Kazuaki Shimazaki

HK PCCW Netvigator might also be starting to throttle. Up until the day before yesterday, I could download at least half of full rate (over 300KB/s) from a good torrent. All of a sudden I'm down to <20KB/s class (my DL is averaging maybe 10 right now). Of course, I'm talking torrents with seeds in at least the dozens and the peers in the hundreds and thousands, new torrents. The possibilities are: 1) Coincidence. 2) Some cable broke in the Pacific (again) which slowed things down. 3) PCCW Netvigator is starting to punish me for excessive DLing by throttling me. Yes, of course I've got encryption on (but not Forced and I'm still accepting "Legacy" connections). Can any other users of Netvigator confirm or refute? UPDATE: Well, for awhile, I thought maybe it is some kind of general problem, but I've seen my other connections going at a 300KB/s clip, while my well-seeded torrents are tooling around at 100KB/s with large up and down dips (one of them, anyway, the other is going <10KB/s right now).

I'm sure you are very touched, xpack. However, next time, you might want to be more discreet about exactly what you are downloading

Working real well here. Only nit is that for some reason the anti-standby function still doesn't work for me, so I have to disable the autostandby whenever I'm downloading myself (so my computer doesn't standby when I close the laptop lid). I know these things are particularly tricky, but it'd be nice if one day it worked...

Not everyone has your massively fast hard drive, Firon. I'm sure I told you what happens to my computer at that speed.

I'm estimating it off the Rate readings. If it is reading five times/second off the cache and only one time off the disk, then presumably the other five times it drew from data already in the cache, so it must have guessed correctly and saved my drive heads from having to maneuver. Or am I reading this wrong? In terms of total data read, I may be losing or I may be winning if everyone just wants similar data, but I would have thought the concerns over disk wear would be over the number of times the heads have to maneuver. If they are doing a few big reads rather than many little reads, in principle I should be ahead in terms of disk wear. Or am I getting this thing wrong. For now, I think I'd force it to activate for a bit longer so I can actually get some experimental data... since I'm going to bed, I'd force the read cache up to 32MB (no worries over RAM or resource consumption now...) manually and see how it does tonight. APPEND: I think I'd stop the experiment. Yes, I know I said the above only 10 minutes earlier, but I just realized from reading the Task Manager display that pretty much the entire cache went onto virtual memory (i.e. my hard drive). In fact, I think I'd stop using caches - what's the point when the program reads from my HDD and just copies it over to the VM part of my HD?

I just managed to find out about 458, and wonder why Ludde is so shy about putting it on - it works perfectly. Yeah, I'm sure it doesn't work perfectly for someone, but I don't exactly see a lot of screaming, and really it isn't like the official releases work for everyone anyway (a long line of complaints in Troubleshooting and Found Bugs testify to this). The new disk cache works like a charm, though the AI might need some work. I had to manually disable the write cache and force the read cache to activate. Apparently, the fact I was Pure Seeding didn't impress the AI. The disk cache took just under TWO MB and seems to vary between 60-100% effective with an average of around 80% (preliminary estimate with some tendency to improve over time). I don't know exactly how harmful all the disk reads are, but if I can eliminate 80% of disk action with a measly TWO megs of RAM, I'd pay it. IIRC, BC's cache claims around that efficiency, and it requires at least 6MB, escalating to 50... Good work, Ludde!

