rigsolve. GPU stack compatibility, resolved.

rigsolve

CI Codecov PyPI Python License: Apache-2.0

Resolve torch, CUDA, and native extension compatibility from sourced evidence.

$ rigsolve check
[FAIL] torch was built for CUDA 12.4, but flash-attn expects CUDA 11
  fix: re-resolve torch and the extension on one CUDA line

rigsolve detects the local GPU stack without importing torch. It resolves driver, CUDA, Python, platform, GPU architecture, torch build, C++ ABI, package coupling, and known broken combinations together.

Install

python -m pip install rigsolve

Requires Python 3.10 or newer. Current target data focuses on Linux x86_64 and NVIDIA CUDA stacks.

Quick start

Inspect the machine:

rigsolve detect
rigsolve doctor

Create a plan:

rigsolve solve \
  --want 'flash-attn==2.8.3' \
  --target 'RTX 4090,driver=580.65,python=3.12,linux'

Typical output:

# Matrix 2026.08.15 (1e066bd53f01); evidence: metadata-backed.
python -m pip install --index-url https://download.pytorch.org/whl/cu126 torch==2.9.0
python -m pip install 'https://github.com/Dao-AILab/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3%2Bcu12torch2.9cxx11abiTRUE-cp312-cp312-linux_x86_64.whl#sha256=4e2f9e39313266b1544b68138b15b91ee6221eccf14f7902b7c6620351340810'

Install and verify on the detected machine:

rigsolve solve --want torch --execute

--execute runs isolated import checks and available GPU probes after installation. Hypothetical --target and --python plans cannot be executed.

Diagnose an installed environment:

rigsolve check
rigsolve check --fix

Explain a request:

rigsolve why 'flash-attn==2.8.3' \
  --target 'RTX 4090,driver=580.65,python=3.12,linux'

Commands

Command Purpose
rigsolve detect Profile the GPU, driver, toolkit, Python, platform, and installed packages
rigsolve solve Resolve a compatible stack and render an install plan
rigsolve check Report known problems in the installed environment
rigsolve why Explain a solution or reduced conflict
rigsolve verify Run isolated imports and available GPU probes
rigsolve matrix Inspect, update, or extend compatibility facts
rigsolve doctor Check rigsolve and NVIDIA tooling

Plans can be rendered as pip, uv, TOML, Dockerfile, JSON, or Colab output. Only solve --execute installs packages.

See the CLI reference for all options.

Evidence

Label Meaning
Metadata-backed Published artifact or documented build axis
Install-tested Exact artifact installed in a recorded environment
Import-tested Package imported and available build metadata was recorded
GPU-tested A minimal kernel ran on the recorded GPU architecture

Evidence depth is not a probability. Local execution verifies the selected stack on the current machine. It does not make a global claim about other GPUs or systems.

The bundled matrix contains 114 sourced facts and one scoped known_broken entry for the flash-attn 2.8.3.post1 filename mismatch.

rigsolve matrix stats

Read the trust model and matrix schema before contributing evidence.

Safety

  • Detection, solving, diagnosis, and matrix inspection are offline by default.
  • There is no telemetry.
  • Native imports run in child processes.
  • verify --contribute writes a local JSON file and uploads nothing.
  • Matrix updates validate the complete payload before replacing the cache.
  • Install plans are reviewable. Installation requires --execute.

Contributing

Use CONTRIBUTING.md for code and matrix changes. Report new incompatibilities with the known broken issue form. Report security issues through SECURITY.md.

Social assets

License

Apache License 2.0

Project links

Data verified by PyPI on Aug 16, 2026
Data provided by the project maintainers, verified at the time the release was uploaded to PyPI.

Key dates

PyPI data
Data sourced directly from PyPI's database.
  • Released:
Latest release

1 maintainer

PyPI data
Data sourced directly from PyPI's database.
Avatar for satwiksps from gravatar.com satwiksps

Credits

Author: rigsolve contributors

GitHub Statistics

Data verified by PyPI on Aug 16, 2026
The GitHub source repository was provided by the project maintainers and verified by PyPI at the time of upload. Stars, forks, and open issues/PRs are derived from that repository and have not been independently verified.

License expression

Apache-2.0
View SPDX License List

Requires

Python >=3.10

Provides Extra

dev

Tags

cuda dependency-resolution developer-tools flash-attention gpu mlops pytorch

Classifiers

Report project as malware

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

rigsolve-0.1.2.tar.gz (146.7 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

rigsolve-0.1.2-py3-none-any.whl (124.6 kB view details)

Uploaded Python 3

File details

Details for the file rigsolve-0.1.2.tar.gz.

File metadata

  • Download URL: rigsolve-0.1.2.tar.gz
  • Upload date:
  • Size: 146.7 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/7.0.0 CPython/3.13.14

File hashes

Hashes for rigsolve-0.1.2.tar.gz
Algorithm Hash digest
SHA256 14108e8f726762a313734278ff79f8f7c7e1a7f35d50097746c34c2a37b6b48b
MD5 aa81f71f73b3bac61db1ade9d479e204
BLAKE2b-256 bd785c445f63fee124e5099fc192b42953f43e59a8fd9b2cc3228dbfa618bb88

See more details on using hashes here.

Provenance

The following attestation bundles were made for rigsolve-0.1.2.tar.gz:

Publisher: release.yml on satwiksps/rigsolve

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

File details

Details for the file rigsolve-0.1.2-py3-none-any.whl.

File metadata

  • Download URL: rigsolve-0.1.2-py3-none-any.whl
  • Upload date:
  • Size: 124.6 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/7.0.0 CPython/3.13.14

File hashes

Hashes for rigsolve-0.1.2-py3-none-any.whl
Algorithm Hash digest
SHA256 e0733da0af2eec48d0aaac164bd352db5072ee187a2873dc9e96cdc818d91e36
MD5 86be67a17313bc3c25011325790f7f63
BLAKE2b-256 dd4bddcbded62250c059fba588e680731ac0bc02d7246b67646f1f1ce4002dfd

See more details on using hashes here.

Provenance

The following attestation bundles were made for rigsolve-0.1.2-py3-none-any.whl:

Publisher: release.yml on satwiksps/rigsolve

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page