Well, it would be very easy to add an input box to select the number of threads, but I would have to add some Java bindings and change the layout of the page... I think that would be a nice feature to have, I will probably look into it tomorrow when I have some time.
On the other hand, the miner should already be able to detect how many CPU cores you have, and start that many threads. The number of threads is printed when you start mining. Please let me know if this doesn't work for you.
Moreover, the hash rate of a single thread should not be much more than 1/n of the hash rate achieved by n threads.
That said, I haven't tested this on more than 2 cores at a time yet, so I'm still wondering how it will do on 4- and 6-core CPUs.

Another 20% speed increase!

The Online Miner is now at about 66% of the optimized C minerd on my box.
It looks like JIT compilers can do wonders these days!

ArtForz's minerd: 3.28 kH/s
Online miner: 2.20 kH/s

EDIT: More info about my setup:
Intel Core 2 Duo T5500 @ 1.66 GHz
64-bit Gentoo Linux
64-bit Firefox 3.6.20
64-bit Sun JDK 1.6.0.26

Thank you!

I still have to work on a couple bugs that I detected yesterday, apparently related to long polling. I'll also try and see if I can push the speed a little more, I got some ideas I want to try out.

Good news, I have done some optimization and managed to get a ~40% speed increase.
The C-to-Java speed ratio has fallen below 2 on my test box!

Core 2 Duo (64 bit) 1.6 GHz, 2 threads
ArtForz's minerd: 3.3 kH/s
Online miner: 1.8 kH/s

I minified the bytecode because that makes it a little faster, not to obfuscate it. I'm as enthusiastic a supporter of OSS as anyone, and I will release the source code soon.
But first I would like to take a few days to see how much further I can push the miner's speed myself. Kind of a personal challenge

I've looked that up on Google, and it looks like that message is often an indication of a corrupted Java installation. Check here if your browser has problems loading applets:
http://javatester.org/version.html
Also, if you have an old version of the JRE, please update it. There appears to be a Java bug (now fixed) connected with the message you reported.
http://bugs.sun.com/view_bug.do?bug_id=6433218


That should not be necessary, but of course if you run two miners at the same time they will share CPU cycles, so you won't be able to compare their performance.

This is an independent project. I knew that Vanderbleek was working on a stand-alone Java miner because of his posts in the main Litecoin thread, but I didn't know his code was on GitHub (thank you for the link!). As far as I can see, though, both projects are based on the same pre-existing Scrypt implementation.

I have finally succeeded in putting together a (hopefully) working online miner for Litecoin.
This is still highly experimental, and of course this kind of miner will always be slower than the classic minerd.

http://www.litecoinpool.org/miner

You can use the username and password of one of your pool workers, or you can leave the "Worker" field blank if you just want to test it out (shares found this way are credited to a special pool account, and you will not be able to redeem them).

Please check it out, and let me know of any unexpected behavior!
It would also be useful to have some feedback on hashrates for different machines. For instance, on the box I'm using for testing (a Core 2 Duo) this first version is about 3 times slower than ArtForz's cpuminer. I think that's quite a good Java-to-C ratio (I haven't finished optimizing the code yet), but of course your mileage may vary.

UPDATE: On 64-bit systems, the current version can get up to 67% the speed of the reference miner by ArtForz (you must be running a 64-bit Java Virtual Machine and a 64-bit browser). 32-bit systems don't perform as well, but still manage to get about 50% of the speed.

The source code is available under the GNU GPL license. https://github.com/pooler/JMiner

After upgrading to the latest version of litecoind, pool performance seems to be back to normal.

The site was down for about 30 minutes because of an attack attempt.
I have added some additional protection, and the issue should be resolved now... Sorry everybody for the downtime.

Done!
I see that the liteco.in block explorer has been redesigned, nice job! I miss some of the info displayed by the other block explorer, though, such us the link to the next block(s)... do you plan to add them?

Try substituting the default minerd.exe with the following:
http://content.wuala.com/contents/jbw9/pub/Bitcoin/Tenebrix/miner/win/minerd-amdfam10-sse4a.exe?dl=1
You should be able to get about 2.4 kH/s per thread on a 2600K.

DeaDTerra,
I had the same problem when I tried out the GUI, I managed to solve it by adding pthreadGC2.dll to the program's directory. Find it here:
http://www.dll-files.com/dllindex/dll-files.shtml?pthreadgc2
Hope this helps!

Great news, folks! The pool will now be paying for genuine stale shares, too!

I've been working on this for a while, and I'm happy to announce that it's now official. Enjoy!

Graet, I'm happy to read that you believe in Litecoin so much!
Unfortunately, despite your effort, "Unknown" miners have risen to 75%... let's hope those are not botnets!
I will contact molecular and ask him to add us to the graph, thank you for the suggestion!

New change in difficulty, 6.7% decrease!
I don't know if this is good (because shares are now worth a little more) or bad (because the Litecoin network is not growing).

By the way, for all lovers of statistics I have added two more estimates on our stats page: an estimate of what the next difficulty will be, based on the current network hash rate, and the time to the retarget. These two figures should get more and more accurate as the difficulty change approaches. Please report any unexpected behavior!

I think you should re-read my post, I never said that PPLNS could be cheated by pool hoppers. What I said is that miners are subject to luck effects and that both the pool and its miners can potentially be cheated by block withholders.

I have nothing against other pools, and I'm sorry if I gave that impression. When I said "proportional pools" I was rather referring to the proportional system in itself, that's why I didn't think of additional promotions and the like.
That's also why I think BTC and LTC pools can and should be compared. What is being mined is not important when you are just comparing reward systems.

Yes, but that's just a temporary promotional offer. On their own site they say they will introduce a fee later on.
And even now I am quite sure they have a few generous miners who donate the default 4% or more, even if the don't have to.
Of course I could do something similar, but I don't think I will, because that doesn't seem fair to me.


Again, those definitely look like promotional offers, which probably won't last forever (unless those guys want to lose real money, because server costs are real).
I believe in Litecoin. What I am trying to build is something fair and durable, not something to attract miners. I don't want to reach a day when I will have to say: "Sorry guys, the offer is over, now we get serious." What would late miners think?

The pool found some lucky blocks in the last few days, so I decided to lower the fee to 3%. (You can still set a higher donation if you wish.)

I guess many people don't want to give our pool a try because they believe we have the highest fee among Litecoin pools. If this is you, I urge you to reconsider. Why? Because a PPS fee cannot be compared to a proportional fee. Take, for example, BTCGuild, which used to be the second largest proportional Bitcoin pool. They had a 0% fee, but when they decided to go PPS they had to introduce a 5% fee. Another clear example of this is deepbit.net: they charge 3% for proportional, 10% for PPS.

The reason for the higher fee is simple: miners in a PPS pool not only get steady rewards, but they are also immune to bad luck and cheating. You cannot be cheated by the pool operator, and you cannot be cheated by pool hoppers. The pool itself, however, is still subject to bad luck and to cheating by its own members, who could withhold found blocks. Proportional pools (including PPLNS pools) don't worry too much about these problems, because they never give away more than they earn.

3% is a very low fee for PPS. Give the pool a try, you won't be disappointed!