Embedable Javascript Bitcoin miner for your website

[lowentropy]

I am very interested in checking this out.  I signed up , got my snippet of javascript, but no confirmation email or login nor password....  Is the outgoing mail problem happening again?

Thanks!

We had some server difficulties earlier... apparently a server with 128MB of memory *can* serve 750 requests per second... it's just not recommended

If you PM me your client id (found in the script), I'll get you registered.

[1713538136]

1713538136

[1713538136]

1713538136

[1713538136]

1713538136

[LeFBI]

We had some server difficulties earlier... apparently a server with 128MB of memory *can* serve 750 requests per second... it's just not recommended

[/quote]
Do you you use the same server for all the getwork requests from the JS Clients as well? I thought you use the amazonaws cloud for everything? I hope the server can handle another view hundred constant workers i'm intending to throw at you
But before i do that...some more statistics would be nice. For example accepted shares, number of current clients for each page, overall hashrate of the javascript pool, a graph with workers/hash rate during the last day/s would be nice too(so you can see fluctuation), stuff like that...


questions over questions:

Just out of curiosity, what is the current global hash rate of the js-pool and how many clients did it need?

The current getwork response from your server is ~600byte. that traffic can be reduced to ~300byte by using gzip. Of course this would increase cpu usage on the server, but did you try & compare this yet? I did set up my own little pool and tested it with ~200 constant workers. there wasn't much of difference in the server performance, but the traffic reduced as described.

I also tested the patch mrb suggested here. I can confirm the speed increase by factor ~1.7. but i can't confirm the accuracy of the patch, i didn't go through the code or test it long enough to find a share. will you implement the patch?
there also is another JsMiner here that seems to have a better performance than the current sha256 implementation.  The miner itself is alpha version and doesn't have any work distribution whatsoever but the sha algorithm seems to be faster(at least it was on my desktop).

And last but not least....statistically....how many shares do you need on bitp.it to get a balance of 0.00000001BTC?
During the last 12h my visitors were constantly mining with +80 to 180 kHash/s and generated 2 shares. but it still shows me 0.00000000BTC. I know 2 shares isn't much at all compared to a classic pool but i wonder how many shares it will need to actually generate something with the jsMiner, so i can increase the number of workers i throw at your pool...

[lowentropy]

Do you you use the same server for all the getwork requests from the JS Clients as well? I thought you use the amazonaws cloud for everything? I hope the server can handle another view hundred constant workers i'm intending to throw at you
But before i do that...some more statistics would be nice. For example accepted shares, number of current clients for each page, overall hashrate of the javascript pool, a graph with workers/hash rate during the last day/s would be nice too(so you can see fluctuation), stuff like that...

... snip ...

And last but not least....statistically....how many shares do you need on bitp.it to get a balance of 0.00000001BTC?
During the last 12h my visitors were constantly mining with +80 to 180 kHash/s and generated 2 shares. but it still shows me 0.00000000BTC. I know 2 shares isn't much at all compared to a classic pool but i wonder how many shares it will need to actually generate something with the jsMiner, so i can increase the number of workers i throw at your pool...

We're running on a single instance at the moment. At least for this web mining project, it would simply cost too much (i.e. more than $0) to scale to larger or multiple instances. We can handle a little more traffic... we recently adjusted the update interval to reduce the load some (if a miner finds a share though, it still submits it immediately).

I'm not sure what you mean by accepted shares - the shares we show you are accepted and part of a round on slush. Unlike an actual pool, however, we opted to only disburse BTC once a week, so your balance will only increase once a week on Saturdays. This is mostly because web mining as a whole is just too slow to justify real-time updates to balance.

The other ideas for features are definitely things we are planning to add, but at the moment we're pretty busy getting our own pool ready

http://bitp.it/beta

[Antwan]

Hi. Can 1 script be used on multiple domains?

[lowentropy]

Hi. Can 1 script be used on multiple domains?

Yep. The console page will show statistics separately for each domain you put it on.

[anujfr]

can the script be on multiple pages on the same domain???

[1bitc0inplz]

can the script be on multiple pages on the same domain???

Yes, there is no harm in that.

[lowentropy]

Hi all, we're taking the site offline for maintenance and upgrades - I'll let you know when it's operational again. Sorry for the inconvenience.

[Zyrkon]

Even though FeFe doesn't share my point of few on browser-mining (because it might drain the power from Laptops), I think we need to improve on the bitp.it client. Doing some hashing with the CPU doesn't work even with dedicated mining anymore. I believe that the only way to do some efficient mining would be using webGl hashing (there is no popup for it either) with a low or middle Aggression factor. We would at least get some MHashes/s per client. Check out http://www.youtube.com/watch?v=rfQ8rKGTVlg .

Best wishes,
Zyrkon

[lowentropy]

Hi guys, bitp.it is back up.

Unfortunately, we are basically at capacity, so we have to turn off new registrations.

Anyone who has an account can still log in, and web miners will continue to work.

Thanks for all your support!

[seleme]

So, I missed this in few hours 

Man, is there any chance you could accept me ? I have video site that gets around 5 millions UV annually and this would be fantastic, specially since those guys stay nice amount of time on the site while watching videos.

Please, please, please 

[lowentropy]

So, I missed this in few hours  

Man, is there any chance you could accept me ? I have video site that gets around 5 millions UV annually and this would be fantastic, specially since those guys stay nice amount of time on the site while watching videos.

Please, please, please  

We really appreciate everyone's enthusiasm

Unfortunately, bitp.it does not make us any money, and the resources we have are just completely maxed out. I should also mention again that while web mining is a fun concept, javascript really is far too slow to make a useful miner, at least until WebCL mining comes of age and sees more widespread adoption.

