I know it will take almost three years to test every single key at the current rate but what will we all do once RC5 is completed? Will distributed.net launch a new contest? What will happen to this juggernaut of computer power we have? Thanks.
RC5 is a waste of time (from a scientific standpoint - I do it for the stats, and to smash the /, and the DPC's ). I'm not going to do RC5 128, as it is about 3.4028236690 * 100^38 times (2 ^ 128 - 2 ^ 64) more difficult to crack than RC5 - 64.
I, along with many others, will do another project, but for myself, it WON'T be rc5-128 (I wonder if D.net will bother with it.... to track that as they do now would be a phenominal task of storage management)
Distributed Net is doing RC5 for the prize money. If there is big bucks riding on RC5-128, they might do it.
Then again, OGR has no money involved, but there is a definite addition to the Mathmatics Hall of Fame for finding the Optimal OGR-24 or OGR-25 values. If RC5-64 finished tomorrow, we would knock out the OGR contest a lot faster.
If you read the history of past contests, DNet has always been lucky and found the solution early in the contest. RC5-64 is 32% thru now, so not quite a third - still early.
If they're prepping for another contest, it has been very "hush-hush"
I think D.net are pretty much committed to doing RC5-72 (or whatever Rc5 contest comes after). This pretty much ensures RC5 for another 5 odd years realisticly. By that thime there will be other contests from other organisations.
I think in the new era of distributed computing coming, Dnet will find it hard to compete unless they undertake paying jobs, which rewards users with money for blocks submitted.
The big 4 projects (ie. RC5, OGR, SETI, and Gamma Flux) are only the tip of the iceberg for distributed computing, compared to what may lie ahead. With companies like Pixar coming forward to get films rendered on PC's on the cheap, who is to say that Microsoft wont get in on the action and use a client to help debug a new windows release.....
vss1980, I've never seen anything (outside of Forum speculation) that made me think RC5-72 was a project. RSA Security hasn't offered prize money. In fact, they just released the patents on RC5, so it won't be the revenue for them to sponsor prizes.
Unless a competitor of RSA Security comes up with another encryption busting contest (like CSC was), DNet will have to move on to more "worthhile" contests.
If you know of a DNet plan file (I've looked thru the archives) that actually talks about RC5-72, I'd be interested. But, don't tell DNet that I want them to do it, 'cause I don't.
JonB, I was just speculating. I have no solid information to assume Dnet will be doing RC5-72, but none to assume that they will not (they did 56, now 64, 72 is the next logical step), but it is pretty obvious at the moment that Dnet are only geared up to do only two projects, ie. RC5 and OGR. Although I do remember reading somewhere on the RSA website that RSA were sponsoring X amount of RC5 contests (Ps. I found the link): "The RSA Security Secret-Key Challenge consists of 13 contests in total, one using DES and the remaining 12 using the block cipher RC5 with a variable keysize" - RC5 & DES contests pages
If RC5 ended and they did not continue it, then OGR would get an almighty kick up the a$$ in terms of speed.
I would rather Dnet came up with something a bit more useful than RC5. I think we have proven that RC5 64-bit encryption is pretty hard (which is not what RSA wanted). Also, I would prefer to take a paying job from D.net or Dcypher.net as opposed to other project organisers.
personally, I think the 2 best distributed programs out there are seti@home and Folding@home. These in my opinion give the most scientific output. I would encourage people to switch to seti@home after RC5 is over simply because TA has a pretty good seti team and if we get some more heavy hitters we could break top 10 or top 8 fairly easily. Protien folding seems very usefull for scientists, but until the program gets updated and we get a TA team then i'm definately sticking to seti@home. I'm not trying to push one program more than other here (seti@home!!!!!!!) please no hatemail or responses about how you guys may or may not think these programs suck.
vss1980, more info on the Pixar thing, please?! : ) Having a little amateur 3D-modelling experience, I find that very interesting, and the idea that the project might pay back at least enough to cover the electricity used would be icing on the cake. I would still rather do a non-paying project with Team Anandtech than a paying project by myself, however. Let's hope they understand the stats angle, and have frequently-updated stats with teams and sub-teams, as well as individuals.
I suppose that after RC5-64, Dnet will migrate to OGR. With having proved the difficulty of RC5-64, the reason for any other level of RC5 would be pointless(unless the P4 somehow managed to pull a 20x multiplier for keys/sec), so they'ed probably pick something a little more scientific, since they've now layed the nessisary infastructure. Although, if we do go to another RC5 contest, I guess it's 7 more years of kicking someone's butt.
This explains some of what you may already know. Basically, Processtree (who are part of Dcypher.net) run Gamma flux which is ray-tracing. Pixar when rendering a scene use computers that are programmed to do not much more than that. I have no solid information, but it is rumoured that Pixar are being approached, or are approaching a Distributed computing company.
You have it right. After rc5-64, DNet will be moving on to OGR. They've said before that the current client will be able to do stubs up into the OGR32 range (I think I have that number right...) There will almost undoubtedly NOT be another rc5 contest of any size. It just becomes too large a keyspace. Remember, rc5-65 would be TWICE as long as rc5-64. rc5-72 would be 512 times as large. That would mean about 1500 YEARS to complete it at our current cracking rate!!!!
Even if all the RC5 clients switched to OGR, we would still have enough work for quite a few years. And by then, there will be other projects in the works...
I'm not sure even Moore's law would help us with further RC5 contests. There just becomes a certain point where the contests become nearly impossible. I've posted it before, and am too lazy to do the math right now, but even if every single person on earth had an eight-way Ghz Athlon computer (in other words, if there were 40 billion Ghz Athlons), it would still take several trillion years to complete RC5-128.
For those who are interested, OGR scales even faster than RC5. OGR24 took between 40-50 days to exhaust the keyspace. OGR25 took 40-50 days to get to 15%. Unless we get a major speed kick (RC5 ends) I'm not sure that we could complete much past OGR26 or 27.
i also so something interesting about distributed computing for aids research, maybe some of us should look into that, it's not all just about the stats (even though we all love them) I think that might be good direction to go into since it can help out so many people.
Some alien race could drop by our little corner of the galaxy, see the silly little contests that we're participating in, flip open his wrist watch, push a few buttons and come up with the answers in just a fraction of a second.
/end sci-fi BS
Computers have come a long way since I've been around them. There is no limit to what they can do, given time. Look at all the researchers working on coming up with a nural network. If you think coming up with the winning RC5-128 key is hard, try simulating the human brain with a few chuncks of silicon and some wire.
I have no doubt that the machines we're using right now will seem like the old Intel 8088's I used to play with as a teen to future generations. Just wait and see!