We are thrilled to announce the much-anticipated launch of Codee 2024.1, which further improves the capabilities of our solution for automated code testing specialized in performance that enables the detection of performance issues early in the software development life cycle.
- The Screening with Ranking Report, which helps prioritize the implementation of rules most suited for the environment where the code will run.
This is one of the key features of Codee 2024.1. These rules are ordered top-down, so that users are suggested to start working on the rules at the top of the ranking to maximize the impact on performance in their own codes.
- The Ranking is elaborated using “risk assessment criteria” similar to those employed in the field of coding standards for security, which favors the adoption by the developers that are already familiar with secure coding.
The SEI CERT C Coding Standard is well established in the industry, and Codee is leveraging its risk assessment criteria as follows:
- Likelihood: This represents the probability that the fix proposed by the checker will achieve a performance improvement (e.g. reduce cache misses).
- Remediation cost: An estimation of the effort required for implementing the fix proposed by the checker (e.g. switching compiler flags, changing data structures).
- Severity: This indicates whether the checker addresses a fundamental aspect of performance in modern hardware (e.g. memory efficiency) or a platform-specific aspect (e.g. data transfers in accelerators).
Similarly to SEI CERT C, these three checker attributes dictate the priority that each checker will have.
- Improved Fortran language support, which shows our commitment to helping developers with modernization of Fortran legacy code, as well as improving the performance of Fortran programs.
Our team has worked tirelessly to enhance Codee’s compatibility and stability for Fortran program constructs. Current focus is also on the automation of new checkers to pinpoint the legacy Fortran constructs used in the code as well as opportunities for optimization on CPU and GPU. Many more enhancements to come in future releases of Codee.
- The GitHub repository ‘Open Catalog of Best Practices for Performance’!
We consider this a major milestone that will strengthen our collaboration developers interested in best practices for performance in C, Fortran and C++. The github repository leverages decades of experience in optimization of C, Fortran and C++ source code and enables contributions by sending pull requests to amend, improve, or include new content. Whether you have insights, suggestions, or questions, our repository is open for issues where you can actively participate in shaping the future of code performance.
We look forward to seeing the positive impact Codee will have on your development workflow, and we welcome your feedback and contributions to make Codee even better.