[RC5] RC5-72 Clarity?
chicks at chicks.net
Mon Jun 4 13:25:44 EDT 2007
On Mon, Jun 04, 2007 at 06:49:56PM +0200, Marcin Sochacki wrote:
> On Mon, Jun 04, 2007 at 01:48:44AM +0200, Danie van Heerden wrote:
> > Should rc5-72 be stopped, and OGR-P2 be completed, most of our installed
> > clients will no longer be serviceable.
> This is a side issue, but in fact D.Net rules state that you shouldn't
> run the client on computers without authorisation. For example, say
> you've been working at a university, when you leave the job, the clients
> should have been uninstalled.
> I know this is sometimes difficult to perform, technically, but frankly
> speaking you are boosting your stats with computers you no longer
> manage. Thus you shouldn't complain about old clients stopping crunching
> when RC5 is withdrawn from the project.
I understand how you got there, but I disagree. For instance, some ports are no longer maintained or current so folks are stuck running old cores on old clients to keep strange machines running.
The only reason I can see shutting down RC5-72 are if the resources are so exorbitant that its not worth it, but I think they've got that dialed in, so what does it hurt?
> > Could I perhaps suggest an automatic
> > core update on new dnetc clients?
> This could open a potential hole for malware, and this is probably one
> of the reasons D.Net never created the autoupdate feature. Also, if you
> still have authority over the machines, you can employ some easy, third
> party solution to update the clients. For instance I've used shared
> Samba folders for my Windows clients and autologon script, which would
> download the newer version if found.
That's great, but what if there are technical reasons that isn't practical?? Some large percentage of the dnet populous isn't that advanced, but they're still contributing. Authority is fine, but if it doesn't come with time, gas, and will, its no use.
I for one would love to see automatic updates as an option to help those who haven't recreated their own IT architecture just to contribute to a volunteer project.
> About the main topic -- if keeping up the RC5-72 infrastructure is not
> too expensive, then IMO it can stay up to leave some place for
> hardcore fans. If it's too expensive -- withdraw it altogether.
> Seriously, there is no point at all in RC5-72. The prize was never a
> factor for me; I crunched for RC5-64 and earlier projects because they
> were relatively short and fun. When RC5-72 was announced it was clear
> that >256 times more work is not going to be done quickly, even with
> Moore's law. So I switched everything to OGR.
The fact that it is so much larger is an attraction for some of us. We do three impossible things before breakfast just to warm up and then sprinkle miracles all around for the rest of the day. Breaking one dang RC5-72 packet is just going to take a little more sweat than most miracles. Hey - one of our machines could have just found the answer so there's less reason for despondancy than many have expressed.
> With today's huge power draw of the CPUs, we should first think
> about the environment.
I figured throwing out the TV and letting my computer take over was good for the environment. How about you? Have you moved into a smaller house? Have you traded in your automobile?
Dictating to other people what their priorities for thinking should be doesn't often lead to desired results. But if people want to have an earnest debate about the enviromental impact of what we're doing, its not hard to make some rough estimates.
> Having spare cycles is not as valuable, as having
> spare energy.
The energy consumption due to dnet is small compared to the amount being wasted by the computer and many other things all the time. Not running dnet isn't going to make a dent in your electric bill unless you're running a few racks of machines in the basement and then you're still in the single digits of percentage.
I waste gas by driving like a nut, but the 20% difference to drive nice isn't worth it to me personally. I hug trees, but if I'm stuck driving a car a little gas can get spent on joy and mirth and passing the muggles still trying to find the accelerator.
> Few people have the latter and I vote that if energy is
> spent for crunching, then it must have a useful purpose.
Your standards of usefulness and my vary widely. If we're making "musts" I hope we will be more precise.
> OGR is such a
> project, so it should be made the default in d.net, if possible also for
> existing older clients (I remember RC5 was the preferred one in default
> install if work units were available for both projects).
I agree for rather different reasons. It is clear than OGR is easier for folks to appreciate and should be the default. The apparent historic philosophy of preferring the longest term projects is contrary to human brain processes and we would be more effective and happier by spending our emotional and technical spare cycles on avoiding explaining this to people again and again. Inevitably it seems we'd free up mailing lists and brain cycles by sorting things from easiest to appreciate to hardest.
The net effect of changing the defaults is that all "well-maintained" clients will get updated and eventually default to OGR, wtfe, and then RC5-72. That means that the only people crunching RC5 will be the intent and the lazy and I'm honored to show up in both categories.
"The problem with troubleshooting is that trouble shoots back!"
More information about the rc5