Information Page
How the tests work
Factors
Port and line speed
 
Related Links
Speed Tips
Trouble-shooting
CRC Errors
ISP Bandwidth

Information Page

The purpose of the test pages is to provide an easy way of measuring Internet connection performance. If you spend a lot of time waiting for your browser to load pages from the Internet then this is the page for you.

How the test pages work

The tests measure the time taken to download a file of known file size from the server. The average throughput can be calculated by applying the following formula:

Average Throughput = Bytes transferred  x  8(bits in a byte)
...........  ........Download time 

Note that Internet 'overload' and telephone line quality can increase the time. Other factors, such as processor speed, memory, browser efficiency, operating system, video card speed and server performance can also affect the transfer time.

Factors that can affect the test

Initialisation string
DTE rate and compression
Error control
Line quality
Processing power
Software
Internet activity

Initialisation string

The initialisation string is probably the most common cause of data speed problems. A browser's dialler program sends commands to the modem before dialling the required telephone number. These commands set-up the modem for different types of connections. Apart from some basic commands, most commands are specific to a particular modem. A command that disables compression on one modem might do something completely different on another modem. Incorrect modem settings in the dialer program may cause very slow connections or other unpredictable problems. Sometimes, it is possible to improve performance by adding to or changing the initialisation string.

DTE rate and compression

DTE rate is the rate at which your computer sends and receives data to and from your modem. If data compression is used, the DTE rate should be set as high as possible. The ITU V.42bis compression standard can achieve 4-to-1 compression, and automatically switches off if data cannot be compressed. The MNP 5 compression standard can achieve 2-to-1 compression, but does not switch off if data cannot be compressed, and slows down data transfer as a result. The table below compares download times for the text test page on connections with a line rate of 14.4 Kbps for DTE rates of 19.2 Kbps and 115.2 Kbps, with V.42bis compression enabled and disabled.

TYPICAL DOWNLOAD TIMES
Line Rate / DTE Rate(bps) Compression ON* Compression OFF
14400 / 115200 3.1 13.8
14400 / 19200 12.3 13.8

*V.42bis data compression

If compression is disabled and the DTE rate is higher than the line rate it would be possible to guess your current line rate from the table below. You should only disable compression in order to perform the test, as it will reduce your perceived speed. Alternatively, the graphic test page will give an indication of the current line rate.

TYPICAL DOWNLOAD TIMES
Line Rate / DTE Rate (bps) Transfer rate (cps) Estimated 24.3K Download Time (secs.) Typical Download Time (secs.)*
33600 / 115200 4200 5.9 6.1
31200 / 115200 3900 6.4 6.6
28800 / 115200 3600 6.9 7.2
26400 / 115200 3300 7.5 7.7
24000 / 115200 3000 8.3 8.5
21600 / 115200 2700 9.2 9.5
19200 / 115200 2400 10.4 10.7
16800 / 115200 2100 11.8 12.0
14400 / 115200 1800 13.8 13.8

*Data compression is NOT used on these tests

Note that compression is most effective on text files like HTML files and has little affect on files that are already compressed such as ZIP files and JPG graphic files.

Error control

There are two common types of error control, LAPM and MNP. LAPM is an error control protocol specified in ITU-T Recommendation V.42. MNP has different levels with levels 2-4 being included in ITU-T Recommendation V.42.
A typical load time for the JPG test page is 9.2 seconds using LAPM (V.42) error control, compared to 12.5 seconds (26% slower) using MNP5 error control. Most modems use LAPM error control by default.
With error control disabled load times for the text test page jumped from around 2-3 seconds to around 40-80 seconds, and CRC errors were also recorded in the PPPLOG.TXT file.

Line quality

The distance you are from the telephone exchange will affect your connection rate. One modem manufacturer recommends that you are less than 4 miles from your telephone company's equipment to achieve speeds above 33,600bps. If you hear noise, crackling or a hum on ordinary voice calls your connections won't be as good. Low throughputs, though, do not necessarily mean a dodgy telephone line. It is possible, for example that a problem exists between different telephone exchanges or in point-of-presence (POP) modems or data links to the Internet Service Provider.

Many telephone companies now use multiplexing equipment, so-called "Pair Gain Devices", in order to increase the number of customers they can connect to their existing cables. From the telephone company's point of view, they are ideal for customers who install a second line, except, it seems that they typically limit line speeds, often to around 26,400bps. Private telephone switching systems, such as those found in offices and hotels, can also limit the maximum line speed.

