File Compression for HPC
One of the most underutilized capabilities in Linux and HPC is compressing unused data files.
Processor Affinity for OpenMP and MPI
Processor affinity with serial, OpenMP, and MPI applications.
Processor and Memory Affinity Tools
Get better performance from your nodes by binding processes and associating memory to specific cores.
Darshan I/O Analysis for Deep Learning Frameworks
I/O Characterization of TensorFlow with Darshan
Prolog and Epilog Scripts
HPC systems can benefit from administrator-defined prolog and epilog scripts.
Run One Program at any Scale with Legate
Run Python NumPy code on distributed heterogeneous systems without changing a single line of code.
When I/O Workloads Don’t Perform
Isolating and diagnosing the root causes of your performance troubles.
Preload Trick
By using the LD_PRELOAD environment variable, you can improve performance without making changes to applications.
New Monitoring Tools
If you like ASCII-based monitoring tools, take a look at three new tools – Zenith, Bpytop, and Bottom.
Desktop Supercomputers: Past, Present, and Future
Desktop supercomputers give individual users control over compute power to run applications locally at will.