| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: Help Cure Muscular Dystrophy - Phase 2 Forum
Thread: Difficulty to receive HCMD-2 WUs |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 18
|
|
| Author |
|
|
KerSamson
Master Cruncher Switzerland Joined: Jan 29, 2007 Post Count: 1670 Status: Offline Project Badges: 
|
Hi everybody,
----------------------------------------After one week HCMD Phase 2, I have to observe the following issue. It is very difficult to receive HCMD WUs !  Usually I add new projects to the project list and I can rely on WCG project scheduler for receiving WUs of all the selected projects. Since the release of HCMD, I was only able to perform around two days of work (with 20 crunching cores). For this reason, two days ago, I decided to devote 8 cores to HCMD only. Being careful, I still left selected the option "If there is no work available for my computer for the projects I have selected above, please send me work from another project". The consequence is that I received HPF2 WUs and not only HCMD2. Because of past problems with HPF2, I modified again the boinc profile and selected only HCMD and IADS. Now, I receive 90% IADS WUs and only 10% HCMD WUs. Why is it so difficult to receive enough HCMD work ? I have the feeling that the balance between projects is not maintain on a fair way regarding HCMD ! ... Cheers, Yves |
||
|
|
smeyer55
Senior Cruncher Joined: Feb 15, 2009 Post Count: 303 Status: Offline Project Badges: 
|
I have the same settings (only HCMD2 selected) and I'm getting all I can handle.
Maybe try turning off the "send work from other projects" box and see what happens. Steve |
||
|
|
ctimm46
Cruncher Joined: Apr 2, 2006 Post Count: 17 Status: Offline Project Badges: 
|
In the interest of being careful, you might want to set your work buffer to a couple days, and then set it only to the HCMD-2 project. Just a thought this way you won't run out of work.
|
||
|
|
KerSamson
Master Cruncher Switzerland Joined: Jan 29, 2007 Post Count: 1670 Status: Offline Project Badges:
|
Hi Everybody,
----------------------------------------indeed I know how to force getting tasks for one project only. The reason for my remark is that the balance between projects is currently surprising regarding previously made experiences. It would be interesting to know from time to time the WCG priorisation rules for distributing WUs. Perhaps a kind of dashboard on the web site. Usually, I rely on WCG for receiving WUs for the projects I selected. Currently, I receive particularly few HCMD WUs (it is only an observation). Cheers, Yves --- PS: By the way, excepted during the beta phase, until today I did not experience any error with HCMD WUs using boinc 5.10.45 with W2K, XP Pro SP3, XP Pro 64 on P4, P4M, P4HT, Q6600, Q9450, and Xeon E5345. ---------------------------------------- [Edit 1 times, last edit by KerSamson at May 20, 2009 8:46:09 AM] |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
There is enough work in unit terms coming out the feeders except it's short so who ever comes first swishes up more tasks per fetch. HCMD [after HCC] had the most yesterday at 70,693 WU's validated of the 414,173 total.
----------------------------------------Anyway, it's 1. an upstart project 2. there are job sizing matters to be resolved [see beta] 3. WCG overall is tested to be stable at 400,000 WU per day, So we need to appreciate the Techs are not opening up the floodgates altogether.
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 1 times, last edit by Sekerob at May 20, 2009 8:45:22 AM] |
||
|
|
nasher
Veteran Cruncher USA Joined: Dec 2, 2005 Post Count: 1422 Status: Offline Project Badges: 
|
right now i am set to 1 project only but my computer uses boinc and i have other projects also que'ed up (very low %) so i can leave mine with only HCMD-2 i havent run out of work yet myself with my 4 computers(5 cores total)
----------------------------------------curently i have my cache size set to .3 days and i seem to always have 1 more to work waiting for me (at least) so i dont have to worry about waiting for work i have a feeling cause of the super short WU's there are alot of machines that have TONS of WU's waiting to crunch now and.. unfortunatly its going to be 15 days (probably closer to 8 left) for alot of this PV Purgatory to clear up.. soon as i get enough WU's to be gold i will swap to Flu (and other projects) but i will need to catch up on days to get Flu gold (but i got bronze so im happy)  |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Members loading up on work of a new project, maybe expecting it drying up soon, is the new reality. Those that continue to run short cache and return at normal pace will thus see on quorum projects an initial jump in PV tasks listed on their Result Status pages... for me no different. Of my 6 pages of HCMD2 PV, 3 pages are results fetched and returned on the 14th. e.g.
----------------------------------------CMD2_ 0001-PP1BA.clustersOccur-RADIA.clustersOccur_ 3408_ 0-- 849476 Pending Validation 14-5-09 10:11:37 14-5-09 18:02:30 0.07 1.1 / 0.0 But honestly, it can't bother me. They get validated eventually and we do know that in normal operation over 90% gets returned in under 4 days i.e. less than 10% takes a bit longer.
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I think the biggest reason for the massive number of PV is the incorrect time estimate for new WU that happened last week. I have a 3 core AMD computer that was set to 2 days buffer. One day that computer downloaded an avalanche of WU because the estimated time for each WU was way too short. When the WU were auto corrected to 1+ hours, I suddenly had 45+ days of WU in my buffer.
Fortunately there were many 10-50 minute WU in that pool. My buffer has shrunk in size to the point where there is a chance that I can crunch every WU and return it within the time allowed, without aborting any WU. My buffer is going back and forth between normal operation and high priority so I may still have to abort some WU. We'll get them crunched eventually, but you will have to wait until May 29 for the last of these WU.  |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Despite the complexity of the algorithm, BOINC remains very naive about estimation.
However, there are server-side limits, too, designed to avoid sending too much work. It may be better to abort some work early to avoid a quota problem later. |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Well, this series of very short HCMD2 jobs was enough of a reason for me to cut the cache quickly, for it's a well known event at WCG with now 9 projects active that fuses blow in buffer sizing and the accompanying rDCF destabilization [See other very recent discussions]
----------------------------------------
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
|