We are an interdisciplinary group doing R+D on GPGPU Computing specifically tailored to Physical Simulations in CUDA.

We are currently doing joint work with a team of T. J. Watson IBM Research Center in characterizing a cognitive computing workload in POWER8+Pascal.

We have been granted by NVIDIA Professor Partner Funding Grants from September 2008 to December 2010.
We were CUDA Teaching Center, and CUDA Research Center from 2011 to the end of the program in 2016.

We collaborated with the LSC through the AGWG, and FAdeA.

Carlos Bederián	Sergio Cannas	Oscar Reula	Nicolás Wolovick
Mauricio Guignard	Marcelo Schild	Nahuel Seiler	Mauricio Mazuecos
Jorge Atala	Ezequiel Torti López	Lionel Hubmann

Former

Ezequiel Ferrero	Axel Dente	Germán Ceballos	Gastón Ingaramo	Matías Tealdi
J. P. De Francesco	Matías Bellone	Facundo Fabre	Leandro Perona	Dionisio Alonso

• Córdoba Province awarded us for our work with INVAP, 2017.
• IBM Faculty Award 2016 to work in the Fathom AI Workload in the Minsky server. Joint work with Augusto Vega at T. J. Watson IBM Research Center, 2016.

• Collaboration with FAdeA in acceleration of CFD using GPUs.
• We were part of the Argentinian Gravitational Wave Group (AGWG), in the LIGO Scientific Collaboration.
  The admission was decided in the Roma meeting, 2012.

• INVAP-FaMAF contract, part 2:
  A team of five people working on a GPU parallelization of a software for automatic object tracking in real-time, August 2014 to October 2014.
• Poder Judicial de la Pcia. de Córdoba-FaMAF contract:
  A team of two people doing password attack using high-end GPUs, March 2014.
• INVAP-FaMAF contract:
  A team of six people working on a GPU parallelization of a software for automatic object tracking in real-time, November 2013 to February 2014.

• Carlos Bederián, "Why GPUs?", Data Science Córdoba, October 7, 2015. [pdf] [video]
• Jorge Atala, Carlos Bederián, Nicolás Wolovick, "Embedded Supercomputing", Hackers&Founders, SemanaTIC, September 2, 2014. [pdf] [photo]
• Nicolás Wolovick, "Referentes Regionales en GPU Computing", La Voz del Interior, August 1, 2014.
• Jorge Atala, Nicolás Wolovick, "De la Play a la Facu: GPUs para la Ciencia", Conferencias Populares, Academia Nacional de Ciencias, July 23, 2014. [video, 2, 3, 4]
• Jorge Atala, Facundo Fabre, Oscar Reula, Nicolás Wolovick, "From Playstation to University: GPUs for Science", Ciencia para Armar, Universidad Nacional de Córdoba, 2014.

