Page 1 of 1
why still using tinker core
Posted: Fri Apr 17, 2009 6:20 pm
by shiryunaga
tinker core have 400 frame (i see from benchmark Fahmon) that make the PPD drop,
that coz error at Fahmon too, complete 4% but @ panel say 1%,
coz that different frame (400), why still using that core
,
Re: why still using tinker core
Posted: Fri Apr 17, 2009 6:31 pm
by 7im
They are not still using Tinker core. Tinker is the default name when adding new work to the Project Summary page. When the page is finished updating, the name changes. Tinker is also the default name when they are testing new types of fahcores that are not yet named, or named but not yet announced.
Re: why still using tinker core
Posted: Fri Apr 17, 2009 6:37 pm
by shiryunaga
ok i understand about 'tinker' name, but why that core have different frame(400), that make progress 4x longer
Re: why still using tinker core
Posted: Fri Apr 17, 2009 6:50 pm
by 7im
Generally, frames equate to how many steps in takes to reach 100% completion. It varies from one work unit to the next. Some take 100 steps, some take 100,000 steps. It only realates to how big a time frame is being processed, not the size or speed of the work unit.
Re: why still using tinker core
Posted: Fri Apr 17, 2009 7:06 pm
by shiryunaga
Code: Select all
Project : 5744
Core : Tinker
Frames : 400
Credit : 384
-- GPU --
Min. Time / Frame : 2mn 20s - 592.46 ppd
Avg. Time / Frame : 2mn 20s - 592.46 ppd
No Cur. Time / Frame
No R3F. Time / Frame
No Eff. Time / Frame
Code: Select all
Project : 5746
Core : Tinker
Frames : 100
Credit : 384
-- GPU --
Min. Time / Frame : 2mn 21s - 2353.02 ppd
Avg. Time / Frame : 2mn 21s - 2353.02 ppd
No Cur. Time / Frame
No R3F. Time / Frame
No Eff. Time / Frame
that frame make count of PPD drop,
i still don't understand that both WU has same credit(384) but why they have different frame
or maybe that just bug from Fahmon, i confuse
Re: why still using tinker core
Posted: Fri Apr 17, 2009 7:10 pm
by toTOW
That's a fahmon bug ... it probably had trouble parsing the psummary.
Try to update your data.
Re: why still using tinker core
Posted: Fri Apr 17, 2009 7:21 pm
by shiryunaga
how to update my data
Re: why still using tinker core
Posted: Fri Apr 17, 2009 7:23 pm
by toTOW
Tools > Download new projects.
Re: why still using tinker core
Posted: Fri Apr 17, 2009 7:59 pm
by shiryunaga
i use Tools > Download new projects but still that frame not change
first i still don't know how much that 'tinker' core frame, coz from
http://fah-web.stanford.edu/psummary.html that core don't have detail
IMO that core make error Fahmon, say that have 400(maybe that the real frame for that core), the panel from Fahmon not display the true progress (say 1% but the true is 4%)
Re: why still using tinker core
Posted: Sat Apr 18, 2009 1:55 am
by bapriebe
shiryunaga wrote:Code: Select all
Project : 5744
Core : Tinker
Frames : 400
Credit : 384
Sounds like your FAHMON is the one that is confused. Here's a benchmark from my FAHMON 2.3.2b:
Code: Select all
Project : 5744
Core : GPUv2 Gromacs
Frames : 100
Credit : 384
-- ADMIN2: GPU2 --
Min. Time / Frame : 2mn 40s - 2073.60 ppd
Avg. Time / Frame : 2mn 47s - 1986.68 ppd
No Cur. Time / Frame
No R3F. Time / Frame
No Eff. Time / Frame
Re: why still using tinker core
Posted: Sun Apr 19, 2009 5:25 am
by bruce
FahMon is confused, but that's because the data on the Project Summary page is confused. You'll notice that there is a column called "Frames" for each project. In fact, for project 5744. There are two entries and in one case, that column is currently blank, which is clearly wrong.
If the project is actually reporting 400 frames, then four frames must be completed to complete 1% of the WU. That used to be true for some projects for the original Tinker core. It might also be true for whatever core is actually be used by that project. (If 100 GROGPU2 is correct, it will be fixed when the extra data no longer appears in psummary.html.
In any case, FahMon knows how to deal with various numbers of Frames, . . . provided that the correct data is reported in psummary.html. I'll be sure that vvoelz is notified of the problem.
Re: why still using tinker core
Posted: Sun Apr 19, 2009 6:51 am
by shiryunaga
so the problem from Project Summary page ( i assume that Fahmon get error coz the ambiguous data from that page), can anyone from Pande Group give clarification about this issue, i just want to know that is really tinker core still use for this project or not, or just mistake list, Fahmon give me data Tinker and GPUv2 by turns from project that have the tinker core, make confuse everyone
Re: why still using tinker core
Posted: Sun Apr 19, 2009 5:52 pm
by 7im
I told you Tinker is no longer in use, it's a default name on PSummary when the correct data is not available.
toTOW is a forum
Moderator, and said the same thing, and added that the problem was Fahmon reading outdated info on PSummary, which backs up my answer.
Bruce is a Forum
Administrator, and explained this again in detail, confirming both toTOW and my statements.
There is NO need for
Pande Group to repeat this same answer again for a 4th time.
As a 5th confirmation, please read the Fahmon thread in the 3rd Party Tools section. This question has been asked before, and answered by the
Fahmon developer,
Uncle Fungus. BTW,
Uncle Fungus is also a Pande Group member.
Yes, it can be confusing, but now that you know, it isn't so confusing. Pande Group is aware if this issue, and tries to minimize its affect, but it will never be perfect. I think that about covers it.
Re: why still using tinker core
Posted: Sun Apr 19, 2009 6:12 pm
by shiryunaga
thx 7im for your answer, sorry if i make noob question
, so there are 1 core (GPUv2) for project 5744, the different name coz the ambiguous data from Project Summary page
, 400 frame maybe coz Fahmon count from frame old tinker core, so last question for this thread, why that Project Summary page not update, so everyone like me not confuse again
Re: why still using tinker core
Posted: Mon Apr 20, 2009 5:44 am
by 7im
The Psummary page is dynamically updated from the data on the work servers. If there is confusing data on Psummary, it is because there is conflicting data on the work servers.
Pande Group is working on 100s of projects, and they occasionally forget to update the list, or overlook the double entry made during a test, or are just too busy looking for cures to all these diseases and didn't have time to fix the problem. This is not the first time it has happened, and it won't be the last time either. Again, thank you for pointing out the problem, but it may or may not be fixed any time soon. Pande Group has a limited staff, and this is rather low on the priority list to fix. Sorry.