conda-forge / conda-forge/tensorflow-feedstock

libtensorflow_cc cuda contains about 1GB of headers (extracted) (cpu contains like 350MB)

Open
#463 6 comments 0 reactions 0 assignees View on GitHub
question
Dominant language
Shell
Stars
93
Forks
93
PR merge metrics
No merged PRs in 30d

Description

### Comment:

```
ncdu libtensorflow_cc-2.19.0-cuda128he143472_253
--- /home/mark/miniforge3/pkgs/libtensorflow_cc-2.19.0-cuda128he143472_253 ----------------------------
1.1 GiB [##############] /lib
926.7 MiB [########### ] /include
3.8 MiB [ ] /info
```

most of it seems like llvm/.../Targets

```
--- /home/mark/miniforge3/pkgs/libtensorflow_cc-2.19.0-cpu_hf7abd0a_3 ---------------------------------
611.9 MiB [##############] /lib
368.8 MiB [######## ] /include
3.7 MiB [ ] /info
```

Contributor guide

Open the contributing guide

Research direction

Start by reproducing the reported ncdu measurements for the libtensorflow_cc-2.19.0-cuda128he143472_253 and libtensorflow_cc-2.19.0-cpu_hf7abd0a_3 packages. Inspect the extracted include trees, especially llvm/.../Targets, and then trace those files back through the tensorflow-feedstock recipe. Done means the source of the oversized headers is identified and an agreed package-size reduction is demonstrated without breaking the packages.

Written by the indexing model from the issue text.

Assessment

Tech stack
tensorflow
Domain
build-system
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.