caio/clang-p2996
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[mlir][sparse] unify support of (dis)assemble between direct IR/lib path (#71880)

Browse Source 
Note that the (dis)assemble operations still make some simplfying
assumptions (e.g. trailing 2-D COO in AoS format) but now at least both
the direct IR and support library path behave exactly the same.

Generalizing the ops is still TBD.
This commit is contained in:
Aart Bik
2023-11-13 10:05:00 -08:00
committed by GitHub
parent 5fdb70be7b
commit af8428c0d9
9 changed files with 312 additions and 297 deletions
Download Patch File Download Diff File Expand all files Collapse all files

13
mlir/lib/Dialect/SparseTensor/Transforms/SparseTensorCodegen.cpp
View File

@@ -435,19 +435,6 @@ static ReassociationIndices getReassociationForFlattening(ShapedType srcTp) {
return reassociation;
}
/// Generates scalar to tensor cast.
static Value genScalarToTensor(OpBuilder &builder, Location loc, Value elem,
Type dstTp) {
if (auto rtp = dstTp.dyn_cast<RankedTensorType>()) {
// Scalars can only be converted to 0-ranked tensors.
if (rtp.getRank() != 0)
return nullptr;
elem = genCast(builder, loc, elem, rtp.getElementType());
return builder.create<tensor::FromElementsOp>(loc, rtp, elem);
}
return genCast(builder, loc, elem, dstTp);
}
//===----------------------------------------------------------------------===//
// Codegen rules.
//===----------------------------------------------------------------------===//