I'd disagree. It is actually better to make a new thread. This is a 90 degree tangent off the original. Wrong analogy. In this case, scientist already released his finding and product, and for free. Ultimately, it all comes down to trust. How do you know that the open-source code is exactly the same as what was published? Or what if I publish one version of the code, and then publish an "update", and I delay the updating of the published source code for a while. Of course, eventually I republish a clean version and source code before people catch on, then more dirty versions, and so on... And if you have to carefully check the real opcodes or analyze behavior to be really sure, than what's the difference between open and closed source? But why must he show his hobby to anyone else? Why should he be forced to publish his secret game-winning strategy? "Greatness" in a program is defined by function and efficiency. In both respects, uTorrent is top class. If I make a study of open source vs close source based on BitTorrent clients, open source would look like a direct path to bloat. --------------------------------------------------------------------------------------- Actually, I'd disagree somewhat with the concept that a program's greatness is based on an Appeal to Popularity. I prefer objective measures like efficiency and function. True, but two counters: 1) Only a very small minority in an open source community has the technical expertise to make real changes, so in both cases most of the community can only suggest and pray. 2) Why should Ludde help you destroy his vision? This is an important point. Apparently more than his main competitors, Ludde has a set of principles, which manifest in decisions as to what features to enter. For example, Ludde will never implement a Right-Click-Ban feature, or BitComet's (unmentioned) Rapid Connection Recycling, and of course any leeching features. AFAIK, he wants his client to be "torrent-friendly". One can see that any uT-derived thing, made by more cynical and less idealistic people, will likely incorporate these features. And it'd be popular too. I've heard of at least a few guys say that they would switch to uT and rid themselves of the Azureus Deadweight if not for the lack of Right Click Ban. Ludde's supporters here may argue good reasons for not including any of these, but if a uT variant appeared boasting these as options, it'd look very tempting... Ludde can't stop people from using Azureus and BC. But why should he be forced to make it easier for people to do things that he does not personally like? As far as he's concerned, if you really like Right Click Ban, you can pay by killing your computer with Azureus. That he should be expected to make it easier for people to have Right Click Ban without paying that price must have been shocking to him... No security can ever be perfect. That does not mean one should not try to make it a little tougher on the bad guys. So, is Azureus (open-source program) a loner for not supporting Mainline DHT network?

I can't register on that stupid site for some reason. I'm sure I typed in that stupid card correctly, and they still won't let me register.

Not much for functionality? The beta introduced 2 things: 1) PEx - great for weak torrents and arguably patches the last major uTorrent deficiency vis-a-vis the giants of BC and Azureus. 2) Genuinely functional PE. I do think, however, the icons accounted for like 9K (I swear uT was 141 or so the version before the icons and the version after those icons was >150K). Considering that the two giant features above used only 10K between them, I'd have to agree with you those icons weren't the best investment of valuable space. It was about then I gave up on my anti-bloat position.

I do know that uTorrent's encryption schemes are more advanced and thus will presumably take more CPU power (assuming similar coding efficiency). There's also the point that there are many more BC clients with PHE than uTorrent clients with PE running out there.

Peer Exchange: A function that gets you a few more peers by trading peerlists around with the guys you did connect to. I suppose I don't have to say how knowing about the existence of a few more peers helps you, especially in crap torrents. Protocol Encryption: A function that encrypts your BitTorrent data. That way, maybe you can fool those "evil" traffic-shaping ISPs.

I suggest EDITED. Try and find whether the stuff you want may be on a public tracker. I tend to avoid private trackers unless I have to, and I've only been forced to use one just now.

[site name removed for content] is apparently a private tracker. Well, private trackers these days are banning BC left and right. Have you ever tried signing up? Because of DHT and BC's rather uh ... liberal interpretation of said DHT, you may have been able to get files from [site name removed for content] even without signing in. It is probably the biggest thing (though apparently not the only thing) that got it banned. The "no entry" sign means it cannot connect to the private tracker. Now, after signing up, turn off anything that says "Antivirus" or "Firewall". It doesn't matter if you think there isn't a problem - just turn them off for a go anyway. Now start your download again. If it actually runs turn your antivirus and firewall stuff back on. If the download continues, great, if the download stops after this, come back.

Really. Do you know what kind of system is he trying to implement? From what little I see, Ludde is against anything that can be abused, and as far as going for limiters seems to be going mostly for behaviorial detection (features like trying to detect if BC threw away the requests or banning for consistently providing bad hashes). Which is good (in a "for torrent" way rather than for user - users want "Kick and Ban" available on a right click), but no doubt makes it hard to implement.

There is. It is called "Options->Torrent Options". To a new user, he's probably going to select his speed and go with the recommendation for awhile, before tweaking it which will be done there. If he is an advanced user that thinks he knows better, he probably won't use a Wizard anyway... APPEND: I just realized you might mean the box marked "connection type," and you want them to be able to select arbitrary uploads and downloads. The problem, I think, is that those are just simple locked presets - IF Connection Type A THEN Recommend Setting A. Allowing someone to just type in his own connection's downspeed/upspeed limits will be somewhat dishonest because all uTorrent will do is round it to one of those presets (and Rounding Algorithm = Extra Code). I can understand about wanting to kick people off if you are maxing out your connections but not your bandwidth, but why do you care what they are giving you if your bandwidth itself is maxed out? uTorrent is supposed to drop totally inactive guys if it needs the connection. But I'm relatively new and I already think Ludde doesn't want to implement something that would kick away guys that are merely slow - same reason why we are never likely to have a conveniently placed "Kick and Ban" switch.