|
AspectMatlab
| McIR and analysis and transformation engine |
McVM and McJIT |
Matlab to Fortran |
Domain-Specific Scientific Languages |
Analysis and Transformation EngineOverviewMATLAB is the most commonly used for engineering and scientific applications. Performance is at the heart of scientific and engineering computing. Tuning a critical loop could lead to drastic reduction in running time of a large scientific or engineering application. Performance improvement usually requires tuning of the source code and compilers to produce the faster code.Data dependence testing is the basic step in detecting loop level parallelism in numerical programs. Analysis and transformation engine computes dependence for the loop statements and applies various transformations to improve run time of programs.In MatLab the loop bounds are mostly determined at run time,so the profiler component of analysis and transformation engine generates data for different runs of the program.The data is then passed to heuristic engine to estimate appropriate ranges for loops in the progam. Dependence Analysis and transformations are applied on the estimated ranges.The focus of my thesis is to avoid expensive run time dependence analysis tests and do most of the computations at compile time.PublicationsAvailabilityMcLab is open sourced under the Apache 2.0 license. The source code for McLab is available here. |
