Then how do we know how many credit a task gives?

The script should then be modified to insert the given credit in place of "pending" (or somewhere in the database from where it is copied to the web page). Otherwise there is no way to see how much credit each task was given.

The fact that only one result is needed for quorum shouldn't be a problem, for example at LHC@Home the vbox tasks are sent out only to one computer and credit is still granted normally. I don't know how they do it, but anyway it works.

Personally I would like to see that CPDN would move to smaller tasks and give credit the way that is built in BOINC. If the tasks were the size of one month (the size that creates one zip file) this maybe could be achieved. Of course this would mean higher volumes of downloads and uploads as the starting point for next months calculation should be uploaded to server to be sent to next computer which will continue to analyze the next month. Also it means that models take a longer time with the extra uploading and downloading but it could lure more crunchers to the project which could counter balance that. Maybe this method has already been studied and rejected, I don't know. Just my two cents.

---

It's on the individual pages for each model.
Although it used to also appear on the page listing all tasks. I guess this is all part of the new look.

There's a few things that I'm going to ask about, so I'll add that to the list.

Les Bayliss wrote:

The update can take a while to show on stats sites, as it's up to them to get the data and put it on their sites...

True, but new data has to be provided by the project. What's the date on the most recent export? I think I knew once where to look, but don't remember...sorry.

---

Having said that, I've found a couple of computers that have numbers in the Credit column, so this is probably a teething problem. I'll word the email a bit differently.

edit
Trickles that have been returned "recently", possibly December, have credit, those from earlier don't. So it's just a matter of refining the scripts to suit the new setup.

Harri wrote

Personally I would like to see that CPDN would move to smaller tasks and give credit the way that is built in BOINC.

It would as you say result in a large increase in the amount of bandwidth needed by the project as well as placing a greater load on those crunchers who don't have unlimited broadband. Another factor which is discussed in various threads is that because of differences in architecture, OS etc. the same tasks will produce slightly different results from different computers. Also it would greatly increase the chances of a task failing. I don't know what the overall fail rate is but it would be multiplied by a factor of 12 or thirteen typically for computer fails as opposed to tasks failing which will happen due to the task rather than the computer from time to time. (If a task may switch from a windows computer to a Linux one the fail rate may be much higher then.)

Personally, I like having tasks that typically last a few days as it needs less intervention from myself but then while I check them occasionally I don't really care about stats.

PS: One of the things included in Les's email to the project people was a request to check if the new URL to download the stats from has been given to the external sites.

I think that the simplest way to display credit would be to use the sample_trivial_validator at https://boinc.berkeley.edu/trac/wiki/ValidationIntro with the --no_credit command line argument in order to let the trickle system grant credit with no interference from the validator. This would mark units as validated, allowing the scripts to display the credit instead of "pending".

I have now credit showing on the tasks pages for a currently running wah2 task. All the already finished tasks are still shown as "pending".

---

I see that my completed tasks which previously showed credit now have pending in that column. I am sure it is all part of the teething problems to do with the upgrade of the site. An email has been sent to the project.

As discussed in numerous threads, this project doesn't actually use pending credit but gives credit on the basis of trickle up files sent so even if a task crashes it may get most of the credit which may or may not balance out the fact that this is not the highest yielding project in terms of credit/hour of cpu time.

Now fixed for all my older tasks but those that completed since the last credit script run are showing zero credit. Not sure what they were showing before they turned to pending? I expect all will be well following the next run of the script.

It's on the individual pages for each model.
Although it used to also appear on the page listing all tasks. I guess this is all part of the new look.

There's a few things that I'm going to ask about, so I'll add that to the list.

The table is looking better now, but the "state" row could do with tidying up.

Both validation pending and invalid seem to be seem to be valid and do we need validation inconclusive?

---

Kevin

I had all mention of Validate and Validation removed from the Tasks list last week.

Any mention of this word now, on individual model's pages, is a "valid" use. Sorry about the pun. :)
I think that the only use now, is when all of the data is back at the research centre, and a check is made to see if it looks OK, or is just garbage. (Which is probably when the checksums don't match.)

But even then credits will be given for trickle-up files received.

PS
If you mean something else, please say so. I don't always think beyond the obvious (to me), these days. :(

PS
If you mean something else, please say so. I don't always think beyond the obvious (to me), these days. :(

I was refuring to the selection filter lines above and below the results table.

State: All (637) · In progress (4) · Validation pending (413) · Validation inconclusive (0) · Valid (0) · Invalid (81) · Error (139)

It seems that both the validation pending filter and the invalid filter produce lists of completed work units, valid produces zero results.

If the pending and invalid results could be combined and renamed completed or valid it would look better.

It might also be handy for some if the "error" results could be split into 2 separate lists, those that failed before the first trickle and those that failed later in processing.

Sorry, the written word is not my best means of communication.

---

Kevin

Also I got no credit for my last 3 tasks which needed very long time and used 278,05 CPU sec. What is going wrong ?.
See here:
Task Arbeitspaket Computer Gesendet Meldezeit oder Ablaufdatum Erklärung Status Laufzeit(sek) CPU Zeit(sek) Punkte Anwendung
20092740 10802345 1189576 10 Dec 2016, 0:48:30 UTC 12 Dec 2016, 16:41:25 UTC Fertig 134,752.85 93,280.50 0.00 UK Met Office HadAM3P-HadRM3P Europe 7.28 windows_intelx86

20093558 10803162 1189576 10 Dec 2016, 0:48:30 UTC 15 Dec 2016, 8:58:20 UTC Fertig 166,307.43 92,924.31 0.00 UK Met Office HadAM3P-HadRM3P Europe v7.28 windows_intelx86

20096235 10805827 1189576 9 Dec 2016, 23:47:47 UTC 11 Dec 2016, 19:04:40 UTC Fertig 108,919.60 91,846.11 0.00 UK Met Office HadAM3P-HadRM3P Europe v7.28 windows_intelx86
Thanks for help
Dr.Mabuse

---

*** Since I'm a fool I prooved that the system is not foolproof ;-) ***

I suspect the credit will appear after the next credit run at the weekend. I have noticed that recently credit doesn't appear till after the first run of the script after the trickles. But there are going to be a few more oddities with the upgrade before it all gets ironed out.

Hi Kevin

Perhaps the "pending" one could be removed, but I'd rather not try to take it further.
There are various process that get run to check the reliability of the returned data, and interpreting those lists may require a knowledge of how BOINC works.
(And I didn't mention the "v" word once. :) )

What's left may need to be accepted, even if not understood. But who knows, some more tweaking more occur in the future.

What's left may need to be accepted, even if not understood. But who knows, some more tweaking more occur in the future.

OK

---

Kevin

It would seem that the credit script didn’t run this weekend.

---

Hi Jim

It seems to be running on Sundays these days. (6pm on the 12th)

I think, but unsure, that last week the evidence of the credit run showed up late Sun or mid Mon? .
I'll wait until 2017-02-14 [:)] before I worry about it (a tiny bit)
/heart/ CPDN

---