Conversation
Add a CUDA build of iscc-web so the experimental semantic codes (iscc-sct text, iscc-sci image; both ONNX) run on an NVIDIA GPU, while the default published image stays the lean CPU python:3.13-slim build. Two tag streams from one repo: :X.Y.Z (CPU, unchanged) and :X.Y.Z-gpu (CUDA). iscc-sct 0.2.0 / iscc-sci 0.3.0 moved onnxruntime out of their base deps into mutually-exclusive cpu|gpu extras, so iscc-web now selects a variant: - pyproject: cpu/gpu optional-dependencies (iscc-sct[cpu|gpu]>=0.2.0, iscc-sci[cpu|gpu]>=0.3.0), onnxruntime in the dev group for local CPU dev/tests, and a [tool.uv] conflicts table (cpu<->gpu, dev<->gpu) so the clobbering onnxruntime variant wheels never co-resolve. - uv.lock carries both onnxruntime and onnxruntime-gpu; --extra cpu resolves plain onnxruntime, --extra gpu resolves onnxruntime-gpu. - Dockerfile: CPU prod-build gains --extra cpu; a single combined prod-runtime-gpu stage builds on nvidia/cuda:12.6.3-cudnn-runtime- ubuntu24.04 with a uv-managed Python 3.13 and --extra gpu. The in-image model bake doubles as an import-resolution check. - CI: pin target: prod-runtime on every CPU build (the GPU stage is now last/default); release.yml and ci.yml publish the -gpu stream (suffix=-gpu) alongside the unchanged CPU tags, keeping latest=false. GPU builds only on release + push (not PRs). Reconcile release.yml's stale model pre-download with the current get_model() API. - Docs: note the -gpu tag, its host requirements (NVIDIA driver + container toolkit + --gpus all) and CPU fallback; deployment specifics live in iscc-infra.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Release v0.3.1.
Changes
-gputags (:0.3.1-gpu,:0.3-gpu,:main-gpu) on a CUDA base, running the experimental semantic codes on an NVIDIA GPU. Thedefault image stays the lean CPU
python:3.13-slimbuild. GPU deployment lives in iscc-infra.Release mechanics
Merging this triggers
ci.ymlto publish:mainand:main-gpu. Publishing thev0.3.1GitHub release afterward triggersrelease.ymlto publish the immutable:0.3.1/:0.3(CPU) and:0.3.1-gpu/:0.3-gpu(GPU) tags.latest=falsethroughout.