SuperCharged EVDO - Free Performance Tip

slyfly · EVDO Fledgling Joined: 31 Jul 2006 Posts: 16

What is the MAXIMUM TCP Window Size one can set?

Now, I have it set to 256960 and it seems to give me greatest speed so far. I got this setting from TCPOptimizer optimal setting recommendation.

jschurawlow · Posted: Thu Sep 07, 2006 2:03 am Post subject:

TZ Connection Booster seems to have some decent optimal settings. After using that it did speedup overall benchmarking. Whether or not it remains at those speeds always and remains stable I'll be keeping my fingers crossed Smile

rtpitts · EVDO Newbie Joined: 30 Sep 2006 Posts: 4

Interesting items here.. can you set the TCP on a PPC 6700 or only on the laptop that you are connected to?

In my kitchen last night I checked speed of using my PPC6700 as a wireless modem attached to my laptop..

I have an old rom version.. like 1.06 and nothing tweaked on my laptop

Download reported to be 700K upload at 100K not bad when I see that expected downloads with Rev A are in the area of 800K

Any thoughts?
_________________
RTPITTS

love2lounge2 · Posted: Thu Oct 26, 2006 10:02 am Post subject: Which is the best TCP optomizer?

Any opinions on which is the best and user friendly TCP optomizer to use. Question

My current speeds are:

Speed Test Result
dslreports.com speed test 2006-10-26 12:18:04 EST:
871 / 65 (Kbps)
(106.3 / 7.9 KB/sec)
Compared to the average of 147 tests from spcsdns.net:
* download is 55% better, upload is 12% worse

noobiedoobiedoo · Posted: Sat Oct 28, 2006 11:11 pm Post subject: SpeedTests inconsistant?

Am i the only one that notices the SpeedTests at dlsrerports differs drastically each time you run the test? I didn't try tweaking anything, just ran the test a few times in a row. Reception was in the -70s, anywhere from (1xEVDO): -71dBm to -74dBm, but my results were way off! I maxed out at 900+ DL and 100+ UP but get as low as 130DL and 38 UL!

Does any one else not notice this or does no one try multiple tests in a row? Perhaps I should try a different speedtest site?

s620 / Sprint / Los Angeles

thanks...

love2lounge2 · Posted: Sat Oct 28, 2006 11:34 pm Post subject: Re: SpeedTests inconsistant?

jackrodgers · EVDO Addict Joined: 23 Mar 2006 Posts: 1131

Using SpeedTest.net on the hotel wifi I got

Ping: 94ms
Download: 1066 kbps
Upload: 809 kbps
Server: Orlando, FL

Connecting my Treo 700p as a USB modem ()
Ping: 365 ms
Download: 452 kbps
Upload: 101 kbps
Server: Orlando, FL

I have had 1010 downloads in other locations with the 700p.

Download the USB Modem software The page will give you full details.

noobiedoobiedoo · Posted: Sun Oct 29, 2006 10:42 am Post subject: Re: SpeedTests inconsistant?

noobiedoobiedoo · Posted: Sun Oct 29, 2006 10:46 am Post subject: Re: SpeedTest innaccurate

Way better!! Thank again, love2lounge2.

love2lounge2 · Posted: Sun Oct 29, 2006 11:07 am Post subject: Re: SpeedTest innaccurate

reasonpolice · EVDO Fledgling Joined: 14 Jun 2006 Posts: 11

love2lounge2 · Posted: Mon Oct 30, 2006 9:25 am Post subject:

rojoyinc · EVDO Fledgling Joined: 24 Oct 2006 Posts: 13

I'm new to EVDO but not to computer and data transmission. I've read over this entire thread and I'm skeptical. First, I tried it on my machines and it didn't make any difference. Then it occurred to me "DOH" I bet they're doing this if they plug the aircard into the laptop directly... that MAY make a difference.

