Right now the plainbuffer support is not at the same level of support than the Blosc backend. In addition, NumPy is making great strides in accelerating computations with MKL, SVML and others (for Intel), so for us it would be difficult to compete with plainbuffers performance wise.
So Christian and me decided to get rid of the plainbuffer support inside ironArray (both C and Python) and revisit this in case we are going to implement a wrapper for R or Java, and plainbuffers could be an advantage for us.
Right now the plainbuffer support is not at the same level of support than the Blosc backend. In addition, NumPy is making great strides in accelerating computations with MKL, SVML and others (for Intel), so for us it would be difficult to compete with plainbuffers performance wise.
So Christian and me decided to get rid of the plainbuffer support inside ironArray (both C and Python) and revisit this in case we are going to implement a wrapper for R or Java, and plainbuffers could be an advantage for us.