https://www.pistack.xyz/tags/scientific-computing/ Recent content in Scientific-Computing on Pi Stack Hugo en-us Thu, 04 Jun 2026 00:00:00 +0000 https://www.pistack.xyz/posts/2026-06-04-genomics-workflow-pipelines-nextflow-snakemake-cromwell-guide/ Thu, 04 Jun 2026 00:00:00 +0000 https://www.pistack.xyz/posts/2026-06-04-genomics-workflow-pipelines-nextflow-snakemake-cromwell-guide/ <h2 id="introduction">Introduction</h2> <p>Modern genomics research generates terabytes of data per experiment. A single whole-genome sequencing run can produce hundreds of gigabytes of raw reads that must be aligned, called, filtered, annotated, and visualized — all before a researcher can draw biological conclusions. Managing these multi-step computational workflows at scale requires specialized pipeline orchestration tools purpose-built for scientific computing.</p> https://www.pistack.xyz/posts/2026-06-01-self-hosted-hpc-container-runtimes-apptainer-charliecloud-podman-hpc-guide/ Mon, 01 Jun 2026 00:00:00 +0000 https://www.pistack.xyz/posts/2026-06-01-self-hosted-hpc-container-runtimes-apptainer-charliecloud-podman-hpc-guide/ <h2 id="why-self-host-hpc-container-runtimes">Why Self-Host HPC Container Runtimes?</h2> <p>High-Performance Computing (HPC) centers face a unique containerization challenge: traditional container runtimes like Docker assume root privileges, daemon-based architectures, and layered filesystem overlays that conflict with shared, multi-tenant HPC clusters. HPC-specific container runtimes solve these problems by supporting unprivileged execution, parallel filesystem compatibility, and MPI integration out of the box.</p> https://www.pistack.xyz/posts/2026-06-01-self-hosted-hpc-mpi-implementations-openmpi-mpich-mvapich-guide/ Mon, 01 Jun 2026 00:00:00 +0000 https://www.pistack.xyz/posts/2026-06-01-self-hosted-hpc-mpi-implementations-openmpi-mpich-mvapich-guide/ <h2 id="why-self-host-an-mpi-implementation">Why Self-Host an MPI Implementation?</h2> <p>Message Passing Interface (MPI) is the foundational communication layer for virtually all parallel scientific computing. Every weather simulation, molecular dynamics calculation, and computational fluid dynamics model depends on efficient inter-process communication across potentially thousands of compute nodes. The choice of MPI implementation directly impacts application performance — different implementations optimize for different hardware (InfiniBand, Omni-Path, Slingshot), different communication patterns (point-to-point vs collective), and different levels of thread safety.</p>