The paper "YaCoS: a Complete Infrastructure to the Design and Exploration of Code Optimization Sequences" got second place in the best-paper award of SBLP 2020!
This paper, jointly authored by André Zanella, Anderson Faustino, and Fernando Pereira, describes an ensemble of techniques that developers can use to autotune compilers. Autotuning means training the compiler to treat each program differently. So, the compiler can apply on each program sequences of optimizations that are good for that program. The space of possible optimization sequences is huge! So, the community needs the support of an infrastructure like YaCoS, which provides different ways to compare programs, lots of benchmarks, and many algorithms to try different optimization sequences.