Some of these arguments are just sad. PG quotes:
"In one lab study, the CPU load of our modem host controller was less than that of the Windows serial driver handling a standard external modem while downloading identical files at identical DCE rates. The fear that a well designed "Win Modem" will tax your precious CPU cycles is simply not true."
Notice that this "ONE lab study" still has to be given an out by including "a well designed 'Win Modem'". Unfortunetly, finding a well designed win-modem, outside of a lab and in a retail package for a good price, isn't all that common. As for what they chose as thier "standard external modem" to compare with, god only knows.
Cybordolphin gives us a link to a modem review. While some people believe that win-modems transfer slower than thier strictly hardware cousins, I don't believe that's been a problem on any well made modem. The problem with win-modems is that they tax your system. I'd like to see that review done while the systems were actually being used for something in addition to the modem tests. I also notice the very limited selection of external modems that were tested. Not much use as a comparison to the internals. BTW, were any of the internals tested not win-modems? Most companies make it much less obvious than USR and thier "Win modem" line.
Hmmm, what was the topic of this tread again...?