Test results - compression and error correction
Franc Zabkar
modems using various EC and compression protocols. These modems were
connected via a short piece of phone cable, ie a "perfect" line. One
was a V34 modem, the other a 56K modem with ACF2 chipset. The results
are tabulated below.
56K V34
DTE-A ---------> DCE-A -----------------> DCE-B ---------> DTE-B
115200 bps 33600 bps 115200 bps
h/w flow control no flow control
Error Data Transfer MNP Throughput Bits
Correction Compression Time block (bytes/sec) per
(sec) size byte
---------------------------------------------------------------
None None 298 - 3356 10.0
V.42 LAPM None 249 - 4016 8.37
V.42 LAPM V.42bis 87 - 11494 2.92*
MNP 4 MNP 5 218 128 4587 7.32
MNP 4 None 263 64 3802 8.84
MNP 4 None 252 128 3968 8.47
MNP 4 None 244 256 4098 8.20
The text file that was the subject of these tests consisted of lines
of the form "abcd...ABCD...012...9 line#". As can be seen from the
results, V.42bis did a good job of compressing this file whereas MNP
did not. In fact the minimum possible transfer time of 87 secs shows
that the DTE rate, not the modem, was the bottleneck for V.42bis.
One statistic that puzzles me is the 298 sec time for an "unbuffered"
transfer. This figure was reproduced consistently, as was a figure of
595 secs for a file of twice the size. My previous tests for the same
file yielded a time of 294 secs (see the thread entitled "RS-232
speed"). While I can't explain the difference, I _did_ use a different
software configuration this time:
Configuration 1 --> 294 secs
DTE-A --------/\/\/\/\-------> DTE-B
Win95B Win95B
HyperTerminal HyperTerminal
Configuration 2 --> 298 secs
DTE-A --------/\/\/\/\-------> DTE-B
Win95B DOS Win95B
Telix for DOS HyperTerminal
The result for configuration 1 suggests that the modems were using an
apparent character size of 9.875 bits. As explained in a prior thread,
this is due to V.14 which allows the DCE to drop one stop bit in every
10 bytes, for clock synchronisation reasons. However, in configuration
2 the modems appear to be ignoring V.14. I can't understand why this
should be the case. I've compared all the modem settings but I can't
see any differences.
One thing I have noticed, though, is that DOS apps service the COM
port much faster than Win apps. For example, while Telix can transmit
the above text file at 11.5 KBps, HyperTerminal maxes out at about
4KBps. Originally I thought that this limitation may be a Win95 issue.
However, I have used the same modem to upload and download highly
compressible data to my own mailbox, and have achieved sustained
transfer rates of 11.5KBps in both directions. For these particular
tests I used a text file consisting of the letter "a" repeated one
million times. For whatever reason it appears that DUN is much more
efficient than HyperTerminal in handling the COM port.
-- Franc Zabkar
Please remove one 'g' from my address when replying by email.