• Jorge Atala, Nicolás Wolovick, FaMAF-INVAP experience in GPU Computing, Workshop on GPU Computing for Space Applications, CETT, June 2017.
• Mauricio Guignard, Marcelo Schild, Carlos Bederián, Nicolás Wolovick, Intro to Parallel Computing on CPUs&GPUs, Facultad de Ingeniería, UCC, May 2017.
• Carlos Bederián, Optimización del Rendimiento de Aplicaciones GPU, ECAR14@CARLA2014, Universidad Técnica Federico Santa María, Valparaíso, Chile, Oct 2014 [Day1, Day2].
• Carlos Bederián, Architecture of Accelerators, 3EAGPGPU, Centro Atómico Bariloche, May 2014.
• Carlos Bederián, Advanced CUDA: streams and multi GPU. ICNPG, Centro Atómico Bariloche, November 2013.
• Carlos Bederián, Nicolás Wolovick, Introduction to Heterogeneous Computing. 20th Summer School on Informatics (RIO 2013), Universidad Nacional de Río Cuarto, February 18-23 2013. [materials]
• Axel Dente, Using Video Cards for Scientific Applications seminar. Physics Department, Universidad Nacional de Río Cuarto, October 2012.
• Carlos Bederián, Nicolás Wolovick, CUDA course in ECAR 2012, 31 July 2012, four days. [Materials]
• Axel Dente, Introduction to CUDA: Scientific Programming in Video Cards, Department of Physics, UNLP, June 2012, 12 hours.
• Carlos Bederián, Nicolás Wolovick, Parallel Computing, elective course at CS program, first semester 2012, FaMAF, UNC.
  This course has been awarded by NVIDIA CTC Program, and NSF/IEEE-TCPP Curriculum Initiative on Parallel and Distributed Computing (Spring 12).
• Dionisio Alonso, Intro to CUDA, Technical Discussions, Intel ASDC, Córdoba, June 2010. [pdf]
• "First Argentinean School on GPGPU Computing for Scientific Applications", May 23-27, 2011, FaMAF, UNC.
• Axel Dente, Ezequiel E. Ferrero, From Games to Science, Condensed Matter Theory Group Seminars, FaMAF, UNC, June 2010. [pdf]
• Peak Performance for an Application in CUDA at WHPC10, La Falda, May 4-5 2010. [Day1Slides], [Day2Slides], [code tarball]
• GPGPU CUDA MiniCourse at FaMAF:
  • Tuesday, March 9, 12 to 13, Dionisio Alonso: Introduction. Concurrent Programming Paradigms and examples (MPI, OpenMP, CUDA). One simple example in the three models: Simpson's Rule. Compilation and Runtime Toolkit. [pdf] [code tarball]
  • Thursday, March 11, 12 to 13, Matías Bellone: GPGPU Architecture from NVIDIA's GeForce 8 on. [pdf]
  • Tuesday, March 16, 11 to 12, Nicolás Wolovick: Concurrency, Contemption and Atomicity Problems in the Single Instruction Multiple Thread (SIMT) model. [PDF] [code tarball]
  • Thursday, March 18, 11 to 12, Matías Bellone: Application: Integrating post-Newtonian Equations. [pdf]
  • Tuesday, March 23, 11 to 12, Ezequiel Ferrero: Applications: Potts Model. [pdf]
  • Thursday, March 25, 11 to 12, Dionisio Alonso: Applications: Magneto Hydrodynamics. Final discussion. Possible collaborations with other groups. [pdf]
• CUDA minicourse, given by Juan Pablo de Francesco, from November 11, 2009.
  Schedule: 1-CUDA architecture, 2-Basic CUDA programming, 3-Real programming part one, 4-Real programming part two, 5-Optimizations.
  Slides.

