is this even possible?

[CraftyOtter]

Lets say I was running the F@H CPU client in 2005 under a WinXP boxen at a University in Hawaii and today my username is still posting a 225 point WU/week from said boxen at said Univerisity.

Assuming the boxen was left running that long , would a CPU client from that era be able to accept WU from today's F@H?

My guess is that someone else is using the same username (common first name) but it made me go hmmmm. Could I still have a borg machine still folding???

 

[APOLLO]

I doubt it but anything is possible. What is definitely certain is the older clients wold not be able to accept some of the newest WUs. Whether an older client dating from mid previous decade is still capable of receiving anything will depend on many extremely unlikely scenarios, like the machine running 24/7 since that time. Reason being, clients expire after a certain set period, so we're talking an extreme unlikelihood bordering on the impossible.

 

[CraftyOtter]

I had 16 boxen setup for F@H before I left, all WinXP in a real server room running 24/7. Hmmm.

 

[APOLLO]

I had 16 boxen setup for F@H before I left, all WinXP in a real server room running 24/7. Hmmm.

Interesting, well, the room would have to be pristine, super air-filtered and temperature controlled 24/7 for systems to be running for many years without due maintenance. I just can't imagine it, but I'm not a server admin either...

 

[zero2dash]

I thought the clients expired after a certain period of time (which is why before SMP2 we had to use the drop in client for the SMP Windows clients). ?

Obviously the cores download as needed and then eventually you could have all of them on a machine left running for so long - the question is though, whether the clients expire (again, I thought they did).

 

[APOLLO]

I thought the clients expired after a certain period of time (which is why before SMP2 we had to use the drop in client for the SMP Windows clients). ?

Obviously the cores download as needed and then eventually you could have all of them on a machine left running for so long - the question is though, whether the clients expire (again, I thought they did).

Yes, they expire but if you do not stop and restart the client they'll still accept work. Old clients cannot receive new work like the 6000-series WU for the single core client but they can still receive other WUs. The question remains, did Stanford force a cessation on the single core clients even if they continued to run since 2005??

 

[Tobit]

I don't recall any of the early classic clients (uniprocessor) ever having an expiration date (but 2005 is 3 years before my time). There is still a bunch of AMBER work out there that will likely continue to run fine on these old clients.

 

[APOLLO]

I don't recall any of the early classic clients (uniprocessor) ever having an expiration date (but 2005 is 3 years before my time). There is still a bunch of AMBER work out there that will likely continue to run fine on these old clients.

OK, that's interesting. Then the updates to the client were not obligatory even after much time elapsed, unlike the SMP client. Knowing this, it's possible but still quite unlikely unless the machines are running without an OS change and with a service install, minor upgrades, etc.

 

[Pocatello]

I don't recall any of the early classic clients (uniprocessor) ever having an expiration date (but 2005 is 3 years before my time). There is still a bunch of AMBER work out there that will likely continue to run fine on these old clients.

+1

I agree.

 

[Zero82z]

I thought the clients expired after a certain period of time (which is why before SMP2 we had to use the drop in client for the SMP Windows clients). ?

Obviously the cores download as needed and then eventually you could have all of them on a machine left running for so long - the question is though, whether the clients expire (again, I thought they did).

OK, that's interesting. Then the updates to the client were not obligatory even after much time elapsed, unlike the SMP client. Knowing this, it's possible but still quite unlikely unless the machines are running without an OS change and with a service install, minor upgrades, etc.

The SMP client versions only expired because it's in beta. Non-beta clients do not have expiration dates as far as I've seen, and I ran the uniprocessor client for a good 3-4 years without ever going to a different version.

 

[APOLLO]

The SMP client versions only expired because it's in beta. Non-beta clients do not have expiration dates as far as I've seen, and I ran the uniprocessor client for a good 3-4 years without ever going to a different version.

I have been back to folding 3 years this spring, and the only time since then that an upgrade was really suggested was in the beginning of the year, so I would agree. However, 5 years is stretching things quite a bit and I'm surprised they may still be running for other reasons.

 

[SmokeRngs]

It's completely possible that a client from then is still running and crunching work units.

The only reason people are accustomed to expiring clients is because most of us are running beta clients and have been for the last couple of years at least. All the SMP clients are beta and many of the GPU clients are beta. There hasn't been a single non-beta SMP client yet.

The single processor clients don't have built in expiration dates once they move out of beta. Stanford will suggest moving to a newer non-beta client when new ones are released and will no longer offer the older client for download in most cases but that's about it. If you'll look on Stanford's client download page, you will find clients for Win9x yet. The newer clients will not run under the old 9x OSes so they keep the final version of the old non-beta clients out there.

Older clients like that will have serious limitations on the type of work units they can crunch which can mean the client may not have any work to do for a while in some cases. This is mostly because some newer cores require newer clients to work. That said, an older client will not get work for which it cannot process.