TCP/IP Parameters Tweaking Tips


TweakMASTER can substantially improve your download speeds for all types of connections from dial-up to DSL, Cable, and Wireless. The Connection Optimization Wizard will walk you through some recommended settings using several different "optimization strategies". After that you may wish to experiment with some other settings in the Advanced Optimization Settings dialog box to obtain the fastest possible connection. What follows are some tips to help get you started.

Note that no utility will modify the speed at which you connect to your Internet Service Provider (ISP) on a dial-up connection. Whether it is 28000, 33600, 44000, 48000 or even higher, that is negotiated between your modem and those of your ISP automatically, and may vary depending on which modem you reach at your ISP, or other factors such as phone line conditions, and so on. What TweakMASTER will do is maximize the actual data transfer rates you can achieve through the modem speed you connect at. Note that data compression and other factors may allow you to download at speeds which may substantially exceed the basic connect speed.

Comparing download speeds

Perhaps it goes without saying but you cannot necessarily "optimize your Internet connection" merely by running the wizard, changing the settings, then trying one site to see if it's faster or not, and then base your decision as to the effectiveness on that. The TweakMASTER Wizard only sets "recommended settings" as they apply to an "average" set of circumstances, but most users' connections do not conform to those "average circumstances."

To test TweakMASTER fully for the best possible connection, one has to have patience, use Connection Throughput Measurement, try numerous different sites, record average download speeds (using our utility DU Meter is one good way), change settings (sometimes requires a reboot to take effect) and compare results. The DU Meter stopwatch (available from DU Meter's right click menu) is a handy tool to time downloads and compare loading times and average data transfer rates.

Users of Connection Throughput Measurement should realize, however, that it measures throughput to a short list of servers at one particular point in time. More thorough experimentation with downloads from a half dozen or more different sites, at different times of day, for example, is often required to determine the true optimal settings that will provide the fastest download speeds for the average conditions you will encounter when using the Internet. You may wish to experiment with file downloads using both HTTP or FTP, as well as comparing loading times for various websites, particularly those with numerous graphics which take a while to load normally.

Preparation for testing

It should also be noted that if you are downloading web pages for comparison, you need to clear your browser's cache each time before running a test. With Internet Explorer this is easily done by loading any page other than one you plan to use for testing, and then going to "Internet Options" under the Tools menu and on the General tab, under "Temporary Internet Files", press the "Delete Files" button. This will ensure that you will actually be downloading all the files from the web page, for a true comparison of page loading times. For other web browser software, refer to the help file included with that browser.

Using the Wizard

Although the TweakMASTER Wizard will set parameters to their "recommended" values automatically, please note that no software can automatically determine which settings will work best for your specific configuration of computer software, hardware, telephone lines and Internet Service Provider (ISP). These are the recommended settings only, based on average conditions and system configurations. Experimentation is the only way to determine the real optimal settings that will provide the fastest average throughput for your Internet connection. While other utilities may "claim" to do it all automatically, in truth this can only be a guess at best, based on average circumstances.

Aside from trying wizard settings and general experiementing in the Advanced Optimizations settings, The most important settings to tweak are:

1. TCP receive window size (pre-Vista).

2. TCP receive window autotuning level (Vista and later).

This will have most drastic effects on the throughput, especially on older computers and on high-latency connections.

3. On the browsers tab, setting "maximum simultaneous connections" to 25 (faster Internet connection) or to 10 (slower one, like dialup) will improve Internet browsing, even though the throughput will not improve on per-TCP connection base.

A good utility is one that makes this clear and allows you maximum opportunity to easily and conveniently experiment with these settings to find what works best for your own specific set of circumstances. Remember there is no shortcut to rebooting your system. You must restart Windows XP each time you change most networking-related settings in the Windows registry in order for them to take effect. On Windows Vista and later in many cases reboot is not needed. You will be prompted to reboot if any of the changes require a system reboot. You can say "no" to the prompt to reboot your computer in order to wait for a more convenient time, but if required, then the changes made will not be in effect until you have rebooted. This is a Microsoft Windows design requirement and cannot be circumvented by any software, regardless of various deceptive marketing claims to the contrary. If there were any way to do this, rest assured TweakMASTER would do it.

Dial-Up (modem) users

For a dial-up connection, try the Optimization Wizard's "Optimization Strategy #1" first. If that doesn't seem to help much, try one of the other strategies. You can also try manually changing TCP receive window to 4x MSS, 8x or higher, or leave TCP receive window at default and try that, and see what happens. You can also try different values for MTU as well: 556, 1006, 1024, 1064, 1152 and 1492 are all values that other people have reported better luck with, and try a corresponding TCP receive window value or two.

Broadband (Cable, ADSL, etc.) users

For a broadband connection, the generally recommended practice is to leave MTU at 1500, and to experiment with different TCP receive window settings on Windows XP and different settings for TCP receive window autotuning on Vista. In general, if your connection is "high bandwidth, high latency", set TCP receive window to a higher setting, and if your latency is low, you can try a lower value.

Note: If your ISP requires a user name and password to connect (your ISP may refer to this type of connection as PPPoE or PPTP) then you may find it necessary or beneficial to set your MTU to a lower value than the standard 1500 used for most broadband connections. You should try 1492, 1452, or 1400 and a variety of corresponding TCP Receive Window values. The exact values you choose depends on the nature of your connection. You may want to experiment with all these MTU values from higher to lower and see which one works best for you. 1492 is the most common MTU value used and the one set by TweakMASTER wizard. But you can also manually experiment with different values in Advanced Optimization Settings dialog box.

General tips

With both dial-up and broadband connections, it is always best to run your comparison tests at the same time of day as Internet congestion can vary greatly from one hour to the next.

If one uses different kinds of connections (for example with a laptop), we recommend one set of settings for each type of connection. You can Save the settings for each type of connection using the Backup button. One can also set the MTU for a dial-up adapter and for the network card differently (they are stored in different parts of the registry and have no effect on one another).

You can also restore your original that you had at the time of installation or reset to Windows defaults by using the Restore Defaults button in Advanced Optimization Settings dialog box or by using the Wizard.

Although "optimizing" these settings for one's connection can have a substantial effect in many cases, it is not effective in every case. Careful experimentation using an objective measurement tool like Connection Throughput Measurement or DU Meter may pay off with a combination of faster settings, but in some cases, nothing really seems to help. No software can overcome occasional Internet congestion, slow servers, bogged down Internet routers or overloaded ISPs. But your own computer will no longer be the weakest link in the chain. You can optimize your own system to permit the fastest connectivity the Internet can provide for you. And understand, above all, that no software can do it all for you automatically, despite many misleading claims and deceptive advertising to the contrary. There are simply too many variables to know what if anything will work best in every case. Experimentation is ultimately the only way to be sure which combinations of settings will work best for your own particular situation. Fortunately, TweakMASTER makes this easy to do!

In any case, even if you are one of those who do not seem to benefit from the optimized settings TweakMASTER can make for you, you can instantly restore your original settings by simply using the Original Settings button in the Advanced Optimization Settings dialog box If you truly found those settings worked better for you than various combinations of settings in TweakMASTER, we would very much like to hear from you telling us exactly what those original settings were and what other combinations of settings you tried.

Note that you can still enjoy the benefits of DNS Acceleration or LinkFox, something that does work for everyone. You should also appreciate the other utilities in TweakMASTER, like PC Clock Synchronization, Ping & TraceRoute, WhoIs, or the Network Connections Viewer, and others. Restoring your original settings as described above will only affect the optimization settings and not these other features, which many people have reported are well worth the program cost by themselves.