30 Years of HPC: Many Hardware Advances, Little Adoption of New Languages

A Hacker News discussion asked why HPC has advanced faster in hardware than in programming languages.

Commenters said many HPC workloads are limited by memory bandwidth and data movement, not just parallelism.

That makes established tools like C++ a practical fit for performance-critical code, reducing demand for newer language designs.

Others noted that real cluster work still often uses Python, R, Perl, and awk, with bioinformatics as a notable exception.