It would appear that the JavaScript versions of the test can
give misleading results when using Internet Explorer versions 3.xx.
The 56K throughput test, for example, takes 5-10 seconds longer
to complete using IE 3.xx, rather than another browser.
Tests using different Internet Service Providers, telephone lines,
POPs and PCs show that the common link is Internet Explorer 3.xx.
I would, however, welcome feedback on this subject.
A brief history follows:
10th June, '99
I have recently noticed that a problem, similar to the CRC error
problem experienced before, exists where the download times for
the JavaScript tests are higher than expected using IE3.xx from
both the main site and the mirror site. As before a "pause"
is experienced, but, this time, no CRC errors are recorded.
20th February, '98
My ISP's System Operator suggests that my problem may have been
due to characters in a PPP frame being wrongly interpreted as
control characters by the PPP client, but without analysis it is
impossible to confirm this.
24th December, '97
The CRC error problem I had been experiencing seems to
have dissappeared in much the same way that it appeared some
months ago.
The problem, a 50% increase in load time, only occured when using
IE3.xx to download JPG files. This does not include the 16-bit
Internet Explorer version 3.01.
A pause was evident during the download of larger JPG files and
CRC errors were also recorded in my PPPLOG.TXT file.
I have not received email to confirm that other Modem Speed Test
Page users experience the pause except for users connected to my
ISP. As I have not knowingly made changes to my PC's setup, I can
only assume that it may be due to upgrade work currently being
carried out by my ISP.
7th December, '97
Recent tests confirm that the problem
only occurs when using IE3.xx to download JPG files. This does
not include the 16-bit Internet Explorer version 3.01.
22nd Nov, '97
Since the problem seems to occur with both IE3.0 and IE3.02, but
not Netscape Navigator 4.03 (or Opera 3.0), it looks like
Internet Explorer causes the pause. I am assuming that I did not
make any changes to my PC's configuration. I do know that my
Internet Service Provider went 56K Flex around a month ago, so I
have not ruled this out. A friend using Win95, IE and the same
ISP confirmed the pause.
I am presently trying to find out why the pause occurs, however,
I would appreciate Email confirming or un-confirming my results.
Please let me know your operating system, modem type, ISPs
maximum connection type (i.e. 28.8, 33.6, x2, K56flex, etc.) and
browser.
On a 28.8Kbps connection a time of 10 to 12 seconds would
indicate no delay, whereas a time of 15-20 seconds may indicate
an occurrence of the CRC pause.
28th Oct, '97
Recently, I have noticed that download times for the graphics and
text test and the throughput test had increased by around 5-8
seconds on my home system (recently changed to Windows 95) and my
work system. Both systems have a standard Windows 95 setup using
Internet Explorer 3.0. The problem, however, is not evident when
using Netscape Navigator 4.03 instead of IE 3.0.
During downloading of the JPG graphic file (test-30k.jpg) data transfer appears to pause for 5 to 8 seconds, when about 70% of the file has been downloaded, before completing the download.
Preliminary tests have shown that the problem occurs during more than 95% of downloads of the JPG file and also sometimes occurs during downloads of the GIF graphic file (test-30k.gif), although in the case of the GIF file, pauses seemed to only occur at lower line speeds (around 14,400bps).
An alternative image the same size gives similar results. I have tried different JGPs, different sizes of JGPs, even large weather satellite pictures, but still get the same results.
It would also appear that each occurrence of the pause coincides with one "CRC Error" as recorded by the Point-to-Point Protocol log file - PPPLOG.TXT.
Last updated on January 9th, 2000
Back to Modem Speed Test Page.
Email: leslielong@eircom.net
© Copyright Leslie Long, 1997-2000