Recently, there has been some controversy concerning the role of signal-to-noise ratio (SNR) calculations in conjunction with optical amplifiers and phase-shift keying. Analysis has shown that the large-signal SNR at the output of an optical parametric amplifier (OPA) or optical parametric oscillator (OPO) can exceed the shot-noise limit when an injected signal beam is used to elicit small-signal gain saturation. Although no small-signal SNR gain is obtained in this way, the large-signal SNR improvement does translate into a communication performance improvement if binary phase-shift keying is applied to the resulting output beam. Analysis has also shown that the large-signal, shot-noise limited SNR can be exceeded when a binary phase-shift keyed coherent-state beam is applied to a preamplify-feedforward-homodyne receiver. In this case, however, there is no communication performance improvement. Indeed, the preamplify-feed-forward-homodyne receiver requires twice as many signal photons, on average, to equal the error probability of ordinary homodyne detection. In this talk we will show that there is no contradiction between these two examples.