| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: FightAIDS@Home Phase 2
Thread: FAH2 behaving poorly: % progress is misreported; and the way it "cuts in line" so its Work Units get done before other projects' |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 10
|
|
| Author |
|
|
nei1
Cruncher United States Joined: Jan 30, 2014 Post Count: 38 Status: Offline Project Badges: 
|
An image of my BOINC client:
----------------------------------------https://drive.google.com/file/d/0B55A9WM8IQb3...Zm9VX0xXNXE2dG83Ri1z/view WCG "Results Status" page reports the BOINC client I got this image from is a "Linux 4.13.0-37-generic." I call it Linux Mint Mate 18.3. I'd like to state two concerns about FAH2 work units, as they appear on my cutting-edge, core-2-duo laptop. 1) As demonstrated in the image, the "Progress" percentage is usually wrong. In this case, the "Elapsed" time is a little less than the "Remaining" time. Yet, the percentage "Progress" reads 74%, which is contradictory; if anything, it should be a little less than 50%. I've seen worse discrepancies than this example. Either the Progress percentage is incorrect, or the "Elapsed / Remaining" times are incorrect. No other work units from other projects that I've seen have this problem, their times and %-progress correlate with each other and the clock on my wall. But, most of FAH2 work units have contradictory times vs. percentage progress. I think that should be cleaned up so incorrect data is no longer provided. 2) When new work units from FAH2 come in, they arrive with a Must Complete Date of about 24-hours-from-now, and work units from other projects are paused and put on hold. None of the other work units from other projects have this behavior; their due-dates are about 5-or-10-days-from-now, and they pretty much coexist without having to preempt each other. Is it that FAH2 work units are considered to be higher priority than the other projects? Or do the folks at FAH2's offices simply need to be told to stop doing that? Or am I totally misconstruing? Thanks, -neil- [Edit 2 times, last edit by nei1 at Apr 4, 2018 8:34:21 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hi nei1,
You have noticed two things that are the way they are just because FAH1 is the way it is, if you follow me. Let me try to explain. The estimates for FAH1 don't settle down to be roughly accurate, as they do for other sub-projects, because there are two different types of work unit (WU) which are of very different lengths. BOINC does not contain a mechanism to learn different estimates for these two different types -- it can't tell them apart. The one you show is a shorter one, so BOINC expects it to run longer than it actually does. The estimate for a particular WU only settles down when it approaches something like 90% complete -- at that point BOINC is fairly sure it really is going to finish early. If it were a long-running WU you would see the opposite behaviour -- the estimate would be too low instead of too high. As for the pre-empting that you see because of the short lead times set on these WUs, that is a conscious decision of the techs and the scientists because they want these back quickly. Simple as that, and not up to us crunchers. If you don't like it, you're only choice is not to run this sub-project, I'm afraid. I'm sorry if that's not what you wanted to hear. It's just the way it is and beyond our control. |
||
|
|
nei1
Cruncher United States Joined: Jan 30, 2014 Post Count: 38 Status: Offline Project Badges:
|
Hi Apis. Thanks for the quick response and for not slapping me around too much.
----------------------------------------I have halted work on FAH2 here and there, mostly as part of an experiment to see if I could get the graph of my Average Work Done to jump around less and smooth out. I don't think that helped; in fact when I looked at the returns a couple of weeks ago, one uniquely long FAH2 work unit was awarded about 4 times more credit (per unit of time) than the other projects were awarding. I think it gave 200 points, which is a lot of credit. Darn; I should have taken a picture of that result, but now it's gone. Since then, I've rejoined FAH2, again. Regardless of its boorish behavior, I prefer to check off participation in all WCG projects, and let them decide what's best to serve my little low-powered computers. (Maybe I'll get a i7-6700 Linux for Christmas.) As for FAH2's "% Completed" problem, at least the problem is not on my end. Thanks for confirming that. Nothing left to do but write my congressman; he's not doing anything, anyway. Happy crunching, -neil- [Edit 1 times, last edit by nei1 at Apr 5, 2018 6:00:47 PM] |
||
|
|
fablefox
Senior Cruncher Joined: May 31, 2010 Post Count: 161 Status: Offline Project Badges: 
|
Thanks for the explanation.
