Access Transformation: Add support for multi-dimensional tensors
Open
- Dominant language
- MLIR
- Stars
- 906
- Forks
- 171
- Avg merge
- 4d 12h
- Merged PRs (30d)
- 32
Description
Our current implementation only supports secret-index reads/writes for single dimensional tensor. We need to add support for multi-dimensional tensor read/write operations at:
1) Secret indices where accesses at each dimension should be transformed, and
2) Mixed secret-and-public indices where only secret-index accesses at respective dimensions should be transformed.
This transformation is part of the data-oblivious transformation pipeline (#777).
Contributor guide
Assessment
This issue has not been assessed yet.