| Adrian's Home PMC Electronics Easynet (ISP) FreePPP Site Email Adrian NB Modem speeds are invariably quoted in
bits per second (bps). File downloads are more usually indicated in bytes
per second (bytes/sec). There are 8 bits in a byte so, in theory, to get the bps just
multiply the bytes/sec by 8.
It has been suggested to me however that due to the start and stop bits which normally accompany each byte a more accurate translation would be based on 10 to 1...which is nice...since it makes the stats look even more impressive! There are various factors, in addition to line quality and Net congestion, which limit your download speed to below that of the theoretical maximum.
The modems require to send control information for synchronization and communicate to check for error correction re-sends, for example. They check the line quality and may renegotiate the connection speed up or down, depending on circumstances. This control information is in addition to the actual data which makes up the file itself.
|
|
56K Modem Performance Using FreePPP and OT/PPP
There does indeed seem to be a limit of 57,600bps when using OT/PPP which prevents these Macs from taking the fullest advantage of the highest speed modems. FreePPP on the other hand includes a "hack" which allows the ports to be run at 115,200bps (or higher). Having read about this difference I thought that I should do some comparative tests. The following bar charts show the results of 5 different downloads. 4 of the files were downloaded by FTP using Fetch v3.0.3, and the "all groups" list was downloaded using the MacSoup v2.3.3 newsreader. The setup of the Mac was identical other than the OT/PPP and FreePPP alternation. The script used with OT/PPP was that provided by the modem manufacturer (Pace 56 Voice Mac, version 2.0.0) . The initialization string used with FreePPP was simply AT&F. Each download was carried out 3 times and the
average of the best 2 is the figure used. In practice I managed to achieve
sufficient consistency/repeatability that a simple one-to-one comparison
would have given virtually identical results; however, I had to prove to
myself that the speeds were truly representative.
![]() ![]() ![]() ![]() ![]() ConclusionConnecting using FreePPP (with my particular set-up) is clearly the faster method. The difference is most significant on files which allow the modem to make full use of its compression features. On these files FreePPP approached speeds almost double those using OT/PPP. It is essential to have AppleTalk OFF to achieve good performance. Even if Appletalk is switched to use an Ethernet card rather than the serial port this still causes a significant degradation of performance on these Performas.Though the port speed setting is not explicit when using OT/PPP the above numbers do suggest that a 57,600bps limit is being imposed. More worrying is that on the text FTP downloads using OT/PPP the transfer did appear to be running at higher speeds but then slowed towards the end. Perhaps a different script would make a difference. Initial connection speeds reported by FreePPP and OT/PPP were identical for concurrent sessions (48000bps or 50000bps being typical). During the tests the peak speed reported by FreePPP was 93,600bps; this is just a wee bit more than the 9,600bps my Mac's modem port is supposedly restricted to....thanks FreePPP, Pace and Easynet! August 1998 |
|