Tue 19 Jun 2018 11:35 - 12:10 at Grand Ballroom C - Exploiting Dynamic Information

In recent years, MATLAB's just-in-time (JIT) interpreter has improved the execution time of for-loops to the extent that loops can outperform equivalent array operations in some scenarios. This has caused systematic translation of loops to array operations, a prevalent approach for performance improvement in MATLAB, to sometimes yield a performance loss. Therefore, we propose a selective strategy to loop translation with selection criteria guided by loop profiling data. As a result, only loops with a high-performance speedup potential are selected for translation to array operations. The results of our experiments confirm the efficiency of our approach and illustrate the cases where systematic translation leads to a performance degradation.

Tue 19 Jun

Displayed time zone: Eastern Time (US & Canada) change

11:00 - 12:10
Exploiting Dynamic InformationARRAY at Grand Ballroom C
11:00
35m
Talk
Petalisp: Run Time Code Generation for Operations on Strided Arrays
ARRAY
Marco Heisig , Harald Köstler Friedrich-Alexander University Erlangen-Nürnberg, Germany
File Attached
11:35
35m
Talk
Profile-Based Vectorization for MATLAB
ARRAY
Patryk Kiepas MINES ParisTech, France, Jaroslaw Kozlak AGH University of Science and Technology, Poland, Claude Tadonki MINES ParisTech, France, Corinne Ancourt MINES ParisTech, France