Per the spec: - This operation is only available for matrices with `MatrixUse::Accumulator` and `Wave` or `ThreadGroup` scope - The validator will ensure that the group shared target memory is large enough for the write.
Per the spec:
MatrixUse::AccumulatorandWaveorThreadGroupscope