GPU-accelerated Computing in Bioconductor

Purpose

Support GPU-accelerated Bioconductor packages through continuous integration, user-friendly packaging of system-level dependencies, and foundational packages for Bioconductor GPU programming.

Aims

1. Extension of Bioconductor’s existing continuous integration (CI) infrastructure.
2. Implementation of flexible and user-friendly packaging of system-level dependencies.
3. Development of foundational packages for GPU programming

Accomplishments

GPU hardware for builds

• New Nvidia builder testing Bioconductor packages using a container
• Exposed GPU Nvidia compiler and system management information

GPU Software build reports

• Release and devel reports

• Support for both GPU required and capable Bioconductor packages
• Flexibility for maintainers to opt their packages into the GPU Software builds
• New biocViews GPU term to help users find GPU-capable Bioconductor packages

Current work

Aim 2: Flexible, user-friendly packaging of system-level dependencies

• ronda Repackage any R package for distribution via Conda
• Rcollectl-GPU Developing a GPU-monitoring R package

Future work

Aim 3: Foundational packages for GPU programming

• Extend Bioconductor data structures to integrate GPU processing.
• Implement a CUDA backend for common operations in single-cell omics applicable to both dense and sparse matrices.
• Best practices for Bioconductor package maintainers and users.