Generating optimal CUDA sparse matrix-vector product implementations for evolving GPU hardware

Concurrency and Computation: Practice and Experience

Volumn 24, Issue 1, 2012, Pages 3-13

  El Zein, Ahmed H ^a   Rendell, Alistair P ^a  

^a AUSTRALIAN NATIONAL UNIVERSITY   (Australia)

Author keywords

CUDA; Fermi; GPU; matrix vector; NVIDIA; S2050; sparse

Indexed keywords

COMPUTER GRAPHICS; COMPUTER HARDWARE; MATRIX ALGEBRA; PROGRAM PROCESSORS;

CUDA; FERMI; MATRIX-VECTOR; NVIDIA; S2050; SPARSE;

GRAPHICS PROCESSING UNIT;

EID: 84855223315     PISSN: 15320626     EISSN: 15320634     Source Type: Journal    
DOI: 10.1002/cpe.1732     Document Type: Article

References (16)

1. NVIDIA. NVIDIA CUDA Programming Guide (2nd edn). NVIDIA Corporation, July 2008.
2. Im EJ, Yelick KA, Vuduc RW,. Sparsity: Optimization framework for sparse matrix kernels. IJHPCA 2004; 18 (1): 135-158.
3. Vuduc R, Demmel JW, Yelick KA,. OSKI: A library of automatically tuned sparse matrix kernels. Proceedings of SciDAC'05. Journal of Physics: Conference Series, Institute of Physics Publishing: San Francisco, CA, U.S.A., 2005.
4. Buttari A, Eijkhout V, Langou J, Filippone S,. Performance optimization and modeling of blocked sparse kernels. IJHPCA 2007; 21 (4): 467-484. (Pubitemid 350011340)
5. Zein AHE, Rendell AP,. From sparse matrix to optimal GPU CUDA sparse matrix vector product implementation. CCGRID, Washington, DC, U.S.A. IEEE, 2010; 808-813.
6. Zein AE, McCreath E, Rendell AP, Smola AJ,. Performance evaluation of the nvidia geforce 8800 gtx gpu for machine learning. ICCS (1) (Lecture Notes in Computer Science, vol. 5101), Bubak M, van Albada GD, Dongarra J, Sloot PMA, (eds.). Springer: Berlin, 2008; 466-475.
7. Davis TA,. University of florida sparse matrix collection. NA Digest 1994; 92.
8. Tesla S2050 GPU computing system. Available at: http://www.nvidia.com/ object/product-tesla-S2050-us.html [August 2010 ].
9. NVIDIA. NVIDIA's next generation CUDA compute architecture: Fermi. White Paper, June 2010. Available at: http://www.nvidia.com/content/PDF/fermi-white- papers/NVIDIA-Fermi-Compute-Architecture-Whitepaper.pdf [August 2010].
10. Bell N, Garland M,. Implementing sparse matrix-vector multiplication on throughput-oriented processors. SC. ACM: New York, 2009.
11. Baskaran MM, Bordawekar R,. Optimizing sparse matrix-vector multiplication on gpus. IBM Research Report 2009; RC24704.
12. Williams S, Oliker L, Vuduc RW, Shalf J, Yelick KA, Demmel J,. Optimization of sparse matrix-vector multiplication on emerging multicore platforms. SC, Verastegui B, (ed.). ACM Press: New York, 2007; 38.
13. CUDPP: CUDA Data Parallel Primitives library. Available at: http://www.gpgpu.org/developer/cudpp/[August 2010 ].
14. Sengupta S, Harris M, Zhang Y, Owens JD,. Scan primitives for gpu computing. Graphics Hardware, Segal M, Aila T, (eds.). Eurographics Association: Aire-la-Ville, Switzerland, 2007; 97-106.
15. Armstrong W, Rendell AP,. Reinforcement learning for automated performance tuning: Initial evaluation for sparse matrix format selection. CLUSTER. IEEE: New York, 2008; 411-420.
16. Lee VW, Kim C, Chhugani J, Deisher M, Kim D, Nguyen AD, Satish N, Smelyanskiy M, Chennupaty S, Hammarlund P,. et al. Debunking the 100x gpu vs. cpu myth: An evaluation of throughput computing on cpu and gpu. ISCA, Seznec A, Weiser UC, Ronen R, (eds.). ACM: New York, 2010; 451-460.