Personal research page
  author: Ivan Šimeček  (xsimecek at fit dot cvut dot cz former xsimecek at fel dot cvut dot cz)
address: Department of Computer Systems, Faculty of of Information Technology, Czech Technical University, Thakurova 9, 160 00 Prague 6, Czech Republic
former:
Department of Computer Science, Faculty of Electrical Engineering, Czech Technical University, Technická 2, 166 27 Prague 6, Czech Republic


Keywords

High-performance, numerical linear algebra, cache memory, cache behavior models, code restructuring, architecture dependent optimization, dynamical loop reversal, loop blocking, loop unrolling, loop unrolling-and-jam.


Research papers



2013


2012
SYNASC 2012: Minimal Quadtree Format for Compression of Sparse Matrices Storage
HPCC 2012: Space-efficient sparse matrix storage formats for massively parallel systems
CSE 2012: A New Approach for Indexing Powder Diffraction Data Based on Dichotomy Method
SOCO 2012: A new approach for indexing powder diffraction data suitable for GPGPU execution
CANA 2012: Adaptive-Blocking Hierarchical Storage Format for Sparse Matrices
SNA 2012: Space-efficient sparse matrix storage formats with 8-bit indices
SNA 2012: Modular Arithmetic for Solving Linear Equations on the GPU
APLIMAT 2012: Heterogeneous cluster for acceleration of linear algebra computations

2011
ICPM 2011(invited talk): General purpose GPU computing and numerical linear algebra
ICPM 2011: Improvement of shortest path algorithms through graph partitioning
ICPM 2011: Experimental grid for numerical linear algebra

SNA 2011: Experimental grid for numerical linear algebra

2010
MEMICS 2010: Parallel solvers of Poisson's equation

SNA 2010: Comparison of the Sparse Matrix Storage Formats

SYNASC 2009: Sparse Matrix Computations Using the Quadtree Storage Format

2009
Acta Polytechnica: Memory Hierarchy Behavior Study during the Execution of Recursive Linear Algebra Library
Acta Polytechnica: The acceleration of the sparse matrix-vector multiplication by the region traversal

2008
PhD thesis: Architecture-Dependent Linear Code Optimizations

MEMICS'08: Sparse Matrix Computations with Quadtrees

SNA'08: Sparse Matrix Computations with Quadtrees
SNA'08: CPU or GPU: that is the question

ČVUT WORKSHOP'08: CPU or GPU: that is the question
ČVUT WORKSHOP'08: An Overview of Factorization of Large Integers Using the GMP Library
ČVUT WORKSHOP'08: Comparison of Different Implementations of BLAS and LAPACK Libraries with Application in Computational Fluid Dynamics
ČVUT WORKSHOP'08: Sparse Matrix Computations with Quadtrees

2007
PPAM'07: Sparse Matrix-Vector Multiplication - Final Solution?

SNA'07: A New Format for sparse Matrix-vector Multiplication
SNA'07: Recursive implementation of high performance numerical algebra library

ČVUT WORKSHOP'07: Cache Misses Prediction by Means of Data Mining Methods
ČVUT WORKSHOP'07: A New Format for the Sparse Matrix-vector Multiplication
ČVUT WORKSHOP'07: An Adaptive Solver for Fluid Dynamics on the GPU
ČVUT WORKSHOP'07: A New Approach for Accelerating the Sparse Matrix-vector Multiplication

2006
Acta Polytechnica: A simple cache emulator for evaluating cache behavior for SMP systems
Acta Polytechnica: Performance aspects of sparse matrix-vector multiplication

SYNASC'06: A New Approach for Accelerating the Sparse Matrix-vector Multiplication

SNA'06: Cache adaptive blocking format

ČVUT WORKSHOP'06: Analytical modelling of linear codes
ČVUT WORKSHOP'06: Cache Misses Analysis by Means of Data Mining Methods
ČVUT WORKSHOP'06: Cache Emulator for SMP Systems
ČVUT WORKSHOP'06: Optimization of SESOL Package
ČVUT WORKSHOP'06: Recursive Implementation of High Performance Numerical Algebra Library
ČVUT WORKSHOP'06: A New Diagonal Blocking Format and Model of Cache Behavior for Sparse Matrices

2005
PPAM'05: A New Diagonal Blocking Format and Model of Cache Behavior for Sparse Matrices

SNA'05: Optimalizace a hodnocení efektivity lineárních kódů

Documenta Geonica'05: Computer-related aspects of mathematical modelling at the Institute of Geonics

ČVUT WORKSHOP'05: Software Cache Analyzer
ČVUT WORKSHOP'05: Possibilities of GPU Computing
ČVUT WORKSHOP'05: Architecture Dependent Linear Code Optimizations
ČVUT WORKSHOP'05: Semi-sparse Cholesky Factorization

2004
HPSEC'04: Analytical Model for Analysis of Cache Behavior during Cholesky Factorization and Its Variants

NAA'04: Performance optimization and evaluation for linear codes

ČVUT POSTER'04: Improving of the performance of sparse matrix-vector multiplication on the modern architectures

2003
PPAM'03: Analytical Modeling of Optimized Sparse Linear Code

2001
Diploma thesis: Parallel matrix-matrix multiplication


Projects

2005

The Cache Analyzer

2006

Cache Emulator for SMP Systems

2009

High Performance Recursive Linear Algebra Library

CARB (new format for sparse matrix-vector multiplication)

2010

Quad-tree sparse matrix library

MCM2010 (software for powder diffraction data indexing)

SMP BLAS (Basic Linear Algebra Subroutines for SMP)

Factoring of large integers using GMP library

Generator of random large sparse matrices with given properties

2011

MasParIndex (Massively parallel software for for powder diffraction data indexing)

UFS file system defragmentation tool

The Linear Direct Sparse Solver on GPU for Bundle Adjustment Method

Heterogeneous cluster for acceleration of linear algebra computations
2012

Space-efficient sparse matrix storage formats for massively parallel systems

Minimal Quadtree Format for Compression of Sparse Matrices Storage

Distributed representation of Sparse Matrix in Quadtree Format

2013

MSH/GEO files generator


Curricula vitae


Degrees and titles. (all from CTU in Prague)
Engineer in Computer Science (1999)
PhD in Computer Science (2006)

Employment and visiting positions.
Sep 1999 -- Sep 2004 PhD student, Department of Computer Science FEE, CTU, Prague
Oct 2004 -- 2008 Assistant Professor, Department of Computer Science FEE CTU, Prague
2008 --- today Assistant Professor, Department of Computer Systems, FIT CTU, Prague




TOPlist