Milestones

What the three thrusts deliver, and when.

Year 1

2025–2026
  • Workflow study and format coverage Thrust 1 · survey the analysis workflows the cyberinfrastructure has to serve, and the data formats they arrive in

  • Compression-quality targets Thrust 1 · pin the quality metrics each workflow needs, so "good enough" is a number rather than a judgement

  • Baseline ports Thrust 3 · the existing GPU compressors building and running on NVIDIA, as the reference the other platforms are measured against

Year 2

2026–2027
  • Language bindings Thrust 2 · high-level interfaces, so using a GPU compressor does not mean writing GPU code

  • Command-line interface Thrust 2 · an off-the-shelf tool for the common cases

  • AMD and Intel back ends Thrust 3 · the portability layer carrying the compressors off NVIDIA-only hardware

Year 3

2027–2028
  • Visualization-integrated interface Thrust 2 · the user interface that puts compression next to the analysis it serves

  • Cross-platform evaluation Thrust 3 · one evaluation across all three vendors, on real application data

  • Release and hand-off Thrusts 1–3 · documentation, packaging, and the training material the education plan calls for

Years follow the award period: the collaborating institutions started August 1, 2025 and Oakland October 1, 2025, all running three years. This page lays the three thrusts out over that period; it is a plan, not a record of completed work.