What's so bad about chewing randoms?

[Bakwetu]

I have an Athlon 700@900 at home where I have no internet connection. Could I just install the client and let it crack random blocks when I have the machine on, bringing the buff-out to work and flush from time to time? What is so bad about cracking randoms, everybody seems to fear it more than a total power-out?

Is this possible to do and/or is there a better way to fix this?

 

[Slacker]

I had a machine doing randoms for a month, it had 18,000 keys in the buff out, when I flushed, only about 3,000 were good, that is, 3,000 of them had not been submitted by someone else already.

 

[Sloth]

If you are going to be walking finished blocks from the machine you can just as easliy walk blocks to it. Just use another machine to get a large buffer file together and you are all set.

Polo (I think) wrote a program to do just what you are looking for. You can search the archives for sneakernet. I am sure thereis a link somewhere for it.


S.

 

[Bakwetu]

Only 3000 of 18000-ouch. I seem to remember a thread here some time ago that there was a fairly small risk of chewing on something already submitted?

I am cracking here at work. Could I perhaps copy my buff-in file to another part of my hd, flush and fetch new blocks, take my newly loaded buff-in file with me home and put back the buff-in file I moved?

 

[JimMc]

Here's the link you need Sneaky

 

[BurntKooshie]

"I am cracking here at work. Could I perhaps copy my buff-in file to another part of my hd, flush and fetch new blocks, take my newly loaded buff-in file with me home and put back the buff-in file I moved?"

DO NOT COPY! Well, copy, then remove it from that folder. The reason for this is because if you simply copy it, then you'll have more than one machine crunching the same blocks! Copy it, but then move the buff.in to another folder, as a backup. DO NOT let them crunch the same keys, that is an utter waste of time!

However, DO make copies of your Buff.out! Only flush it once....but have more than one copy. If you don't know if you flushed it...you can reflush them. Sure, it looks suspicious on Mika's (if you do it on different days, because it only does day to day dupe checking) if flushed on different days, but at least you save the blocks.

What you are talking about is called "sneakernetting"....Polo made a program called "Sneaky" for JUST this situation. I don't rememeber the link, but i'm sure someone else does

 

[Bakwetu]

Burntkooshie
I know I don't want them to chew on the same blocks, that's why I thought I'd make a copy of my buff-in file, place this copy somewhere else on the hd, flush and fetch with the original buffers, remove the original (now full) buff-in and place the copy in its place. How does that sound?

Thanks for the sneaky link, I'll check it out

 

[redwing]

Just move the buff-in to a disk. This buffer can than be copied to the machine without internet access. The machine with internet access can than perform a fetch to get a new buffer. For the buff-out, move the buff-out of the no-internet PC to a floppy. Flush the buff-out of the internet machine, then copy the buff-out from the floppy and flush again.

 

[Bakwetu]

I dl:ed the sneaky program. Anybody knows what do I do with it?

 

[Bakwetu]

Redwing
Can I just remove my buff-in file to a floppy and use fetch to get a new buff-in file? In that case it's easy.

 

[ViRGE]

Ya, randoms are pretty nasty, since the odds are overwhelmingly good that someone's allready done them. Contrary to popular belife, the randoms aren't from the remaining keyspace. They are instead from a predetermined range that Dnet currently has open, as they only accept blocks in that range. Hence, the odds are much worse that 1/3 of not having the block count.

 

[spencer]

Bakwetu, You need to put a copy of sneaky.exe in the distributed.net folder on the machine without internet service and the one that has internet service that you plan on using to download workunits with. Then just run sneaky.exe

 

[Deeko]

Ahhh randoms are baaad. I just did my "superflush" recently, I had about 22K stored up, and only 8K of them ended up going through, Yuck!

 

[blade47]

As the contest continues and more blocks are submitted the odds of your randoms not counting will increase. Every block that's returned to DNET increases the odds that your random has already been cracked by someone else.

 

[Joe O]

Bakwetu,


<< Can I just remove my buff-in file to a floppy and use fetch to get a new buff-in file? In that case it's easy. >>


Yes, that is the easiest way to do it without Sneaky. Do it at work while the client is still running. This is important, because if you stop the client it will put the &quot;work-in-progress&quot; back on to its buff-in file.
At home, the easiest way to take the file from the floppy and combine it with your current buff-in file is to execute

<< dnetc -import a:\buff-in.RC5 >>

from within the directory where dnetc resides. If you are running Windows you can create a &quot;shortcut&quot; to do this.
Then you can move (keeping a backup) the buff-out file to a floppy.

At work, you can execute

<< dnetc -import a:\buff-out.RC5 >>

from within your directory that contains dnetc.

All this can be done without stopping/pausing the clients.

This also works for OGR. Just have shortcuts with &quot;.OGR&quot; instead of &quot;.RC5&quot;.