I seem to remember a series of letters or articles in QEX about the difficulties of characterizing SDR radios and how IP3 was no longer the thing. I'll see if I can find it.