Processing power

Since HTML pages contain formatting information mixed in with plain text, they require processing to turn them into the formatted pages that appear on your browser. A faster processor and more memory will speed up this processing.

Software

Some browsers are faster than others. Also, the test returned a value of around two seconds using Windows 95, Internet Explorer 3.0 (32-bit), a 33.6 Kbps modem and setting the DTE rate to 115.2 Kbps. Changing to a 16-bit version of Netscape Navigator returned 5 seconds. Tests at lower DTE and line rates, though, produced similar times.

Internet activity

If the 'Internet' is busy the test page will not load as quickly. Try the tests at 'off-peak' times such as late at night or early on the morning. Repeating the test five or ten times at intervals should give you an idea of how busy the Internet is. If your fastest time still comes nowhere near the time in the table above you may have a problem.

Port speed and line speed

Probably the most confusing aspect of modems is the difference between port speed (Data Terminal Equipment speed) and line speed (Data Communication Equipment speed).

Port speed
Determining the Port speed
Changing the Port speed
Line speed
Changing the Line speed

Port speed

Port speed is the speed between your computer's serial port and your modem. The port speed can usually be set to 110, 300, 1200, 2400, 4800, 9600, 19200, 38400, 57600, 115200 and recently 230400 and above.

Line speed is the speed between your modem and the modem at the other end of the telephone line. Depending on the modem, the line speed can be 2400, 4800, 7200, 9600, 12000, 14400, 16800, 19200, 21400, 24000, 28800, 31200 and 33600 bps. Recent modems can also have line speeds greater than 33600 bps.

Unfortunately, many modems will report the port speed to your communications software by default. This means that, for example, Dial-Up Networking shows a message such as "Connected at 115,200 bps" when you connect to the Internet. If the modem is first told (by means of an AT command like W1 or W2 for many modems) to report the line speed, the message in the above example might read "Connected at 28800bps".

INF files are special device or application configuration files. Windows™ uses parameters contained in a modem's INF file, to configure the modem. One parameter would control what form the modem's connection report string takes, and in turn how the string is interpreted. If the connection speed response cannot be interpreted, for example, when the modem connects with a line speed not included in the INF file, Dial-Up Networking reports the port speed instead.
It is important, therefore, to use the correct INF file for a particular modem.

Determining the Port speed

The port speed setting in the Windows 95 Control Panel (Control Panel | Modems | Properties) is used by applications such as HyperTerminal.
To determine your port speed under Windows 95's Dial-Up Networking:

  1. Click the My Computer icon on the Desktop.
  2. Double-click the Dial-Up Networking icon.
  3. Right click the icon for your Internet connection and then choose properties.
  4. In the General tab click Configure.
  5. The Maximum speed listbox indicates the current port speed setting.

Note: The port speed is recorded in the modem log file.

Changing the Port speed

To change your port speed under Windows 95's Dial-Up Networking:

  1. Click the My Computer icon on the Desktop.
  2. Click the Dial-Up Networking icon.
  3. Right click the icon for your Internet connection and then choose properties.
  4. In the General tab click Configure.
  5. Change the speed in the Maximum speed listbox.

Figure 1 shows a Windows 95 modem properties window. Port speeds from 110 to 115200 can be selected. Typically the port speed should be set at between 2 to 4 times the maximum line speed, so for a 28,800 or 33,600 bps modem the port speed could be set to 115,200 bps. Older PCs may have trouble working at this speed so setting a lower port speed may be necessary. Selecting a port speed of 19,200 bps, however, would be a backward step if the modems have a line speed of 21,400 bps or more.

Windows 95 Modem Properties window - 12Kbytes
Figure 1

Line speed

In general, the line speed is determined automatically for each connection by the two modems. If line conditions allow, the modems connect at the highest compatible speed. If one modem is capable of 33,600bps while the other is only capable of 14,400bps they can only connect at 14,400bps at best.

Changing the Line speed

Modems usually have an AT command enabling the maximum line speed to be set. If slight variations in line conditions cause the modems to retrain (re-negotiate the connection) the modems can be forced to connect at a lower speed using an AT command in the 'extra settings' box. An example is shown in figure 2.

Windows 95 Advanced Connection Settings window - 10Kbytes
Figure 2


Test Page | Test Information | Leslie Long's Homepage | Email
© Copyright Leslie Long, 1997-2001