[Daisuke] Posted January 27, 2008 Report Share Posted January 27, 2008 Hi all,Basic spec:Athlon XP 2600+2GB ramASUS motherboard (can't remember specs exactly atm, but it is pretty old)Windows XP Pro SP2I have been using uTorrent for a while, and last year it was all fine and dandy. I am a careful person when it comes to updating my windows etc. but this last update happened WITHOUT my consent in any form. I had turned off automatic updates completely, only allowing it to notify me when it happened.I have also tried the things followed in this thread: http://forum.utorrent.com/viewtopic.php?id=15992 which frankly dropped my speeds to 1.5kbps max and left them there.Ports are forwarded correctly and I have tried well seeded torrents (open office) and got no better speeds. I have been looking about, and as I used to use bitcomet I checked their forums and found this:http://forums.bitcomet.com/index.php?showtopic=12791340&st=0&p=41278entry41278The reply that I am presenting in particular is post #5. This basically states that the latest windows update contains a revision for the TCP/IP stack, and he recommends not to patch it as it is not worth it. I also found this:http://filesharefreak.com/2008/01/09/the-winxp-tcpipsys-3244-update/Wherein the author notes a change in the version number of the TCPIP.sys file, which is 5.1.2600.3244 <-- the version that I am now using.The microsoft bulletin for the two updates I have found in my Add/Remove programs panel to have been installed in 2008 are as follows:http://www.microsoft.com/technet/security/bulletin/ms08-001.mspxhttp://www.microsoft.com/technet/security/bulletin/ms08-002.mspxThe first link takes you to a fix titled Vulnerabilities in Windows TCP/IP Could Allow Remote Code Execution (941644) and it deems it a critical update for XP SP2.Having followed advice on what the half open limit should be (I've heard people say higher is better, and people say it shouldn't be higher than 10) I cannot find a value that will allow good speeds. Due to the timing of my speeds drop, I am certain that it is to do with the windows update kb941644. As such I am going to performa s sytem restore to before this update happened, and stop it from updating again (even though I didn't allow it to before and it did it anyway) and see if that helps my speed. If it does, I shall recommend avoiding the windows update.This thread is half looking for people's thoughts on the matter, and to confirm whether I am the only one affected by this update.Thanks,Dai Link to comment Share on other sites More sharing options...
jewelisheaven Posted January 27, 2008 Report Share Posted January 27, 2008 Halfopen does not affect speeds. It does control how FAST you connect to aggregate peers. The standard limiting problem still apply when connected to peers. You may be shaped. They may be shaped. You are not the only person experiencing trouble after the update. Searching on the forum post-"Patch Tuesday" will verify that. If you are running default settings for those advanced options (* means non-default) the problem is external. If you are noticing a discrepancy in the speedtest results and your experienced speeds be sure you're not confusing Kbps (speedtest) with KiBps (uT and Windows metric for speed). A 512 Kbps is equal to 64 KiBps of which you can expect 80-90% utilization sustained. If you'd like to try 1.8 from the Announcements forum self-encapsulated so it doesn't muck with your current settings you may find increased speeds due to workarounds for some current traffic shaping implementations. Link to comment Share on other sites More sharing options...
[Daisuke] Posted January 27, 2008 Author Report Share Posted January 27, 2008 Thanks, jewelisheaven, I shall try the 1.8. I am not confusing any speed terms in my mind, just the terms I write. As far as the lists have shown, traffic shaping is not done by my ISP, but mybe that has changed since the list was last updated. Link to comment Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.