[mlir][NFC] Fix comment explaining ConverVectorLoad

[tyb0807]

Summary

The new number of elements should be the original one divided by a scale factor computed from old and new bit width

[dcaballe]

Thanks! :)

(I think we should be using github for new changes. IIRC, Phab is about to move to read-only mode)

mlir/lib/Dialect/Vector/Transforms/VectorEmulateNarrowType.cpp

matchAndRewrite(vector::LoadOp op, OpAdaptor adaptor,
     // Here only the 1-D vector load is considered, and the N-D memref types
     // should be linearized.
     // For example, to emulate i4 to i8, the following op:
     //
     // %1 = vector.load %0[%c0, %c0] : memref<3x4xi4>, vector<4xi4>
     //
     // can be replaced with
     //
-    // %1 = vector.load %0[%linear_index] : memref<12xi8>, vector<2xi8>
+    // %1 = vector.load %0[%linear_index] : memref<6xi8>, vector<2xi8>
     // %2 = vector.bitcast %1 : vector<2xi8> to vector<4xi4>
     //
     // TODO: Currently, only the even number of elements loading is supported.
     // To deal with the odd number of elements, one has to extract the
     // subvector at the proper offset after bit-casting.
 
     auto origElements = op.getVectorType().getNumElements();
     if (origElements % scale != 0)