Jump to content

Greg Hazel

Established Members
  • Posts

    725
  • Joined

  • Last visited

Everything posted by Greg Hazel

  1. As I said here: http://forum.utorrent.com/viewtopic.php?pid=388163#p388163 these issues are probably related to the NIC driver. uTorrent should not be able to cause a bluescreen by using IPv6. Please contact your NIC manufacturer or Microsoft and let them know about the bug.
  2. The limit should be obeyed properly. Why do you think I is not?
  3. If you stop the torrents, does browsing work again? Where did you set the number of half-open connections to 50? In uT or Windows?
  4. Please install the Debugging Tools for Windows, and get uTorrent to crash again: http://www.microsoft.com/whdc/devtools/debugging/installx86.Mspx Submit the resulting dump (it should be > 60kB)
  5. Please run the logging build, 13911, for about 5 minutes. Exit uTorrent than post the log files here.
  6. uTP is enabled by default - no need to configure anything.
  7. You just mean that you have to set the limit slightly lower with uTP, right? When set properly uTP behaves just as well as TCP, right?
  8. It's odd. TCP does not seem to affect ping times the same way - but you said video playback in the TCP unthrottled case was worse than uTP unthrottled, did you not?
  9. Also, you don't show it in the uTP video, but I assume upload is unlimited there as well?
  10. In both of those videos NetLimiter says "Limiter enabled". What limits does it have?
  11. Worse ping times with uTP? At the same rate as TCP? (externally measured) That seems unlikely.
  12. Can you capture a wireshark log while it's happening? Also, next time it happens, please crash uT like so: In Vista, use Task Manager, right click on the process. In XP: ntsd.exe -p 4100 -e 4100 -c ".dump /mf c:\jit.dmp;q" where 4100 is the PID, and c:\jit.dmp is the resulting dump.
  13. This is probably due to a bug that was fixed for the next version. It happens on routers that use the LAN source address when they hairpin packets.
  14. But it does, if you set the limit, right? You said: So 2150 for TCP and 2110 for uTP? That seems pretty high to me.
  15. Ok, so it sounds like uTP is better, unthrottled. Not perfect, but better. Right?
  16. How is this different from the log you pasted previously? Wasn't that uTP unthrottled too? If you use TCP unthrottled, do you see errors in video playback as well, similar to the errors you see with uTP unthrottled? The logging mechanism does not capture statistics for TCP, so you're going to have to try to play video externally and see if it is disrupted.
  17. Can you give me the results for the "TCP, unthrottled" test? No, this is not caused by uTP, this is due to net.calc_overhead.
  18. Thanks. I'll send you a test build.
  19. Well, it doesn't make much sense that TCP only would upload slower on 1.9 than on 1.8.1. Nothing really changed there. Also, please try with uTP only using the logging build (13911) and post the logs here.
  20. After you set it to TCP only you need to stop all torrents or restart the client, otherwise existing uTP connections will stay open.
  21. It evaporated. You could buy me a new one. I'll put up a donation page.
  22. Next time it happens, please crash uT like so: In Vista, use Task Manager, right click on the process. In XP: ntsd.exe -p 4100 -e 4100 -c ".dump /mf c:\jit.dmp;q" where 4100 is the PID, and c:\jit.dmp is the resulting dump.
  23. Do you have another uTorrent client running on your local network?
  24. So to be clear, here are the tests you have run: TCP, throttled (1300kB/s limit ?) uTP, unthrottled I could also use results for: TCP, unthrottled uTP, throttled (1300kB/s limit ?) The current theory is that your router does not have a significant buffer - this would explain the inability to affect ping times while uploading unthrottled. So instead of delaying a packet, it would simply drop it. Packets that make it through have no extra significant delay. In this case I believe uTP will perform no worse that TCP, but basically the same. Your tests will prove that one way or the other.
×
×
  • Create New...