Published January 9, 2026
| Version v0.56.1
Software
Open
xdslproject/xdsl: v0.56.1
Authors/Creators
- Sasha Lopoukhine
- Fehr Mathieu1
- Emilien Bauer2
- Alex Rice
- Anton Lydike3
- Nicolai Stawinoga
- jumerckx
- Tobias Grosser1
- Michel Weber
- Chris Vasiladiotis2
- Dalia Shaaban
- George Bisbas4
- Edmund Goodman1
- kingiler
- Nick Brown2
- David K
- Joren Dumoulin5
- Hugo Pompougnac
- kayode-gif
- Emma Urquhart
- Hendrik_Klug
- Théo Degioanni6
- KGrykiel
- Prathamesh Tagore7
- Christian Ulmann8
- Josse Van Delm9
- Max Manainen10
- 1. University of Cambridge
- 2. University of Edinburgh
- 3. The University of Edinburgh
- 4. Imperial College of London
- 5. MICAS (KU Leuven)
- 6. NVIDIA
- 7. ETH Zürich
- 8. NextSilicon
- 9. KU Leuven
- 10. Rivos Inc.
Description
Happy New Year!
Changes
xDSL framework
- core: (rewriter) Fix replace_op name hints @ed741 (#5534)
Dialects
- dialects: (riscv_snitch) migrate DM ops to declarative assembly_format @szerdick (#5540)
- dialects: (riscv_snitch) migrate dmstr to declarative assembly_format @szerdick (#5532)
- dialects: (riscv) add shift right immediate canonicalization @nmjfry (#5518)
- dialects: (riscv) annotate rd type in RdRsRsOperation @superlopuh (#5526)
- dialects: (scf) add if/else branch short circuit canonicalization pattern @sueszli (#5515)
- dialects: (riscv) add canonicalization pattern for OR and XOR with const @sueszli (#5514)
- dialects: (scf) Add
scf.execute_region@jumerckx (#5504)
Transformations
- transformations: (convert-pdl-to-pdl-interp) generate_rewriter @jumerckx (#5530)
- transformations: (riscv-lower-parallel-mov) Add support for moving floats @anominos (#5512)
- transformations: (riscv-lowering-parallel-mov) Add helper to insert MvOp @anominos (#5533)
- transformations: (lower-riscv-func) avoid GreedyRewritePatternApplier @superlopuh (#5535)
- transformations: (riscv-lower-parallel-mov) fix some lint errors @superlopuh (#5527)
- transformations: (convert-pdl-to-pdl-interp) _generate_rewriter_for_replace @jumerckx (#5524)
- transformations: (riscv-lower-parallel-mov) Cache output register index @anominos (#5522)
- transformations: (riscv-lower-parallel-mov) Use xor swapping for cycles with no free registers @anominos (#5508)
- transformations: (riscv-lower-parallel-mov) Refactor to use existing builder methods. @anominos (#5513)
- transformations: (scf) Add canonicalization pattern for
scf.execute_region@jumerckx (#5505)
Backend
- backend: fix register allocation for operations with multiple outputs @anominos (#5538)
- backend: (llvm) add type conversion @sueszli (#5525)
- backend: (mps) init project @sueszli (#5520)
- backend: (llvm) add LLVM output functionality to xDSLOptMain @sueszli (#5519)
- backend: (llvm) add numba/llvmlite boilerplate @sueszli (#5516)
Testing
- testing: use unittest.mock instead of manual override for pyclip_copy @superlopuh (#5531)
Dependencies
- dependencies: lock file maintenance @renovate[bot] (#5529)
- dependencies: lock file maintenance @renovate[bot] (#5498)
Miscellaneous
- misc: add type stubs to used pyclip functions @sueszli (#5537)
Files
xdslproject/xdsl-v0.56.1.zip
Files
(4.5 MB)
| Name | Size | Download all |
|---|---|---|
|
md5:d34002cadc1b9b0b7ad0ad5cfa57efca
|
4.5 MB | Preview Download |
Additional details
Related works
- Is supplement to
- Software: https://github.com/xdslproject/xdsl/tree/v0.56.1 (URL)
Software
- Repository URL
- https://github.com/xdslproject/xdsl