Lightning-AI / Lightning-AI/lightning-thunder
check memory location of things tagged STATIC_MEMORY_LOCATION by default
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 1.5k
- Forks
- 121
- PR merge metrics
- No merged PRs in 30d
Description
...in the prologue.
This could be tricky because it needs the real tensors, in particular if we try to move materializaton to later.
Implementation strategy could be
- add the data_ptr as an optional field to the TensorProxy
- include this in the check metadata
Note that pruning the checks would give the option to remove it, which is a good thing IMO.
Contributor guide
No contributing guide indexed for this repository
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by tracing the prologue and the check-metadata path around TensorProxy, focusing on how real tensors and materialization timing affect memory-location checks. Done means the default STATIC_MEMORY_LOCATION checks can inspect the relevant tensor locations, with pruning able to remove those checks as noted.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python, pytorch
- Domain
- compilers
- Issue type
- Feature
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100