* Fix uninitialized new operator

* Fix

* test

* Fix memory bug

* Add SmallString

* Make shmem buffer elim more aggressive

* fix non domination issue

* Fix

Use a new ArgumentList class instead of malloced pointers to pass
arguments to drivers.

This...

* Add parallel lower fix

* fixup

* Update APIs to match LLVM to llvmorg-16-init-8549-gcbc378ecb87e

* Stop the ridiculously fra...

* Enable shared library build.

* This patch can enable Polygeist as shared library. aka, when...

* Correct struct pointer offset calc

* fix test

* fix test

* WIP: alternative lowering for memrefs

Introduce an alternative lowering scheme for memref t...

* Use funcop for indirect call

* Add test

* Add comment

* Handle builtin_expect

* add warning

* Fix ext_vector_type test; the test did not correctly checked the input

* update ext_vector_...

* Bump llvm

* Fix polygeist-opt tests

* Do not use opaque pointers

* Do not return null...

* Early inner serialization

* Add loweraffine

* Add serial licm improvement

* Fix yield...

* Add aggressive parallel merging

* New infra

* bugfix

* Continue

* functioning

*...

* Affraise improve

* Handle ext on affraise

* Eliminate read-write copy buffers

* Fix

* Fix test

* add doublebuf test

* Fix test

* If simplification pt 2

* fix bug

* More affine simplifications

* Fix bug

* Affine ...

* Affine if simplification

* fix and add test

* Improve debug info

* More debug info

* Preserve pass locations

* Enable constant rank reduction

* Fix tests

* Fix for with break to while lowering

* Fix conversion

* Ensure same fcmp order as clang

* Update test

* Simplify return value from a for with a break

* Fix bug

* Add a pass converting scf.for(scf.if) to scf.while

The following construct

scf.for %...

* Negate lookthrough

* Improve for break perf

* Add if remove unused

* Fix tests

Storing undefs into cuda shared mem results in wrong semantics because it causes all threads in ...

authored-by: pietro.ghiglio <[email protected]>

[BUGFIX] correctly emitting calls with addressoff operator

Formatting

moved test to cgeist...

* Rename mlir-clang to cgeist

* Remove todos

* Add support for some builtin std:: functions

clang::Builtin::BImove
clang::Builtin::BImove...

* CUDA Stream support

* Async lowering [WIP]

* Fix lowering to moccuda

* Convert to mal...

* Handles basic syntax using ext_vector_type type.
* Improvement is needed to support more comp...

* Call device stub from host code for global functions

* Update tests

* Don't reverse inpu...

- FuncOp namespace
- include path for moveLoopInvariantCode

* Fix barriers betting removed when they are required for interchange

* Fix distributing afte...

* Upgrade mem2reg

* Fix test

* Correct barrier elimination

* Fix affine raising

* Fi...

* Don't prefix malloc/free

* C++ Foreach

* Fix lookup where no base is created in llvm abi

* Add test

* Bump LLVM

* Fix API chan...

* Prefix ABI

* Fix format

Co-authored-by: William S. Moses <[email protected]>

* Fix Parallel LICM condition

* Fix ParallelLICM

* Fix inheritence base lowering

* Fix ...

* Fix

* Only allow constant rems

* foo

* Fix emit direct callee

* Add fetch add

*...

run-clang-tidy -fix tools/mlir-clang/Lib/* -checks=-*,llvm*
run-clang-tidy -fix lib/polygeist/* ...

* Add Try

* Fix placement new

* Fix if interchange

* Fix allocation region

* Fix format

* Fix test

* Fix sizecheck...

Parallel loop distribute cache min cut optimisation

Properly remap values to cloned recomputations in parallel regions

Large scale refactoring to eliminate code duplication