F u r u y á Posted February 21, 2008 Report Share Posted February 21, 2008 First of all, I'm not an english native speaker, so please forgive me when I make english mistakes. Second of all, I thought that, instead of keeping those tests only to me, it might be useful if I post it publicly.I was writing some kind of "narrative" but I thought that many readers would find it boring and maybe would stop reading it (not that you'll read it all the way ). So I chose to put the data in the "dryiest" way I could. So here we go.From old tests in the past I knew that BitComet was able to mantain consistents speeds, while in uTorrent the speeds had some fluctuating behavior. Even so, I'm a uTorrent user, for many other reasons.Test methods* The test was done in a 40 minutes window.* The torrent used had about 1000 seeds and 300 peers.* That torrent was the only task running in the tested clientsIn the picture below and from now on, I'll use the following abbreviations* mc: Max connections per torrent* mg: Max global connections* ho: Half-open connections* u : Max upload speedPlease click in the images to open it in its full resolution. In this test, what matters is just the DU Meter window (bottom of the screen). I put the whole screen just for you to see the time and uTorrent window. The DU Meter window shows the instantaneous download(red)/upload(yellow) speed each second, and the graphic "walks" from the right to left; so in the very right of DU Meter window you'll see the current speeds.4h32 AM* I marked with green rectangles the speeds region for different clients.* Beneath the retangle, you find the used client and the configurations.* Speed regions marked with yellow ballons means that two files which was set to high priority was still incomplete. (I chose a .nfo file and a sample file, for obvious reasons). * Also, note that I've set high priority just for uTorrent (not intenionaly, the ideal was making a test with all files with normal priority or setting that high priority in BitComet too).From the speed chart I concluded that setting some files to high priority can choke up your download speed until those files are completely downloaded. See in the second uTorrent retangle (mc70 mg80) that, once the download of these two files has finished, the speed started slowly increasing.As for the difference between the clients, we can cleary see that BitComet manages to get in the max speed really faster than uTorrent, which took a long time to reach the max speed (even discounting the time wasted for the high priority files).4h39 AMuTorrent was able to mantain the download at full speed. Right in this time I changed the configurations to: mc90 mg100. (In the past I made some tests and found that download speed was *very* sensitive to max connection per torrent parameter. If you want, see those tests in the end of this message).5h12 AMEven with mc90 and mg100, uTorrent kept the download at full speed.(ps: it's so nice to see the DU Meter window just filled with a red rectangle :cool:)(ps: I know that mc90 mg100 is the "theoretical" best settings for my settings, but in practice I see that lower values gets me more consistent speeds)Now the question is* Why and how BitComet can get the torrent to constant full speed so quickly? Does it cheats? (that's why it's banned in many private trackers?)* There is a way for me to configurate uTorrent to NOT make the speed fluctuates?Consistency is all! I preffer a constant and solid 90KB/s download than a 30-200KB/s oscillating download.To me, uTorrent can only achieve a constant full speed when the torrent has a very nice swarm, as the one used in this test. In some private trackers, with about 300/20 (Seed/Leech) I can get full speed with uTorrent too.But, with a swarm worse than 300/20, which is the most common swarms I find, BitComet shows its potencial for achieving and mantaining the max download speed. When I'm in a hurry to download the file and when the tracker is public, I use BitComet. Once the download if finished, I go back to uTorrent to seed it.:: ----------------------------- ::The old tests I mentioned:20/02/2006 to 21/02/2006All measures has been made with BitComet with a single torrent job. The other configurations I didn't mentioned was the same in all measures (the only thing that has changed was mc).20/02 - 21h00 mc7521/02 - 02h00 mc7521/02 - 03h00 mc7521/02 - 04h30 mc7521/02 - 05h15 mc7521/02 - 06h30 mc5021/02 - 17h55 mc5021/02 - 18h30 mc5021/02 - 19h00 mc5021/02 - 19h30 mc5021/02 - 20h00 mc5021/02 - 20h30 mc5021/02 - 21h13 mc50 (end of the download)Here, 19 to 22h are "peak time", so in these times it's normal that decreasing in the speed consistency. edit 1: Just see the difference between mc50 and 75! With mc50, the speed gets much more a constant behavior.28/12/2007 to 29/12/2007Five measures has been made with uTorrent; one with BitComet. With both clients, the same torrent was the only job running.:: uTorrent ::28/12 - 20h00 mc7028/12 - 21h00 mc7028/12 - 22h00 mc7028/12 - 23h50 mc70 (the download speed peak was a concomitant little download I've made from a website)29/12 - 00h30 mc70:: BitComet ::29/12 - 10h50 mc90Commentary: even with a higher max connections value, BitComet could achieve constant full speed! What I know from my tests is that the more I increase mc the more the download speed tends to be unstable with the undesired "rollercoaster" effect. So with lower values of mc I think uTorrent should mantain constants speeds, which haven't happened.Final words: plz don't think I'm a BitComet fan!!! As I stated before, I'm user of uTorrent!! I'm just sharing the results of my tests!Thanks for your attention! Link to comment Share on other sites More sharing options...
DreadWingKnight Posted February 21, 2008 Report Share Posted February 21, 2008 Consistencies between individual downloads can't be guaranteed simply because of how random your peer lists are. On swarms larger than your total per torrent connection limit, you never connect to the same peers twice in a row. Link to comment Share on other sites More sharing options...
Switeck Posted February 23, 2008 Report Share Posted February 23, 2008 How many upload slots did you allow per torrent?How fast were you uploading on the torrent?Also how fast uploading in total?Was uTorrent reaching and sustaining the upload speed limit you set it to? Link to comment Share on other sites More sharing options...
F u r u y á Posted February 25, 2008 Author Report Share Posted February 25, 2008 Hi Switeck> How many upload slots did you allow per torrent?3 in BitComet; 2 in uTorrent.> How fast were you uploading on the torrent?> Also how fast uploading in total?The torrent was the only task running (the total upload speed was equal to torrent upload speed). Because the Seed/Leech ratio was so good, the torrent was barely uploading, uTorrent says that the average was 5.9KB/s (but, during the download, it was lower than that, because after the download has finished, the upload speed began to increase).> Was uTorrent reaching and sustaining the upload speed limit you set it to?As I said in the last answer, the upload speed wasn't at its limit (seed/leech ratio too high), and its max was set to 10KB/s (this is the best limit here, because this way the torrent is able to sustain the upload speed at 10KB/s and the download don't get harmed by the upload). My true max upload speed is about 14.4KB/s, see here, so I think that is the max safe upload speed I must set. Link to comment Share on other sites More sharing options...
Switeck Posted February 25, 2008 Report Share Posted February 25, 2008 For what it's worth, you likely won't get consistent speeds on very overseeded or underseeded torrents. Link to comment Share on other sites More sharing options...
AllGamer Posted February 25, 2008 Report Share Posted February 25, 2008 the best consistent test is OpenOfficenot sure if you guys mentioned it already or not Link to comment Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.