The MultiplyAdd functions that take a Bias with interpretation should convert the Bias vector to the output type (if needed) in the linalg.h header before passing it into __builtin_LinAlg_MatrixVectorMultiplyAdd built-in and eventually dx.op.linAlgMatVecMulAdd op.
Spec update: microsoft/hlsl-specs#857
The MultiplyAdd functions that take a
Biaswith interpretation should convert theBiasvector to the output type (if needed) in thelinalg.hheader before passing it into__builtin_LinAlg_MatrixVectorMultiplyAddbuilt-in and eventuallydx.op.linAlgMatVecMulAddop.Spec update: microsoft/hlsl-specs#857