Skip to content

docs: design PDL NumPy and CUDA integration - #1303

Merged
fglock merged 1 commit into
masterfrom
docs/pdl-cuda-design
Sep 8, 2026
Merged

docs: design PDL NumPy and CUDA integration#1303
fglock merged 1 commit into
masterfrom
docs/pdl-cuda-design

Conversation

@fglock

@fglock fglock commented Sep 8, 2026

Copy link
Copy Markdown
Owner

Summary

Adds a design proposal for numerical performance and interoperability in PerlOnJava, covering PDL, direct CUDA, NumPy-compatible arrays and interchange, optional Python/NumPy bridging, and a measurable performance strategy that accounts for JVM warmup.

The design includes:

  • realistic compatibility boundaries and architecture options;
  • direct CUDA as an independent route when PDL integration is too costly;
  • NumPy-style semantics, .npy interchange, DLPack considerations, and an optional persistent CPython bridge;
  • resident device pipelines, fused plans, startup/break-even accounting, and NumPy/CuPy/PDL baselines;
  • milestones, acceptance gates, validation rules, progress tracking, and handoff requirements.

Validation

  • make check-links — 504 links checked, 0 errors.

Documentation-only change; no source or runtime tests are required.

Describe realistic compatibility routes, direct CUDA fallback, NumPy
interoperability, performance break-even measurements, milestones, and handoff.

Generated with [OpenAI Codex](https://openai.com/codex)

Co-Authored-By: OpenAI Codex <175728472+openai-codex[bot]@users.noreply.github.com>
@fglock
fglock merged commit 7ab1fea into master Sep 8, 2026
2 checks passed
@fglock
fglock deleted the docs/pdl-cuda-design branch September 8, 2026 13:15
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant