Raisetolinalg - #412

Draft

arpitj1 wants to merge 171 commits into

llvm:raisetolinalgfrom

arpitj1:raisetolinalg

Commits on Jun 6, 2024

Commits on Jun 11, 2024

Partial changes from coding session to implement fusion of linalg.generic and for op
arpitj1
committed

Commits on Jun 19, 2024

Incremental changes to fuse linalg and for loop- Logic for shifted operands and map for linalg.generic
arpitj1
committed

Commits on Jun 25, 2024

Commits on Jul 2, 2024

Some compile fixes
arpitj1
committed

Commits on Jul 3, 2024

Fixed all the compilation issues. Sample MLIR not raised
arpitj1
committed

Commits on Jul 16, 2024

Bug fixes, generating some output at getLinalgArgMap
arpitj1
committed

Commits on Jul 17, 2024

Almost implementated remap in affine dim for multi idx
arpitj1
committed

Commits on Jul 24, 2024

Added submap op support and refactored the code to use submap
arpitj1
committed

Commits on Jul 30, 2024

bunch of fixes. Now able to generate raise linalg code
arpitj1
committed

Commits on Jul 31, 2024

Commits on Aug 6, 2024

Fix for syms and dims calculation
arpitj1
committed

Commits on Aug 7, 2024

Commits on Aug 21, 2024

submap canonicalizer implemented
arpitj1
committed

Commits on Aug 22, 2024

Added reduction loops for linalg
arpitj1
committed

Commits on Aug 28, 2024

Fix for incorrect for loop dims
arpitj1
committed

Commits on Sep 5, 2024

Linalg.generic 4 loop cases raised- todo: reduction and some if-else cases failing
arpitj1
committed

Commits on Sep 18, 2024

Adding test case for all passing raising and lowering, example case of debufferizing added which works for tiling and fusion
arpitj1
committed

Commits on Oct 12, 2024

Commits on Oct 15, 2024

WIP commit
arpitj1
committed

Commits on Oct 18, 2024

Added submap of submap canonicalizer with test- failing
arpitj1
committed

Commits on Oct 25, 2024

Commits on Jan 17, 2025

Commits on Feb 1, 2025

Canonicalization gets rid of memref capture by loop
arpitj1
committed

Commits on Feb 7, 2025

Commits on Feb 8, 2025

Commits on May 8, 2025

Generic solver to represent linalg.generic as kernel.def ops
arpitj1
committed

Commits on May 12, 2025

Adding cases for generic solver
arpitj1
committed

Commits on May 28, 2025

Backup of previous edits
arpitj1
committed

Commits on Jun 11, 2025

Temp changes for kernel dialect
arjaiswal
committed
Enabled kernel dialect correctly running on sample IR with kernel defn collection
arjaiswal
committed

Commits on Jun 12, 2025

Commits on Jun 13, 2025

Crash fix
arpitj1
committed

Commits on Jun 26, 2025

Commits on Jun 27, 2025

Fixed a bunch of bugs in raiseToLinalg while raising polybench
arpitj1
committed

Commits on Jun 28, 2025

Fixed raise to linalg and canonicalizer to generate subview
arpitj1
committed

Commits on Jul 31, 2025

Fixed submap simplification, improved raisedToLinalg to work with non constant bounds, added debufferization to work with allocs
arpitj1
committed

Commits on Aug 3, 2025

Commits on Aug 4, 2025

Able to raise gemv
arpitj1
committed

Commits on Oct 15, 2025

Commits on Oct 17, 2025

Commits on Jan 10, 2026

Implemented improvement in linalg debufferize to work through inversesubmap, and a more shophisticated approach taken for tracking current tensors via a tree of regions
arpitj1
committed

Commits on May 13, 2026

Commits on May 14, 2026

Commits on May 15, 2026

Commits on May 23, 2026

Commits on May 24, 2026

Commits on May 25, 2026

Commits on May 26, 2026

PVA backend: lower kernel.launch to libpva_operator for int8/int16 conv2d + 4 image filters on Jetson Orin

arpitj1
and
claude
committed

Commits on May 29, 2026

Commits on May 30, 2026

Add tensor LLM lowering paths
arpitj1
committed

Commits on May 31, 2026

Commits on Jun 1, 2026

Commits on Jun 7, 2026

Commits on Jul 21, 2026

Raise nested reductions and lower tensor products to cuTensorNet
arpitj1
committed

Commits on Jul 24, 2026

Commits on Jul 25, 2026

Generalize MFEM tensor contraction matching
arpitj1
committed