I have not had one of these work units finish yet on any machine. I aborted some on one machine.
I have a bunch on these on another machine which have passed the 5 hr mark.

Anybody else had this problem with these w/u's

Ta
Nairb

Yep, they are long suckers, but they do finish
http://www.enigmaathome.net/result.php?resultid=19552986



Fish

Thanks for the info. I did have 1 of these w/u pass the 11 hr mark before I aborted it. So I was not to keen to try the others. But I did leave 1 w/u to run over night and it did finish too (12 hrs). A bit longer than the 40 min ones I had before.
I will let all the others one run to a finish too.

Ta
Nairb

Yep, these workunits are long, especially on slower machines, but they should run without problems.

---

M4 Project homepage
M4 Project wiki

If you think 12 hours is long, don't try any ClimatePrediction WUs. They have report deadlines of a year in advance, for good reason.

---

Click Here to see My Detailed BOINC Stats

I have actually done many ClimatePrediction w/u's. And some of them seem to last a lifetime. I think the short w/u of enigma give the feeling you are doing someting. A 3 month Climate w/u give the feeling of getting nowhere fast.
It was the change to 13 hr w/u that was a surprise. I think the machine will only manage about 70% of them.
But its good to have a project which a lower power machine can still do useful work without needing a vast amount of memory/diskspace and a huge cpu.

But I wont try the dual pentium 75 socket 7 machine ..... it usually takes 30 days to do 2 seti w/u's at a time....... The dual pentium pro machine... has already done some small w/u's

Nairb

I think I will get my last batch done on time, but just barely.

---

Pluto will always be a planet to me.

Just abort anything that won't finish in time, that's fine. I also like the short workunits, but a large number of workunits in progress stresses the server to the point where it's unable to hande the load anymore.

---

M4 Project homepage
M4 Project wiki

For now I set the long tasks to low priority, so they are send only if there's nothing shorter in the queue.
I'm considering patching the scheduler to send long tasks only to fast machines, because older PCs especially those running standard apps may be unable to finish them in time or even finish them at all (longer task - higher probability of something going wrong).
Btw, those pldrv210s are not that long.... There's a batch called m4pldrv64, it's the same ciphertext but running over full M4 keyrange instead of M3. These are really long, only 336 were send few days ago (most of them were bugged and will fail to start), onlu few came back so far with average processing time ~18 hours.

---

M4 Project homepage
M4 Project wiki

There's a batch called m4pldrv64, it's the same ciphertext but running over full M4 keyrange instead of M3. These are really long, only 336 were send few days ago (most of them were bugged and will fail to start), onlu few came back so far with average processing time ~18 hours.

And of course my old laptop gets a couple of those :)
Looks like they'll come in at about a hundred hours!




Fish

Perhaps you could implement some sort of work unit length preference setting. Rosetta@home does this, for example.

I've experimented with automatic WU assignment (no long workunits for slow hosts), but it didn't work very well - there were "no work available" issues if the queue was full of longer WUs. Same thing will happen with manual selection, unless I'll also tweak the feeder to keep enough workunit of any type.

---

M4 Project homepage
M4 Project wiki