Another year of Progress in Machine-assisted Refactoring: what's new in Coccinelle

The most energy-efficient computing platform for large-scale numerical calculations nowadays is the GPU.
Porting old HPC codes for GPUs can be difficult.
As a result, efforts may lag behind what would be needed towards full or efficient use of GPUs.
Inefficient use of GPUs leads to inusually low performance but high energy expense; that may range into several kWhs per node -- and thus easily into hundreds of EUR of energy costs from a single simulation.

The Coccinelle project was established to ease maintenance of the Linux kernel and its drivers' code, in the C programming language. Nowadays it belongs to the toolkit of kernel drivers' maintainers. Without it, progress in the Linux Kernel development would be slower and its quality would be lower.

We are working to achieve an ambitious goal -- enabling Coccinelle use in large-scale code HPC-oriented refactorings (with emphasis on GPUs and C++), thus extending the lifetime of existing codebases, and easing their adaption to the new programming models required by the most recent hardware.

This poster tell the last year's progress of our collaboration, evidencing new features and new usages of Coccinelle: language constructs now supported, transformations now possible.
This poster may interest users of C, C++, but also of any other language interacting with these.