When using CXR as the image output format, the entire image is saved as an uncompressed 32-bit image. In many cases this results in an astronomical file size that takes a long time to save, when most uses only require 32-bit.
It would be incredibly useful for the user to be able to define the bit depth (32 bit float, 16 bit float) and compression (ZIP, ZIPS etc.) for each render element - much like you can in the standard 3DS Max EXR setup.
(specifically it is ZLib compression in a block of 16 scan lines, it is a part of OpenEXR format)
Just to clarify: the CXR images are not uncompressed, there is compression used.