-based hardening strategies, which are validated through fault-injection.

[1] Thomas G. Stockham,et al. High-speed convolution and correlation , 1966, AFIPS '66 (Spring).

[2] Jacob A. Abraham,et al. Fault-Tolerant FFT Networks , 1988, IEEE Trans. Computers.

[3] L. Carro,et al. An Efficient and Experimentally Tuned Software-Based Hardening Strategy for Matrix Multiplication on GPUs , 2013, IEEE Transactions on Nuclear Science.

[4] David H. Bailey,et al. The Nas Parallel Benchmarks , 1991, Int. J. High Perform. Comput. Appl..

[5] L. Carro,et al. Neutron-Induced Soft Errors in Graphic Processing Units , 2012, 2012 IEEE Radiation Effects Data Workshop.

[6] J. Krüger,et al. Linear algebra operators for GPU implementation of numerical algorithms , 2003, ACM Trans. Graph..

[7] Vijay S. Pande,et al. Hard Data on Soft Errors: A Large-Scale Assessment of Real-World Error Rates in GPGPU , 2009, 2010 10th IEEE/ACM International Conference on Cluster, Cloud and Grid Computing.

[8] Kevin Skadron,et al. A hardware redundancy and recovery mechanism for reliable scientific computation on graphics processors , 2007, GH '07.

[9] Mark D. Lerner. Algorithm Based Fault Tolerance in Massively Parallel Systems , 1988 .

[10] S. Pontarelli,et al. A New Hardware/Software Platform and a New 1/E Neutron Source for Soft Error Studies: Testing FPGAs at the ISIS Facility , 2007, IEEE Transactions on Nuclear Science.