Performance evaluation of LU matrix decomposition using the SYCL standard

The object of this study is the performance SYCL standard tools when solving LU matrix decomposition problem. a fairly new technology for parallel computing in heterogeneous systems, so topic evaluating on specific tasks field relevant. In study, algorithm parallelized square was implemented by means and C++, an experiment conducted to test implementation system with several types processors. During testing, program received matrices various dimensions as input, output execution time schedule selected processor. obtained results, presented form tabular graphic data, show advantage over ordinary C++ more than 2 times using graphics It experimentally shown that almost not inferior speed executed central Such results are caused both high possibility parallelizing itself, great work developers its optimization. indicate speeding up solution similar algorithms systems processors optimized data parallelism. can be used justifying choice problems or parallelization scheme.