JANE: Closest city you can get is best: it doesn't really matter, since you're after a "real-world" throughput # anyway! (E.g. In the best case you'd run the speed test TO THE GAME SERVER, since that's the # you really need!)
The setting varies the size & frequency of the packets that you send & receive between your Halo game client and the server. Hence you don't want it too high, since that will result in an overload of your connection's ability. (Hence "set it to the ONE BELOW what you get," you see....)
If your speed varies a lot (can you say "satellite"/"cable?") then play with the setting IN-GAME (bring up menu with <ESC> then "Change Settings" and the rest per the above instructions), varying it between the lowest that you EVER measured and your "usual" and see if you can even tell the diff.
Max: If you only did the speed test once, then do whatever you want, because it may not be an appropriate representation of your real line speed anyway!
If 500 is your MAX over 3+ runs, then yes, set it to 386k. If ANY ONE of your tests got over 512, give 512 a try, and then after you've gotten used to the performance, knock it down to 386 and see if you like it better.
Ediseye/All: I have my doubts about the accuracy of speedtest.net's measurments. It give me #s WAY higher than broadbandreports.com, or any other measuring method that I use.E.g. My #s via both: (Latency(ms) / Down(Kbps) / Up(Kbps) )
BroadbandReports.com: 12 / 8507 / 1704
SpeedTest.net: 12 / 14380 / 4470
I do not get the sort of numbers given by SpeedTest.net unless it's for bittorrents. (I.e. Multi-connection, simultaneous accesses in parallel.) [Note that in either case the ISPs compression may be causing misleading results as well, but that's hard to take out of the equation....]
I think SpeedTest.net might be using some method that maximizes bandwidth by using multiple connections, which is okay for a "browsing"/"downloading" speed number, but
is NOT a true measure of what you'll get in any real-time gaming situation....
All: Play with this setting! Results seem to vary a lot, but in the general case fall in the area that I've suggested.
Your mileage may vary!Also: Be SURE you keep in mind that this setting varies with YOUR PROFILE, so when you switch profiles
(go "undercover") you might want to make the same change there. Also remember to change it when/if your connection changes, too! (You'll WANT TO kill yourself if your on a portable and try to play a LAN game while set on 128k! All the while everyone ELSE will be killing YOU!
)