better MaterializeMergeSchedule.cpp with %lane indexed batch computes

support for tensors of index values
2026-05-22 21:52:28 +02:00
parent 495186503c
commit c77ffa9c56
20 changed files with 398 additions and 300 deletions
@@ -264,7 +264,7 @@ llvm::FailureOr<ResolvedContiguousAddress> resolveContiguousAddressImpl(mlir::Va
        return mlir::failure();

      auto sourceStrides = computeRowMajorStrides(sourceType.getShape());
-      byteOffset += linearizeIndex(offsets, sourceStrides) * subviewType.getElementTypeBitWidth() / 8;
+      byteOffset += linearizeIndex(offsets, sourceStrides) * getElementTypeSizeInBytes(subviewType.getElementType());
      value = resolveAlias(subviewOp.getSource(), knowledge);
      continue;
    }