adjusting to latest imex

Author: fschlimb

README.md

@@ -4,14 +4,14 @@ A tensor implementation following the [array API as defined by the data-API cons
 It supports a controller-worker execution model as well as a CSP-like execution.
 
 ## Setting up build environment
-Install MLIR/LLVM and IMEX from branch refactor (see https://github.com/intel/mlir-extensions/tree/refactor).
+Install MLIR/LLVM and IMEX from branch dist-ndarray (see https://github.com/intel-innersource/frameworks.ai.mlir.mlir-extensions/tree/dist-ndarray).
 ```bash
 git --recurse-submodules clone https://github.com/intel-sandbox/personal.fschlimb.ddptensor
 cd personal.fschlimb.ddptensor
+git checkout jit
 conda env create -f conda-env.yml -n ddpt
 conda activate ddpt
 export MPIROOT=$CONDA_PREFIX
-export MKLROOT=$CONDA_PREFIX
 export MLIRROOT=<your-MLIR-install-dir>
 export IMEXROOT=<your-IMEX-install-dir>
 ```
@@ -21,7 +21,7 @@ python setup.py develop
 ```
 If your compiler does not default to a recent version, try something like `CC=gcc-9 CXX=g++-9 python setup.py develop`
 
-## Running Tests
+## Running Tests [non functional]
 __Test are currently not operational on this branch.__
 
 ```bash
@@ -58,11 +58,11 @@ dt.fini()
 ```
 Assuming the above is in file `simple.py` a single-process run is executed like
 ```bash
-python DDPT_IDTR_SO=`pwd`/ddptensor/libidtr.so python simple.py
+DDPT_IDTR_SO=`pwd`/ddptensor/libidtr.so python simple.py
 ```
 and multi-process run is executed like
 ```bash
-python DDPT_IDTR_SO=`pwd`/ddptensor/libidtr.so mpirun -n 5 python simple.py
+DDPT_IDTR_SO=`pwd`/ddptensor/libidtr.so mpirun -n 5 python simple.py
 ```
 
 ## Contributing

src/jit/mlir.cpp

@@ -396,21 +396,17 @@ std::vector<intptr_t> JIT::run(::mlir::ModuleOp &module,
 static const char *pass_pipeline =
     getenv("DDPT_PASSES")
         ? getenv("DDPT_PASSES")
-        //    :
-        //    "func.func(ptensor-dist),convert-dist-to-standard,convert-ptensor-to-linalg,arith-expand,canonicalize,arith-bufferize,func.func(empty-tensor-to-alloc-tensor,scf-bufferize,linalg-bufferize,tensor-bufferize),func-bufferize,canonicalize,func.func(finalizing-bufferize,convert-linalg-to-parallel-loops),canonicalize,fold-memref-alias-ops,lower-affine,convert-scf-to-cf,convert-memref-to-llvm,convert-func-to-llvm,reconcile-unrealized-casts";
-        //    :
-        //    "builtin.module(func.func(ptensor-dist),convert-dist-to-standard,convert-ptensor-to-linalg,arith-bufferize,func.func(empty-tensor-to-alloc-tensor,scf-bufferize,linalg-bufferize,tensor-bufferize,bufferization-bufferize),func-bufferize,func.func(finalizing-bufferize,convert-linalg-to-parallel-loops),canonicalize,fold-memref-alias-ops,expand-strided-metadata,lower-affine,convert-scf-to-cf,convert-memref-to-llvm,convert-func-to-llvm,reconcile-unrealized-casts)";
         : "func.func(ptensor-dist,dist-coalesce),convert-dist-to-standard,"
           "convert-ptensor-to-linalg,canonicalize,convert-shape-to-std,arith-"
-          "expand,canonicalize,arith-bufferize,func-bufferize,func.func(tosa-"
-          "to-linalg,"
-          "empty-tensor-to-alloc-tensor,scf-bufferize,tensor-bufferize,linalg-"
-          "bufferize,bufferization-bufferize,linalg-detensorize,tensor-"
+          "expand,func.func(tosa-to-linalg,canonicalize,linalg-fuse-"
+          "elementwise-ops,empty-tensor-to-alloc-tensor),memref-expand,arith-"
+          "bufferize,func-bufferize,func.func(scf-bufferize,tensor-bufferize,"
+          "linalg-bufferize,bufferization-bufferize,linalg-detensorize,tensor-"
           "bufferize,finalizing-bufferize,convert-linalg-to-parallel-loops),"
           "canonicalize,fold-memref-alias-ops,expand-strided-metadata,convert-"
-          "math-to-funcs,convert-math-to-libm,lower-"
-          "affine,convert-scf-to-cf,convert-memref-to-llvm,convert-func-to-"
-          "llvm,reconcile-unrealized-casts";
+          "math-to-funcs,convert-math-to-libm,lower-affine,convert-scf-to-cf,"
+          "finalize-memref-to-llvm,convert-func-to-llvm,reconcile-unrealized-"
+          "casts";
 JIT::JIT()
     : _context(::mlir::MLIRContext::Threading::DISABLED), _pm(&_context),
       _verbose(false) {
@@ -477,9 +473,9 @@ void init() {
   ::mlir::bufferization::registerBufferizationPasses();
   ::mlir::arith::registerArithPasses();
   ::mlir::registerAffinePasses();
-  ::mlir::registerMemRefToLLVMConversionPass();
   ::mlir::registerCanonicalizerPass();
   ::mlir::registerConvertAffineToStandardPass();
+  ::mlir::registerFinalizeMemRefToLLVMConversionPass();
   ::mlir::memref::registerMemRefPasses();
   ::mlir::registerReconcileUnrealizedCastsPass();