LHC@home

I run in dialup mode, but this could apply to anyone.

lhc sent me 10 WU's earlier today, and a strange thing happened.
As soon as a WU came in, it started up.
Then when the next WU came in, the first was "preempted" and the second started up.
Etc, until I had three preempted, one running, and six waiting for work.

SO, instead of one slot occupied with lhc, I had four occuppied by lhc.

An examination of report deadlines confirmed that they were not shipped to me in timestamped order, but the lhc client insists on executing in timestamp order.

I guess this more of an interesting observation than a bug.
However, if all ten WU's had come in the wrong order, that would be more inconvenient.

I'm running BOINC 5.2.5 on this machine.

Ned

---

happens to me all the time, sometimes i have 10 WUs and more preempted cause of that, pesty memory hog ...

---

I've noticed the same thing on 5.2.6 also

---

- da shu @ HeliOS,
"Free software” is a matter of liberty, not price. To understand the concept, you should think of “free” as in “free speech”, not as in “free beer".

This project uses absolute deadlines that are assigned when the workunit is created. Most projects use relative deadlines that are assigned when the result is downloaded. They do that here because they need high turnaround. For the same reason they normally send out five workunits when they need a quorum of 3.

---

BOINC WIKI

BOINCing since 2002/12/8

Your computer is also running in EDF mode due to it's connect preference being more than 80% of half the earliest wu deadline (80% to take into account the crunching time of a wu). As I've seen deadlines on LHC ranging from 10 days to 7 days to 5 days, it is difficult to set a connect preference to stop BOINC going into EDF mode, especially when your on dial-up. The only way to overcome this is after the downloads are complete, exit out of BOINC completely and re-start or set your preference to remove from memory when preempted.

It is interesting to note that Rosetta which has fewer members has a higher RAC when compared to LHC. I think the admins need to ponder why it is so. I know one of my team members who had a RAC of over 1,200 has now stopped LHC altogether due to the poor amount of credit being granted when compared to some other projects. I also find this an issue along with the variable deadlines causing EDF mode which causes cache levels to be screwed with causing my computers to ask for work right when LHC is out of work. This is only a problem since I like to see BOINC have the "right" amount of work cached for 4 days of real work since a couple of my computers are also on dial-up.

Live long and crunch.

---

Paul
(S@H1 8888)
BOINC/SAH BETA