We are excited to announce the release of Codee 1.3, which further improves the first static code analyzer specifically designed to boost the performance of C/C++ applications.
The major feature of this release is the cost model used to guide the performance optimization of a loop. This new analytical model estimates the cost of executing a loop and decides whether memory optimizations and vectorization bring actual performance benefits.
The new vectorization cost model implemented in Codee 1.3 proposes vectorization only for eligible loops using a set of heuristics based on memory access patterns, conditionals and loop trip count. For non-vectorizable loops, the reasons preventing vectorization are reported through the remarks RMK010, RMK011, RMK012, RMK013 and RMK014.
The new memory cost model implemented in Codee 1.3 detects loops that are eligible for the loop interchange optimization, which is reported through the recommendation PWR039.
Finally, Codee 1.3 augments the existing entry-level performance optimization report by classifying the loops according to how difficult it is to optimize them for performance. The new option --evaluation --level 2 includes low, medium and high difficulty.
We will be happy to meet with you to show you a demo or reply to your questions!
Building performance into the code from day one with Codee
Leave a Reply