• N. Wolovick, GPUs para ML, un enfoque práctico, Seminario Grupo de Análisis y Procesamiento de Grandes Redes Sociales y Semánticas, 2020. Video
• M.D. Mazuecos Pérez, N.G. Seiler, C.S. Bederián, N. Wolovick, A.J. Vega, Power Efficiency Analysis of a Deep Learning Workload on an IBM “Minsky” Platform, CARLA 2018. [slides] [in CCIS 979]
• M. Guignard, M. Schild, C.S. Bederián, N. Wolovick, A.J. Vega, Performance Characterization of State-Of-The-Art Deep Learning Workloads on a Minsky Platform, HICSS-51, 2018. (ISSIP-IBM-CBA Student Paper Award!)
• E.N. Millán, C.J. Ruestes, N. Wolovick, E.M. Bringa, Boosting Materials Science Simulations by High Performance Computing, ENIEF 2017.
• E.N. Millán, N. Wolovick, M.F. Piccoli, C. García Garino, E.M. Bringa, Performance analysis and comparison of cellular automata GPU implementations, Cluster Comput (2017). doi:10.1007/s10586-017-0850-3
• Carlos Bederián, Nicolás Wolovick, A Project-based HPC Course for Single-box Computers, EduHPC'16, SC16, 2016. (best paper award) [pdf] [slides]
• E.N. Millán, C.S. Bederián, M.F. Piccoli, C. García Garino, E.M. Bringa, Performance analysis of Cellular Automata HPC implementations, Computers & Electrical Engineering, Volume 48, November 2015, Pages 12-24, ISSN 0045-7906, DOI: 10.1016/j.compeleceng.2015.09.015.
• Carlos Bederián, Presentación del Grupo de GPGPU Computing, Quinta Jornada de Doctorandos de Computación, FaMAF, UNC. [PDF]
• Jorge Atala, Carlos Bederián, Andrés Bordese, Gastón Ingaramo, Facundo Gaich, Julia Medina, Maximiliano Rosetti, Jorge Sánchez, Matías Tealdi, Nicolás Wolovick, Real-time FullHD Tracking-Learning-Detection on a 2-SMX GPU, GTC 2015, Silicon Valley, 16-20 March 2015. [PDF]
• Fabio Bustos, Nicolás Wolovick, Una Experiencia en GPU Computing entre FaMAF e INVAP, Jornadas de la RedVITEC, UNC, 20 November 2014. [pdf]
• Emmanuel N. Millán, Paula Cecilia Martinez, Verónica Gil Costa, Maria Fabiana Píccoli, Marcela Printista, Carlos Bederián, Carlos García Garino and Eduardo M. Bringa, Parallel implementation of a Cellular Automata in a hybrid CPU/GPU environment, CACIC2013, Mar del Plata, Argentina, October 2013.
• Axel D. Dente, Carlos S. Bederián, Pablo R. Zangara, Horacio M. Pastawski, GPU accelerated Trotter-Suzuki solver for quantum spin dynamics, arXiv:1305.0036, 2013.
• Emmanuel N. Millán, Christian Ringl, Carlos S. Bederián, Fabiana Piccoli, Carlos García Garino, Herbert M. Urbassek, and Eduardo M. Bringa A GPU implementation for improved granular simulations with LAMMPS, HPCLatAm13, July 2013.
• Ezequiel E. Ferrero, Federico Romá, Sebastián Bustingorry, Pablo M. Gleiser, Dynamical heterogeneities as fingerprints of a backbone structure in Potts models, Phys. Rev. E 86, 031121 (2012). arXiv:1209.1743.
• Ezequiel E. Ferrero, poster q-state Potts model metastability study using optimized GPU-based Monte Carlo algorithms, Statmech2012, Mainz, Germany, September 2012. [PDF]
• Axel D. Dente, poster GPU enhanced spin dynamics with Trotter-Suzuki evolutions, HPCLatAm12, CABA, Argentina, August 2012. [poster] [slides]
  This work was awarded at the HPC Day@41JAIIO.
• Ezequiel E. Ferrero, Juan Pablo De Francesco, Nicolás Wolovick, Sergio A. Cannas, q-state Potts model metastability study using optimized GPU-based Monte Carlo algorithms, Computer Physics Communications Volume 183, Issue 8, August 2012, Pages 1578-1587. arXiv:1101.0876.
• Carlos S. Bederián, Axel D. Dente, Boosting quantum evolutions using Trotter-Suzuki algorithms on GPUs, HPCLatAm11, 40JAIIO, Córdoba, Argentina, August 2011 [pdf]
• Juan P. de Francesco, Ezequiel E. Ferrero, Nicolás Wolovick, Sergio A. Cannas, Simulation of the q-state Potts model using GPU, 8th TREFEMAC, Mar del Plata, Argentina, May 2010. [pdf]
• Frank Herrmann, John Silberholz, Matias Bellone, Gustavo Guerberoff, Manuel Tiglio, Integrating Post-Newtonian Equations on Graphics Processing Units, Class.Quant.Grav.27:032001, 2010, arXiv:0908.3889v2.
• Matías Bellone, Manuel Tiglio, Frank Herrmann, John Silberholz, Integration of Post-Newtonian Equations on GPUs, poster presentation at HPC2009, 38 JAIIO, Mar del Plata, Argentina, August 2009. [pdf]

• Juan Fraire, LCD, OpenCL-Accelerated Simplified General Perturbations 4 Algorithm, October 22, 2013. [slides]
• Jorge Atala, INVAP, Inside Tile64, September 6, 2013. [slides]
• Manuel Ujaldón, UMA, NVIDIA, Descubriendo Kepler, August 2, 2013. [pdf] [photo]
• Esteban Pellegrino, CAB, Monte Carlo Simulations for Neutronic Problem using GPU, August 25, 2011. [pdf]
• Ezequiel Ferrero, CAB, GPU Implementation of a Long-Range Phi4 Model with Pseudo-Spectral Techniques using CUFFT", August 10, 2011.
• Arnaldo Tavares, Tesla Business Development Manager - Latin America, Roadmap of the Tesla Line, April 7, 2011.
• Carlos Bederián, GPGPU-FaMAF, Ricci Flow Equation Simulation in S³ using GPU, November 24, 2010. [pdf]
• Carles Bona, Relativity and Gravitation Group, Universitat de les Illes Balears, Spain, Overview of the Symphony Code, November 17, 2010.

• One Titan Xp (3840 CUDA cores, 12GiB GDDR5X) GPU Grant Program, August 2018 [photo].
• Three Jetson TX2 development kits, aquired with the Córdoba Province award.
• One Tesla K40c (2880 CUDA cores, 12GiB GDDR5) EduHPC'16 best paper award prize.
• One GTX TITAN X (Maxwell) (3072 CUDA cores, 12GiB GDDR5) donated by NVIDIA as part of GEC 2016.
• Two GTX GTX 980 (2048 CUDA cores, 4 GiB GDDR5) donated by NVIDIA as part of GRC 2015.
• One Tesla K40c (2880 CUDA cores, 12GiB GDDR5) borrowed from Jorge Atala. It was a give away from GTC2015.
• Four Jetson TK1 developement kits, two acquired with group's fundings and two donated by NVIDIA. [photo]
• Oculus Rift, development kit donated by Hernán Amiune.
• Two GTX 650 (384 CUDA cores, 1GiB GDDR5) given as part of the INVAP-FaMAF contract.
• One Tesla K40c (2880 CUDA cores, 12GiB GDDR5) donated by NVIDIA as part of CRC Program. [photo]
• Two GTX Titan (2688 CUDA cores, 6GiB GDDR5) property of LANAIS and funded by "ANPCyT Bicentenario". [photo]
• One Tesla K20c (2496 CUDA cores, 5GiB GDDR5) donated by NVIDIA.
• FaMAF-IFEG-IATE Cluster "mendieta": 8×(2×Xeon E5 2680 + 2×Tesla m2090 + IB QDR + 2×GigE). [photo]
• Two Tesla C2075 and four GTX 480 donated by the Early Adopters program of NSF/IEEE-TCPP Curriculum Initiative.
• Intel DX58SO2 motherboard with Intel Core i7-980@3.3GHz, 24GB DDR3 1366MHz, power supply Powercooler PS-1200EM, property of LaNAIS Group, holding two Tesla C2070 boards (448 cores, 6GB GDDR5 ECC RAM each), donated by NVIDIA (CTC and CRC program).
• Intel DX58SO motherboard with Intel Core i7-950@3.07GHz, 16GB DDR3 1066MHz, power supply Powercooler PS-1200EM, property of LaNAIS Group, holding two Tesla C2070 boards (448 cores, 6GB GDDR5 ECC RAM each), donated by NVIDIA. [photo]

Retired

• Intel DX38 motherboard with Intel Core 2 Duo E8400, 2GB DDR3 1333MHz, power supply Power Cooler PS-850EM.
• Two NVIDIA GeForce GTX 280 (240 cores, 1GB GDDR3 RAM), donated by NVIDIA.
  Both were seeded to new GPGPU Groups in Argentina, one at IATE-UNC-Córdoba, the other at CAB-Bariloche.

• Pablo R. Zangara, Patricia R. Levstein, Horacio M. Pastawski, Role of energy uncertainties in ergodicity breaking induced by competing interactions and disorder. A dynamical assessment through the Loschmidt echo, Papers in Physics 7, 070012 (2015).
• Pablo R. Zangara, Denise Bendersky, Horacio M. Pastawski, Proliferation of effective interactions: Decoherence-induced equilibration in a closed many-body system, Phys. Rev. A 91, 042112 (2015).
• Pablo R. Zangara, Axel D. Dente, E. J. Torres-Herrera, Horacio M. Pastawski, Anibal Iucci, L. F. Santos, Time Fluctuations in Isolated Quantum Systems of Interacting Particles, Phys. Rev. E 88, 032913 (2013)
• Pablo R. Zangara, Axel D. Dente, Aníbal Iucci, Patricia R. Levstein, Horacio M. Pastawski Interaction-disorder competition in a spin system evaluated through the Loschmidt Echo, Phys. Rev. B 88, 195106 (2013)
• Renée Onnainty, Esteban M. Schenfeld, Mario A. Quevedo, Mariana A. Fernández, Marcela R. Longhi, and Gladys E. Granero, Characterization of the Hydrochlorothiazide: β-Cyclodextrin Inclusion Complex. Experimental and Theoretical Methods, J. Phys. Chem. B, 2013, 117 (1), pp 206–217.
• Gualdesi M. S., Briñón M. C., Quevedo M. A., Intestinal permeability of lamivudine (3TC) and two novel 3TC prodrugs. Experimental and theoretical analyses, Eur. J. Pharm. Sci. 2012 Dec 18;47(5):965-78.