unforum

[thalamus]

[meat_paste_dave]

Yup, that works.

Cheers,

meat_paste

[sackofpotatoes]

This effort was just mentioned on Boingboing:

http://www.boingboing.net/2006/04/12/perplex_city_players.html

Hopefully this will get us some more clients!

Good thinking whoever submitted this to them...

[rising]

A friend introduced me to PXC and pointed me here with the intention I'm sure of lending cpu time. I'm got the client going on a couple of machines, one of them being a dual core, dual processor xeon in the office which I don't use a lot right now - so it might as well help here . I have a couple of questions about the client

1. Last I read in this thread the client we're using was based on rccrypt 1.6 but the creator of rccrypt has mentioned a speed increase if using 1.4 which is also likely to be the version MC used. Is this still the case and wou ld you consider using the faster version?

2. I've not heard much said about the p flag which xor's with the previous block. Is the assumption that as the default is it not being used that that is how MC used it.

3. When the code decides if it has a potentially correct answer is it assuming the entire response is printable characters are a large percentage say 80%. Obviously this increases the number of false positives which may or may not be acceptable.

[sackofpotatoes]

I would just like to point out that at 12pm PST today there were 338 active clients, and at 9pm PST there are now 440 active clients! 4000 work units processed in 9 hours - not too shabby.

Keep recruiting!

[chimera245]

To answer rising:

1) Though I supposedly used the 1.6 code as a base, in the bits of code I actually used the two are the same, and I then adapted them to .NET and made some performance improvements of my own

2) We are assuming that there is no xoring to be done, because as you say this is the default RCCRYPT behaviour and it would add exponentially to the complexity of the task.

3) We are going for 'all characters below ~' as our candidate success indication. This will bring in a bunch of false positives, but it should be worth it.

[Guin]

We have now been offered a sock puppet as part of the prize

I have gladly accepted the offer and know this will only encourage a great deal more people to participate.

I am away until Monday evening (will pop by later today though) so if there is anything urgent re teh site pm me / mail me.

We have a gmail account now set up (thanks guys) so others will be able to check the mail if its non site related.

oooooh I love it when a plan comes together....

[Envoy]

I've also signed up and am cranking on the code in the background.

I don't even own the card. Rawk.

[rising]

Thanks chimera245 for taking the time not only in answering my queries but the required translation of my sleep deprived English as well (re-reading is painful).

Regarding your answer to 3. Although it would create even more false positives, have you considered the slightly more relaxed indicator of a given percentage of characters below '~'. The only reason I bring this up is having performed many experiments myself I've often wished I had recorded more information the first time round.

Yes it means more processing in phase 2 (the increase in false positives could possibly be calculated in an upper bound?) but reprocessing in this instance really isn't feasible. Maybe I'm just being paranoid but any changes to your client now would only mean reprocessing certain work units if nothing came up in phase 2 the first time and I'd hate for all of this to finish without getting what you came for

Feel free to shoot it down in flames oh and just for the records well done on all the work you've put into it, seriously impressed.

[SocialElite]

how much traffic can the website itself stand? It was a good idea to spread this to boingboing, but have people thought of spreading it to other high traffic sites that also might help? anyone here feel like trying to get us an article on slashdot?

I know that tycho from penny arcade was pretty big into ILB, so I shot him an email both teling him that he should check out PXC and (essentially) begging him to post a link to 13labour.tk on their main page.

anyone have any other suggestions of sites that might catch the eye of "our kind of people"?

[mac_monkey]

It's been submitted to slashdot more than once...we're just hoping we can withstand the traffic and that the mods at unfiction turn off guest posting, so that troll exposure is minimal

[sackofpotatoes]

I'm running the client on a dual CPU Linux box. It seems that the client is only using one of the processors though. I'm not a linux person, so if someone could help me get that second CPU cracking, I'd appreciate it. Thanks.

[rising]

sackofpotatoes, my suggestion to you would be to make another copy from the original zip and not place it in the exact same location. One of the files in the directory holds which work unit you are currently processing so starting the same program more than once processes the same work unit multiple times (what I found under windows).

So open up a terminal per process and simply launch one from each folder. Each one will ask you to submit your email details once and then you simply process a work unit per processor. It's a lot easier than multi-threading the application

[Flynn]

[sackofpotatoes]

That's got it - thanks!