Due to the shapes and probable aspect ratios of the samples you probably will not be able to relate the results for the different shape samples if you are just applying a compressive load in some way.
A set or average of two samples will not provide a high degree of comfidence. If you cast additional samples, you will also have a problem with a curing factor and assurance that the same materials were used.
What do you mean by nominal reinforecement? Is this in addition to the fibers? If you are casting reinforcement in the sample, you will not be measuring the properties of the concrete itself. It was always fun to add a rebar to a cylinder, but the results were worthless.
Beams are normally tested in flexure to measure tensile strength or the effects of the reinforcement amount and placement. Depending on the load application and dimensions your failure could be measuring different properties.
Cylinders are normally twice as high as the diameter and compression testing usually results in a shear failure.
Cubes, having idential height and width, will give you results closer to pure compression with less measurable effect of the fibers.
Dick