Neronut

Could this be because that is the max number that you connection can handle (maybe the ISP plays a part, or the bandwidth to keep them connected)?

THere is a CPU usage fix? Is it possible I could get it from someone? THanks in advance.

Two items: 1) I have noticed something similar to rafi's peak scenario, although it seems to happen as the torrent is contunally running. WHat it does is it will increase in speed and then all of a sudden drop back to where it was, and it does this continually. After a few minutes or so it evens out and sticks to the slower speed. 1.1) Sometimes it does exactly what rafi is describing. 2) I have noticed that I have to keep the upload slower then what I want it to be at otherwise it transfers too fast and screws over the download. Its like the original case, where the overhead was causing issues, but not as harsh. I have net.calc_overhead set to true and it improves it but there is a difference. For example the upload rate of 27.7 kB/sec in μTorrent is reading as 31.6 from a throughput meter. μTorrent also never hits my throttle amount of 30 kB/sec, 28 is as close as it gets. Scott

Question: When the final 1.9 is released will we need to have net.calc_overhead on for uTP to not to use more then the throttled limit?

I'm still having this wacky issue with the download measurement with build 13885 - its measuring the downloaded data in GB instead of MB.

I have just noticed there is a new clock icon in the RSS feed viewer. What does this mean? The normal RSS icon is also there below it. Edit: I was using Build 13827 and noticed that the download meter was nonsensical

I have uploaded some Wireshark data along with a capture of my speeds in both uTorrent and in a Network monitor. Link: http://www.mediafire.com/?h3oi2502izi Hopefully this is what is needed. Edit: Also, does bt.transp_disposition have any other options that are not listed, I accidently set it to 2 and although the overhead is there it seems to be faster. Maybe just a coincidence.

What type of information am I looking for from Wireshark? Like what should I filter off?

My download speed was going around 10kB/sec where as the throughput was around 30-50kB/sec. The upload was almost double, 20kB/sec registered in the application was ~40-45kB/sec throughput. Since my upload is max 52kB/sec on good days it was most likely the upload slowing down the download which was slower then the actual download rates in the application. I can do a more in depth analysis if needed.

One thing I have also noticed about 1.9 is that it will have more IPv6 peers then 1.8.1. Although I only have an IPv6 network/Internet connection it used Teredo tunneling a lot better than 1.8.1. My download speeds were much slower, however, then compared to using 1.8.1, again the extra overhead. All IPv6 peers were using 1.8.1 and in uPT mode. Also does 1.8.1 make uTP connections on its own or does it have to be initiated by the other side? Will it identify them like 1.9 does with the IP [uTP] as 1.9 does? Edit: Haha, I answered my own question by rereading the initial post XD except how it denotes the connections.

@Troller: I also have tried throttling the upload and download and this has not helped. It seems to be some kind of peer traffic, but even with just one peer connection and no transfer its high. I guess I can try and Wireshark it to see whats up.

Hi, I have been watching the network usage with the current version of 1.9 at the time of this writing and it is high even though nothing is downloading. 1.8.1 sat nicely at idle but 1.9 uses an excess of ~40 kB/sec down and 30 kb/sec up. I have experienced slight slower than normal downloads and normal uploads and no crashing, just this persistent transfer to seemingly nowhere, although it must be to a client(s). I have tried disabling the DHT and that wasn't the maker of the traffic. Just thought I'd get this out there in case anyone else was experiencing similar issues. Scott