What are your opinions on using old PCs for BOINC projects? I have a couple of modern PCs but I also connected a couple of obsolete PCs (a PIII and even a K6). Are the old PCs welcome because every little bit helps the project or are they frowned upon because they can delay others with faster PCs (especially in getting their credits granted)?
I'm new to BOINC but I'd like to contribute as much as I can and was also hoping to get some more life out of the old PCs. Personally I'm of the opinion that an old 500MHz PC running constantly isn't too much slower than a modern PC running 6 hours a day. But I don't want to annoy anyone...
Copyright © 2024 Einstein@Home. All rights reserved.
Opinions on using old PCs
)
I have a 400 MHz Pentium II running 24/7 and sometimes my credits in SETI are pending because people with much faster CPUs have not completed a WU. But this seldom happens in E@H.
Tullio
I have a 500MHZ Celeron and a
)
I have a 500MHZ Celeron and a 600MHz P3 both running projects, and am happy they can help.
RE: . But I don't want to
)
>I don't think anyone should be annoyed if you are returning WUs before the deadline. We have an old 400MHz PII laptop that does its share and everyone seems happy. I know that the E@H project will be happy to have all you can donate....Cheers, Rog.
RE: RE: . But I don't
)
I certainly agree that E@H will take whatever it can get. I would have one suggestion though for older machhines comming online with the project: I would try to make sure the machine is operating at its fullest. A while back I decided to buy a utility program for finding bugs that can slow things down. I used it on my older PIII 863 and found a truckload of errors that were probably up to no good. It seemed to work better after that. The program I bought was Bug Doctor. Hope this helps.
Gerry Rough
(Click for detailed stats)
RE: RE: . But I don't
)
Agree that nobody *should* get annoyed so long as results come in before the deadline.
In practice, in any group of people there are some who will get annoyed at anything that is not exactly as they like it, and distributed computing is no exception. A very small minority do get irritated if they have to wait for work to be verified, sometimes they get vocal about it.
My view is that the project team set the rules, including selecting deadlines. It is not right to let a very few speed nutters impose a more restrictive set of rules on us just for their own pleasure. Over half the credits listed in my sig below were gained on machines running at 700MHz or less.
[pre] MHz
1x Dell optiplex 700MHz celeron ............. 700
2x 500MHz P3 compaq small form factor (sff).. 1000
1x 600MHz P3 compaq sff ..................... 600
1x 733MHz P3 compaq sff ..................... 733
3x 866MHz P3 compaq sff ..................... 2600
2x twin 665MHz P3 HP Kayak servers .......... 2660
----
8293[/pre]
Total 8.3 GHz
Purchase cost 10 boxes second hand under £250 in total. How else can you get 8 GHz for £250 eh? In terms of crunch per buck purchase cost, old boxes are deifintiely the way to go.
In terms of crunch per buck power costs, new boxes are better - but then you have to factor in the savings in heating: I only turned the heat on in my lounge twice this winter, for a total of maybe 16hours - for the rest of the winter my BOINCing has kept me warm! In summer I turn off most of my slow boxes because of the power usage.
If you've already got the box, and are willing to donate the power usage, I'd say come in and you will be welcomed by the project team and welcomed by the vast majority of crunchers.
Gerry offers good advice - I'd say either re-install the operating system from scratch and apply all servie patches, or run a program of the kind he suggests.
But don't expect to please *absolutely* everyone - it is not possible, neither here nor anywhere else
~~gravywavy
PS - talking of old machines, they are also networked by old technology - I have a thin-wire coax net running to link 9 of the boxes to the firewall box, all of it reclaimed kit from elsewhere ;-)
~~gravywavy
RE: [ In terms of crunch
)
The way I see it this is a "run what ya brung" deal. If you can afford or have access to later model systems, then, run those puppies. If not, run what you have. Most of my systems are ones that I saved from the landfill over the last several years, and they do just fine.
With the new and improved science app the older systems are even more worthwhile. Just brought several PII 300's and a old dual Pentium Pro back online running C47. Even at 200Mhz x 2 that old dually runs 2 WU's every 24 hours.
I agree. Anything that can
)
I agree. Anything that can run and you're willing to run it, then use it. That said, about a month ago I signed up to the Climateprediction project and my 1.7 GHz P4 has now clocked up 300 CPU hours one it's first WU and there are still 2500 hours to go. I would not recommend using anything less than a 2.0GHz P4 with lots of memory and at least 15GB of disk for that project. Even my new 3200+ 64 bit laptop which has run up 340 CPU hours is still only 25% through a WU.
It's a worthy project, but not, I suggest, for those with smaller machines.
I'd personally never run
)
I'd personally never run CPDN, from experience.
The project fragments NTFS partioned hard drives more than any other distributed project I have run so far. Excessive fragmentation slows down your system and causes your hard drive to work harder to recover the smae amount of data. Hard drives are just like car engines, the harder the work, the quicker they die.
I'd stick with Einstein. :)
Overclock with the MSI G31M3-L and Intel E8600 3.33Ghz
Intel D865GLC Socket 478 Motherboard Review
Overclock your ASUS 1005HA netbook and crunch more
OK, there are a few files
)
OK, there are a few files that seem to have up to 300 odd fragments, but they don't seem to be hit more than once every 48 time steps (app once every 3-4 minutes) and most of these files are "open" anyway, so it's no real problem. Also it's only the application that accesses the fragmented file that is slowed.
Re working the disk hard, I would think an HDD would be worked harder if the Media Play is in regular use.
You are right in thinking the project uses lots of disk space though, 1.25GB for Phase I (of 5), which is why I recommend people with small disks don't run it.
Alright then, I'll keep the
)
Alright then, I'll keep the old PCs connected to BOINC, at least until my electric bill starts getting out of hand.
I found a couple more motherboards in my basement (a 600MHz P3 and a ~350MHz P2). If and when I dig up a hard drive for the P3 I think I'll connect it as well. The P2 might be pushing it though. I have a 450MHz AMD-K6 currently running Einstein and it takes almost 4 days to finish a WU.