To put this in perspective, bitp.it has made less than 1 BTC so far, and at the expense of being constantly hammered by 750 requests per second. One of our users even seems to have us integrated with an ad service, but even so, the payout is small.

The landscape for web mining will change significantly when javascript can reasonably expect to have direct access the the visitor's graphics card pipeline. We will definitely reinvest our time and money in expanding the bitp.it service once that happens.

Until then, any miners out there interested in helping us beta test our new pool are invited to sign up: http://bitp.it/beta

We wil be sending invites in batches over the next few weeks, and would love everyone's feedback in these early stages.

[tonto]

I, for one, think that web visitors should at least be told what's going on, even if just a little white lie ("We're testing your CPU for blah blah blah...").  If I visited a website and noticed my CPU spike, you can bet I'd close that window and never come back (and possibly report to various malware entities).  Or at the very-very least, the home page should not have that script running, and only if they go further into your site.  A proper homepage needs to load fast, get the information to your visitors, or you're not going to have many visitors. 

Just try to think of the ethical side.  Also like mentioned earlier in the thread, maybe for registered/paying customers, the script is disabled (or give them the option to disable it).  Above all else, tell visitors what's going on, or you run the risk of being marked as malware and/or lawsuits.

[TheSeven]

I just got 50KHash/s in Firefox 4.0.1, DOM worker mode on a Core i7 620M with an improved sha256 algorithm. The bitp.it frontpage gets like 2.3KHash/s on the same machine. That's a 2000% performance increase.
I'm putting this into public domain, hoping for some donations if it increases your earnings: http://pastie.org/2054699

DISCLAIMER: I have no idea whether this is bug free, as it isn't exactly trivial to find a share with a JS miner to check. It seemed to work fine for some demo data though.

Some stats:
Firefox 4.0.1, DOM worker: 50KHash/s
Firefox 4.0.1, UI thread: 20KHash/s (but only 15% CPU load due to 30 milliseconds of delay after every 1024 hashes)
IE 9.0.8112.16421, DOM worker: 13KHash/s (possibly no DOM worker support, so falling back to UI thread?)
IE 9.0.8112.16421, UI thread: 13KHash/s
Chrome 13.0.742.91, DOM worker: 12-81KHash/s (swinging badly, no idea why)
Chrome 13.0.742.91, UI thread: 9KHash/s

[lowentropy]

I just got 50KHash/s in Firefox 4.0.1, DOM worker mode on a Core i7 620M with an improved sha256 algorithm. The bitp.it frontpage gets like 2.3KHash/s on the same machine. That's a 2000% performance increase.
I'm putting this into public domain, hoping for some donations if it increases your earnings: http://pastie.org/2054699

While we appreciate your optimizing the code (and I will try and get this code included by default soon), you can't actually put it in the public domain... it is copyrighted and published under an open source license, which we retain, and which applies to any modifications. I'm sure you had only good intentions, I just wanted to make this clear.

[TheSeven]

I just got 50KHash/s in Firefox 4.0.1, DOM worker mode on a Core i7 620M with an improved sha256 algorithm. The bitp.it frontpage gets like 2.3KHash/s on the same machine. That's a 2000% performance increase.
I'm putting this into public domain, hoping for some donations if it increases your earnings: http://pastie.org/2054699

While we appreciate your optimizing the code (and I will try and get this code included by default soon), you can't actually put it in the public domain... it is copyrighted and published under an open source license, which we retain, and which applies to any modifications. I'm sure you had only good intentions, I just wanted to make this clear.

I can still put the modifications into the public domain, as they aren't derived from anything. (I'm referring to the actual sha256 algorithm here, not the minor adaptations to the rest)
As far as I understand that should be perfectly compatible with jsMiner's license which is basically a disclaimer only anyway.

[lowentropy]

I can still put the modifications into the public domain, as they aren't derived from anything. (I'm referring to the actual sha256 algorithm here, not the minor adaptations to the rest)
As far as I understand that should be perfectly compatible with jsMiner's license which is basically a disclaimer only anyway.

I'll make sure to separate the SHA implementation into a separate file and put that file under the public domain, then. Again, thanks for your help! 

[TheSeven]

I can still put the modifications into the public domain, as they aren't derived from anything. (I'm referring to the actual sha256 algorithm here, not the minor adaptations to the rest)
As far as I understand that should be perfectly compatible with jsMiner's license which is basically a disclaimer only anyway.

I'll make sure to separate the SHA implementation into a separate file and put that file under the public domain, then. Again, thanks for your help! 

You don't need to. Public domain means that you can do basically do what you want with it, including redistribution under non-public-domain terms.
My intention is just that everyone can cherrypick this algorithm without being tied to your copyright statement.

[LeFBI]

thanks for your work TheSeven! but i think there's something wrong with the modification. when "workerDone(data)" is called on line 346, the data variable contains a string which is 96 byte shorter than it should be
instead of:

000000019151f35c7844f5d981233dcf95d6c72ba29d93b78079fb970000116400000000e97bf2a 8ccde95e524153f377a42151513dc343cf43da46222e5f756469ebfe34df8b3651a1d932f000000 0000000080000000000000000000000000000000000000000000000000000000000000000000000 0000000000080020000

it looks like:

469ebfe34df8b3651a1d932f0000000000000080000000000000000000000000000000000000000 0000000000000000000000000000000000000000080020000

i noticed this because all shares my users found during the last hours with your modded script got rejected

message: upstream RPC error
code: -4

[hex]

https://wiki.mozilla.org/Platform/GFX/WebCL

Looks like WebCL is not close