Supernovae Observational Cosmology using GPU High Performance Computing

Supernovae Observational Cosmology using GPU High Performance Computing

COLISTETE JR., R; GIOSTRI, R.

Artigo:

We performed computing calculations of SNe Ia (type Ia supernova) in observational cosmology using CPU (Central Processing Unit) and GPU/CUDA (Graphics Processing Unit/Compute Unified Device Architecture) in 7 different programming methods : CUDA called from C/C++, Python with CUDA (Py-CUDA), Wolfram Mathematica with CUDA, C/C++, pure Python, Python with NumPy and Wolfram Mathematica calculated in CPU. With CUDA, we obtained speedup of approximately 3 to 7 hundred times with respect to CPU when performing calculations of the SNe Ia distance modulus (m0). So we confirmed that CUDA (Compute Unified Device Architecture) is an excellent choice for GPU High Performance Computing (HPC) architecture applied to observational cosmology calculations.

We performed computing calculations of SNe Ia (type Ia supernova) in observational cosmology using CPU (Central Processing Unit) and GPU/CUDA (Graphics Processing Unit/Compute Unified Device Architecture) in 7 different programming methods : CUDA called from C/C++, Python with CUDA (Py-CUDA), Wolfram Mathematica with CUDA, C/C++, pure Python, Python with NumPy and Wolfram Mathematica calculated in CPU. With CUDA, we obtained speedup of approximately 3 to 7 hundred times with respect to CPU when performing calculations of the SNe Ia distance modulus (m0). So we confirmed that CUDA (Compute Unified Device Architecture) is an excellent choice for GPU High Performance Computing (HPC) architecture applied to observational cosmology calculations.

Palavras-chave:

DOI: 10.5151/phypro-vii-efa-016

Referências bibliográficas

• [1] Wang, Y., et al. Dark Energy. Weinheim, Germany, Wiley, 2010.
• [2] Weinberg, D. H., et al. Physics Reports, v. 530, Issue 2, p. 87-256, 2013.
• [3] Wolfram, S. Mathematica: A System for Doing
• [4] Mathematics by Computer, 2nd edition. Redwood City, Addison-Wesley, 1991.
• [5] Nickolls, J., et al. ACM Queue, v. 6, n. 2, p. 40-53, March/April 2008.
• [6] Klckner, A., et al. Parallel Computing, v. 38, issue 3, p. 157-174, March 2012.
• [7] Filippenko, A. V. Annu. Rev. Astron. Astrophys, v. 35, p. 309-355, 1997.
• [8] Riess, A. G., et al. Astron. J., v. 116, p. 1009-1038, 1998.
• [9] Permutter, S., et al. Astrophys. J., v. 517, p. 565-586, 1999.
• [10] Drell, P. S., et al. Astrophys. J., v. 530, p. 593-617, 2000.

Como citar:

COLISTETE JR., R; GIOSTRI, R.; "Supernovae Observational Cosmology using GPU High Performance Computing", p-62-64. In: <strong>Anais do VII Encontro Científico de Física Aplicada [=Blucher Physics Proceedings, v.3 n.1]</strong>. São Paulo: Blucher, 2016.
ISSN 23582359, DOI 10.5151/phypro-vii-efa-016