----------------------------------------And I thought I'm completing a wingman someone just aborted last minute due to hardware crash or something. Only after kept getting it I'm sure it's under high priority. |
||
|
|
nei1
Cruncher United States Joined: Jan 30, 2014 Post Count: 38 Status: Offline Project Badges:
|
Hi Fablefox,
Me too. I thought FAH2 was having an emergency. With 24 hours notice before Time Due, they certainly seem to be having a perpetual emergency. But after a year of "emergencies," it seems that's just the way they set up their work units to behave in our clients. I keep thinking, "You have to select your battles wisely." I'm sure I can find something else about which to raise a ruckus, instead of complaining to medical researchers whom are probably angels. (When it's all over, you get the most points if you've argued with angels as little as possible.) Another reason not to complain too much is that regardless of FAH2 cutting in line all the time, it's not stopping any of my other projects' work units from getting done [by their Time Due]. The cutting in line could also be explained, I imagine, by a scenario where FAH2 research generates work units and immediately submits them, and they really could use the answers ASAP. And maybe the other projects already have their work units planned out months in advance for whatever reason. And so, the two different styles of behavior. If the other projects don't complain, then who am I to complain about FAH2. And I think the more powerful the client computer, the less of a potential problem it would be. cu, -n- |
|||
|
||||
|
PecosRiverM
Veteran Cruncher The Great State of Texas Joined: Apr 27, 2007 Post Count: 1053 Status: Offline Project Badges: 
|
My way to get around this is I only run a 2 hour cache (not good if anything happens).
----------------------------------------   |
||
|
|
nei1
Cruncher United States Joined: Jan 30, 2014 Post Count: 38 Status: Offline Project Badges:
|
My way to get around this is I only run a 2 hour cache (not good if anything happens). Hi, PecosRiver. 10-4. It sounds like you're also bothered by the way FAH2 cuts in front of the other projects, and you're also trying extreme measures to get FAH2 to behave correctly? I sympathize. I gave up trying to get FAH2 to behave correctly. That means I selected all of the WCG projects, and the Boinc cache is holding 1.0 day's work, about 10 work units for the old laptop. Sometimes, there are no FAH2 work units, and everything is systematic. Sometimes, FAH2 breaks in, but I'm getting used to it. -n- PS: Googlemaps > Pecos River > No Parking On Bridge / No Fishing From Bridge. Looks like beautiful country and a delicate ecology. Not to mention wikipedia. Long Island used to have a delicate ecology; now we ain't got any. |
||
|
|
l_mckeon
Senior Cruncher Joined: Oct 20, 2007 Post Count: 439 Status: Offline Project Badges: 
|
To those having trouble with FAH2, why don't you just opt out of FAH2 in "My Projects" and crunch other jobs, rather than messing around with cache sizes?
|
||
|
|
nei1
Cruncher United States Joined: Jan 30, 2014 Post Count: 38 Status: Offline Project Badges:
|
> opt out of FAH2
----------------------------------------Yes, WCG gives us the option to choose which project(s) to participate in. Personally, they all seem important to me. [Edit 2 times, last edit by nei1 at Apr 12, 2018 2:29:28 PM] |
||
|
|
fablefox
Senior Cruncher Joined: May 31, 2010 Post Count: 161 Status: Offline Project Badges:
|
Actually I don't mind. I've ticked several research and only recently I ticked FAH2 because I've gained 1 year on Smash Childhood Cancer. Currently I'm working on Microbiome/TB/FAH2 with FAH being intermittend (just 6 day shy of 1 year).
----------------------------------------All research are important so I've decided to gave 1 year to each. When I first started I just tick one and stick until finished. After I realized that Mapping Cancer Marker is open ended I decided to give at least each research 1 year, and just help complete whatever reaching their end stage. |
||
|
|
|