| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: Help Defeat Cancer
Thread: Increase in Process Times |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 7
|
|
| Author |
|
|
Impx
Cruncher Joined: Sep 7, 2005 Post Count: 9 Status: Offline Project Badges: 
|
I've noticed another large increase in run times for HDC units.
Is it just random or is the project increasing the amount of CPU time needed for one unit? I havent seen any annoucement from the project team about this if its on purpose. (Maybe its just random units taking some extra time? Bad/larges units even?) But it would be nice to know if its been increased on purpose. Previously these used to run at 3-5 hours on my PC. A month or so ago they where at 10-12 hours. My last lot of units this week where at 20-25hours. Im on a 3.4g P4 with 2gig ram. I almost aborted the units when it approached 20 hours but decided to leave it which it finished shortly after. My next lot of units now have a similar time of about 20-25 hours. Ive compared this to other projects and they still process in "normal" times. So it seems to be project based. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hi Impx,
I don't get enough HDC units on my BOINC computer to be sure, but my recent units are running in about the same time as the FAAH units. We are currently using more filters than we were at the start of the project. We knew at the start of the project that we were going to be running the lengthier units toward the end, after using the shorter units to identify bugs, but I have not heard any detailed information about this. Lawrence |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Confirmed: After the SEWELL1A series, the latest series are constantly double the time on my machine from the previous series, in fact the last 5 varied from 4.5 to 5 hours, opposed to 2 to 2.5.
----------------------------------------Where we ran previously 10 to 12 CPU years daily, its now risen to 17 CPU years per day. As there is no 'No Work Available' messaging, it seems that the crunch power is not able to keep up with what the labs are sending in. http://www.worldcommunitygrid.org/stat/viewProject.do?projectShortName=hdc
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 1 times, last edit by Sekerob at Dec 5, 2006 3:16:16 PM] |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
To add, if u look at the project statistics page, u see that the running average is 4.34 hours Per Dec.04. On that day we had WU average of 6 hours exact.... almost 40% up.
----------------------------------------
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
Impx
Cruncher Joined: Sep 7, 2005 Post Count: 9 Status: Offline Project Badges:
|
Hi. My most recent 3 units are now taking 40 hours to process.
Did we increase the processing time again in the last week or so? How about some communication on these increase in processing times? I am thinking of stopping running these HDC units for 2 reasons. 1) No communication on changes to the project for like the 2nd or third time. (Thanks to those that posted a confirmation on the previous increase on process times, Muchly appreciated) 2) Large units that take 40+ hours to run. Unless units are required to be this size for the science or for the server backend its quite annoying when running 2-3 projects on Bionic. (Since no one tells us whats happening I have no idea why they are increasing processing times so much.) As I near the return time on the workunits for HDC I normally have to change the priority on this project to be higher so they finish in time. So I need to micro manage the projects. (And no my PC is not slow with a 3.4gig/2gig of ram. Even a FAAH unit only takes ~10 hours. Rosetta takes ~2 hours.) While I believe this is the most beneficial project around (my family has been effected by cancer) the project has to come to grips with the community and how it needs to envolve the community to keep them interested and the support high. If there is one thing I am disappointed by its the communication on distributed computing projects and honestly HDC seems to be lacking this more than others. This is not meant to be a flame just a communication of frustration from a avid supporter for your feedback to try to help you guys out with information on the project. While I havent stopped yet I will be watching for any improvements before I make a final decision. And if I do quit I will check back every now and then to see if there have been any improvements. Thanks for your time. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
There is a known issue affecting HDC right now. The details are hazy, but it is possible you are affected by it. The techs are working on it. We'll let you know what's going on as we find out.
As for the work unit size... it's supposed to work as a black box. You should never need to work about how long a particular unit takes. HDC has fairly predictable units, but some of the WCG projects vary enormously from unit to unit. So, while updates are important, and usually provide interesting, useful information, there is no actual guarantee that you will be told technical details like the size of upcoming work units. But if you're curious, please ask! WCG pass on queries like that to the scientists, and while a response may take a while, we'll get you an answer eventually. You are right, though - I haven't seen an HDC update for a while. WCG do remind the project scientists from time to time, but so long as the project runs smoothly and the scientists provide the updates they are required to, then that's enough. In all fairness, the scientists have done a good job of keeping us up to date when issues emerge, if they have been a bit tardy in providing routine updates. WCG are aware of this problem, but meanwhile we'll keep reminding them to remind the scientists to update you. ;-) |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
With HDC, disk swapping can add significantly to the processing time, an almost new 'personal' record keeper last night on BOiNC was 6:14 Hours, uninterrupted.... that's on a 2.53ghz ht 1.5gb ram machine, 24/7.
----------------------------------------If these get interrupted and have a checkpoint regression, than on a 550mhz machine with limited ram, they'd be easily getting to the times Impx is observing. On UD Agent it is likely longer because of the way this agent writes to disk.
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
|