I'm using the card in a Kyocera router, Then several machines are sharing that connection. Making the TCP changes made really minuscule differences here... and that's logical. Someone said something about a funnel a few messages back. And that this Dr TCP changes the funnel size. Actually shouldn't be true - though the scoop of data may be bigger with a larger TCP send/receive window - the actual "line/connection" with the aircard and router are the funnel, not the computer. Even a slow computer will be faster than 1000mbps EVDO. Changing the size of the window or data scoop in the computers TCP really should make very little difference. The only performance increase of "less wasted talk between computer and card or computer and router. I made the change and all I could see is that the data send/received during a upload or download was a larger amount of data. Since the scoop of data is larger, there is less ACK/NAK chatter between host and client. Less chatter between scoops means more work done and less chatter between the two sources. So more work done with less chatter using a larger scoop for the data will show some performance increase. But really MINUSCULE!

In some instances - making the scoop of data being transferred will SLOW the performance... for example if the system has a poor connection and needs to resend data packets (I'm calling scoops here) because if noise, static, interference... etc. Is the line/connection is poor, then more packets of data will have to be RESENT - and since the packet sizes are larger, this will slow performance. (as some say they've had happen here). I suspect they have poor connections, low bars or other sorts of interference.

On the other hand - those that can benifet from a larger packet (those with clean transmissions will see a tiny bit of performance increase). Those here that are claiming massive improvments are just mistaken and are getting better results due to other reasons. (connection, signal strength etc. I wrote xmodem and smodem transfer protocol programs back in the 80's for CP/80 machines and going to larger packets was a bit of a performance enhancement, but not if their were errors from noisey phone lines. If several packets went through without errors I programmed in that it would step up to a larger packet (which is what this setting changes) and it would increase performance a bit. If the packet was then to big and the checksum was incorrect the packet had to be sent again and after a resend took place, the transfer protocol would fall back to the smaller packet (smaller scoop of data) size which made the host and client talk more often between smaller packets.

Even still - as I sat and through about it. Even with the aircard in the LAPTOP - you're only changing your scoop size in the laptop to the card. The card still needs to exchange data packets with the cell tower and using a larger scoop on your laptop isn't going to increase bandwidth to and from the cell tower. A poor connection there - means more noise, static, drop outs, more errors for each packet of data sent. This is why I experienced SLOW data rates with my setup downstairs getting only 3 bars of signal, yet my signal upstairs was 7 bars and my data rate increased about 8-10 times what it was at 3 bars. More signal means less interference and less interference means fewer -retransmissions of the same data packets- So it's only logical that a better signal will increase speed, where as poor signal means more errors and as a result more re-sends of the same data packets. (SLOWER bandwidth).

sorry for the ramblings - just thinking out loud, but honestly - changing this setting isn't going to do much and only for those that have a excellent connection already. just my 2 cents.
_________________
PortraNET.com

n6gn · Posted: Tue Oct 31, 2006 5:01 pm Post subject:

litwav · EVDO Newbie Joined: 02 Nov 2006 Posts: 1 Location: Florida

Bought a omni directional antenna from booster antenna and my signal pick up to 4 bars from 2 (41%-78%).
Only getting 1RX CDMA reception, but the end of this month will have EVDO
from Alltel. Upgrading the towers as we speak, shoud be online by end of month according to Alltel. I was getting peak download speeds of only 70
now gettin max at times of 144.
Will pernamently set antenna up on roof this week end, should get even better reception. Will check back for a update next week. KPC 650
is the card I'm using by the way.
Went with EVDO because only have dial-up in my area right now. kind of live in the woods ya know. I have 2 towers about 10 miles away from me, 1 NW
and the other SE of my home. With the antenna and set up using my router
to network my home to 3 puters. Hope to get the EVDO router next month
but until then using my laptop ICS for the other puters to get online.
I have to use the hardwire connection ICS because can't get it to work on my wireless 802. thru the router (belkin). Hey its better than dial-up anyway just wanted to let it be known.