unforum

[poozle]

[SimonBitdiddle]

[chimera245]

Hi,

The test we are using to see if the solution has been decoded is that every character in the decode (160 from memory) has an ASCII value < 127.

In actual fact this includes a number of unprintables as well as all printable characters, but the likelihood of a false positive is fairly low (2 ^ 160 to 1 against).

I haven't really worked on Sk1zz's new code much yet - it is (or was at least) somewhat restricted to particular processor architectures - though I have to say under those appropriate circumstances it runs like hot s**t off a shovel.

Even so however it is only a factor of 10 or so faster than the current code on those environments. IF we got to a full keyspace (255 characters) as some are proposing - the new key space 255 ^ 8, which is an astonishing 80000 times larger. Under those circumstances even with a universally applicable super fast client solving the problem becomes an issue that would require a step change in number of clients.

[cwmajors]

Note, Chimera, that I don't propose we go immediately to a full keyspace- I just want to prepare for the possibility that the current keyspace won't find the key. I think we should start considering efforts to increase the number of clients (and, to coincide with that, increase the ability of the server to handle those clients).

I hope we do find the key within the space we're looking now. But if we don't, I'd like to make sure we have a system in place so that people don't see it as a defeat and wander away from the project - ideally, the clients would just clear the current keyspace and keep working. (I assume this is the idea already.) And I do think we should start brainstorming some other publicity ideas to increase the installed client base.

For example, I've been considering writing a little app I like to call "Bonzi 13th Labor Buddy." Waddaya think?

[SimonBitdiddle]

I was also wondering why we started with the capital letters or didn't divvy the keysearch into sections like numerals, lowercase, numerals and lower case, capitals, numerals and capitals, and then numerals and both cases.

You'd think that the password would start off from the lower parts of the ASCII table, to keep the keysearch relatively quick.

... Unless they /really/ dislike us.

[mac_monkey]

[GuyIncognito]

@chimera 245:

First of all, thanks for all the time & effort you put into the client and into patiently answering questions about it! Props for that!

On that note, here's mine:

You mentioned a couple of pages ago that sometimes drops in the number of running clients represent people losing interest in the whole thing. So what happens to their unprocessed WUs? Do they get redistributed?

E.g., someone runs the client and get's assigned a packet of 10 WUs, say "Bastardg" to "Bastardp", but after processing WU 1 of 10 loses interest and never runs the client again. What would happen to "Bastardo" then?
Would "Bastardo" get flagged as processed as soon as someone's client downloads the sequence of 10 WUs it is part of, or would you be able to see which WUs actually have been tried? How would these "gaps" in the sequence of processed workunits get reassigned?

Other than that, I also have complete confidence in the client and the keyspace we are searching...finally, here's my chance to contribute something to a solve!

[chimera245]

[quote="GuyIncognito]what happens to their unprocessed WUs? Do they get redistributed?[/quote]

I recycle unsolved units that have been allocated too long periodically. This shows as a blip in the number of units allocated.

[UKver2.0]

This has been said before, but nobody reads back more than a few pages - I'd like to think MC wouldn't donate prizes to the effort unless they knew this was the way to go. They monitor these forums and #syzygy and therefore know what the search space is that we're using. Michael Smith was even quoted as saying how long he thought it would take the client to find the answer (although he was probably talking about the time it would take to search the entire space so as not to give any clues). "Patience Willow."

[cwmajors]

I hope nobody has misinterpreted me as thinking we're on the wrong track here. All along, I have had nothing but confidence in the procedures Chimera and the rest of the people responsible for the effort so far are using - I'm helping myself. And even if I'm right in thinking the key is not in the space we're searching now, I still think we're doing the right thing by eliminating the "printable" keyspace - we have to start somewhere, and the keyspace we're using is as good or better than any I could propose. I just don't want there to be a huge revolt (against MC -or- Chimera) if we get to 100% of the keyspace we're currently searching and the key, for whatever reason, isn't found.

[Cinana]

Well, add me to the ranks of computers doing it.

Thank you Chimera for helping me get it understood

My computer is currently on unit 1380337742 = RFHN as of about 5 mins ago.
I am processing from beginning of start to end (1/10 thingy) at just under 4 mins.

Could someone be so kind to PM me the correct terms for each of these bits of info? (ie: unit, hex, ascii letter I already understand I think... its the other "thingys" I don't quite understand. And to explain these thingys to my husband, and friends at work, they think I am nuts.

In regards to branching out, and getting more people to participate...I am working my friends into this mess. Also going to hijack a couple of my husbands computers as well.

[jbd]

This has already been pointed out in the blog thread, but it could use repeating here:

[Furry Mark]

I'm guessing if we need a 'second run' that we will need a new release of the client, since the client provides the last four characters of the keys being checked. This could be a bit of a pain for those of us running on multiple systems.
On the positive side, if we have to upgrade the client it will provide an opportunity to add the superfast optimisations (and maybe some extra features like automatically starting at idle priority... )

[Sammykat]

Sammykat

[chimera245]

Looks like the web server decided to reboot itself at 3.02am this morning to install a critical security update.

Bloody Microsoft.

It then waited until I got up this morning (about 6.15am) to come out of power save mode. Go figure.

Anyway all is well and back up again.

There will be a maintenance window coming up at some point where I do an upgrade on the SQL Server back end - but probably not for a week or two.