Hi all,

I'm now living in Ft. Worth, Texas after a big move from Littleton, Colorado (Possibly more on that in the real life forum; It was quite a nightmare). Anyways, tonight I was messing around with my new fiber option internet service through Verizon (dubbed FiOS) and found some interesting results.

My package is the 20mbit down / 5mbit up (20,000 kbit down / 5,000 kbit up) plan. I've been doing various tests to see If I'm actually getting that full speed. First, the internet speed tests are giving me very sporadic results. I've gotten everything from 20 down 2.3 up to 7 down and 4 up. The only test that seems to give me accurate results is the PC Pitstop flash bandwidth test. So, I use that test along with downloading a file from nZone.com and uploading a large file to Filefront.com to test. I figure actual download rates are more important than some test anyway.

I've been hearing that XP may require "tweaks" to the TCP/IP and HTTP stack to get full speeds if using a fast service. I also heard that Vista has a more dynamic stack that doesn't require it. So I tested in Vista first (To make sure the OS was not limiting my speeds). Here's the results:

PC Pitstop Flash Test (Dallas server):
IE7 - 20.7 Mbps down / 4.58 Mbps up
Firefox 3.0 - 20.8 Mbps down / 2.33 Mbps up

Download Tests:
Crysis Demo (1.77GB) from nZone.com using IE7: About 18,320 kbps download rate
Crysis Demo (1.77GB) from nZone.com using DownThemAll! Accelerator (5 chunks): about 19,840 kbps average download rate
Crysis Demo (1.77GB) from nZone.com using Firefox 3.0: 4,808 kbps download rate

Upload Tests:
38.8MB file uploaded to FileFront using IE7: 4,200 kbps
38.8MB file uploaded to FileFront using Firefox 3.0: 2754 kbps

I was quite surprised by the results. It turns out that I indeed was getting nearly advertised speeds while using IE7 or a download accelerator, but it appears while using Firefox I only get half my upload, and a quarter download. What's even more interesting is the PC Pitstop flash test reports my upload at 4,600 kbps upload in IE7 but only 2,300 in Firefox! So it affects the flash based test, AND actual rates. I ran the same tests in XP and got almost identical results; the only thing that was different was my download seemed to peak at 16 KB/s instead of 20 KB/s like in Vista.

Now my question is this: Is there a way to tweak Firefox to correct this problem? Using FasterFox seemed to have no effect; I have pipelining and 8 max connections enabled in Firefox's About:Config already. Furthermore, even though the speeds in XP aren't impacted as much as I thought, they are still a little bit less than Vista. Programs like the PC Pitstop Optimizer 2.0 analyze and show you settings in the TCP/IP stack, and Firefox settings to change to improve speeds. Should I do this or leave it at default?

- Thanks, Cornflake

P.S. I chose nZone for my download tests because it supplies great download rates, and allows download accelerators, so my download isn't limited by the server sending it to me (Its hard to find many sites that can deliver 20mbits when downloading something).

Those results are surprising. I actually use IE7 for most of my downloads. I'm usually downloading installer packages that I don't want to keep so IE7 is setup to download to a temp folder. Firefox is setup to save the downloads so I use that for things I want to keep. What does this mean? Nothing, but I'd like to see what the solution is, if there is one.

Mark

Yeah I was pretty surprised. I thought the OS was affecting my speeds but it was mainly Firefox -- I guess IE is good for something after all . Now what I'd like to do, is find a way to test my speeds without using any browser. Do you guys know of a way that I can initiate a file transfer through, say, a command prompt?

Also, here's a couple of screenshots showing what the PC Pitstop optimizer suggests changing to optimize my speeds. Do you guys think changing this would actually help at all?

if you have a linux system, you could use wget to download a pretty big file, as long as it has an http address you can use with it. You could also use bit torrent to download something (again, a big file would be better). With bit torrent you would want to find something that has a lot of seeders and other peers. An x86 installation image for linux would be a good thing to try it out on as they tend to have a lot of people seeding and also downloading.

you can use putty if you dont have linux and use wget also, again use a big file with http address.

imnuts wrote:

if you have a linux system, you could use wget to download a pretty big file, as long as it has an http address you can use with it. You could also use bit torrent to download something (again, a big file would be better). With bit torrent you would want to find something that has a lot of seeders and other peers. An x86 installation image for linux would be a good thing to try it out on as they tend to have a lot of people seeding and also downloading.

Ah, good suggestions I used uTorrent to download the most seeded Ubuntu distro and was able to achieve about 2.4 MB/s download which is nearly 100% of my rated download I also used d*mn Small Linux (DSL) on my laptop and used Wget to download a large file. I got over 1.5 MB/s on one file, and nearly 2.0 on another. I'm guessing I was hitting the file server's max speed because i was just making one connection.

In the meantime, I did some more research. It turns out a lot of other people have run into problems with Firefox and their upload speed... See these threads here:

Large Upload Speed Difference between Firefox and IE7: EVDOforums.com

FireFox: Slow Upload Speed Fixed! - 3dGameMan: Kickass Forum

In these threads the users are showing the same symptoms - particularly, their upload being reduced by about half when using Firefox (In the first thread above, the user had less than 1 meg upload yet his speed was still being halved). The second thread does mention a fix... by changing the "DefaultSendWindow" to 78000 with a registry entry. What does the SendWindow setting do anyway? (It's not a TCP/IP setting apparently). Why would it only affect one browser? How come it didn't show in the PC Pitstop optimize program?

Strange indeed!