________________________________________________

BOINC Manager - (possible) user interface bugs, wishes:

_____
1) Advanced view, "Disk" tab:
You can easily generate wrong graphical display of used/free/BOINC used
disk space by simply resizing the window of BOINC Manager
(it's harder to get almost true display).













_____
2) % display:
There is about 2% difference between "Progress %" display in BOINC Manager
(Simple and Advanced view) and BOINC Screen Saver "Overall % done".

I noticed this after power shutdown (no UPS at the moment) at about 50% done
but I last checked this (before power shutdown) at about 20%-30% - it was OK
(the same % in Manager & Screen Saver).

I tried to correct this by Suspending the two Tasks, exiting the BOINC Manager
(from the tray), restarting it (from start menu) and resuming the tasks
but with no success.



_____
3) Time estimation:
"Time to completion" estimation is highly inaccurate - almost twice.
At the beginning (0-10%) it was about 52 hours.
After 63h at 70% done it shows 23h "to completion" so it will take maybe 90 hours to 100%.
(63/70)*100 = 90h (and not 63+23 = 86h) [Yes, it took 90h]

After completing the first task the second task's total Time was corrected
from 3+78 = 81h total to 13+116 = 129h total.
My estimation is total of 105h = (12.83h/12.16%)*100
(maybe simple formula is better than complex)

The computer was almost idle all this time (it was "BOINC busy").
The BOINC Screen Saver blanks the screen after 5 minutes.
The Power Management sleeps only the Monitor, not the System or HDD.





_____
4) I manually started first the Task which has the closest Report Deadline.
Maybe you have to consider ordering the tasks by Report Deadline, not the Name.
(by the way - you can press the buttons Name, CPU time, Progress ... but with no effect)

_____
5) The "dangerous" buttons "Abort" (task) and "Abort Transfer" are too close to
"normal" buttons (Suspend/Resume, Retry Now) and can be used accidentally.
Maybe there is confirmation "behind" them but I don't have the nerve to try it!
(Why !? anyone will "Abort Transfer" of 10 - 20 KB result file or "Abort" 90% completed task ?)

_______
P.S.
I would like to know more about logic behind analyzing the data -
a few pages in popular science language will be good if you add it to help menu.
It can be also a .PDF file with pictures along the text to help for better understanding.
Explain spike, gaussian, triplet, pulse, chirp, fft, RA, Dec ...

Info offered in interesting way can bring more supporters to your program.

The readme says only "analyzes the data looking for signals of extraterrestrial origin".
I have to make guesses of what is going on by looking the Screen Saver's output.

You may say I have to look and understand the program sources but despite I studied
chemistry (implies - & some physics) and used to work as C programmer in the past
I don't know much of astronomy and I don't have time to dig too deep,
but I am still curious as many others.


Just for fun you can add small picture of part of the night sky (not very zoomed in)
showing the nearest stars/constellations (by name too) with every Task data.
Show (mark) also the point (with red dot/circle/haircross ?) of currently running Task.
This picture can be shown in Simple View, in Advanced view's new button (Show sky) or new tab,
and/or in Screen Saver's Data info (alternating 5-10 seconds with the text you use now).


------------------------




Host Image with PicTiger

Item #2. BOINC is using whatever data the project application provides in order to display the % complete. This information will lag what the application itself knows by a small amount.

Item #3. Many of the projects have non-linear % completed for their appli. i.e. 50% happens at 6 hours of a 10 hour run. Mostly this is because the run times are intractable.

The estimate of remaining run time is based on a weighted average between a calculation based on the original estimated run time and the % complete, and the CPU time already spent and the % complete. The calculation gives more weight to the second of these two as the % complete increases. Many of the projects have widely varying run times where the run time cannot be calculated in advance. Because of this, the original estimate is closer to the worst case - so that work has a higher chance of being returned on time.

Item #4. The CPU scheduler normally attempts to run work in round robin mode with tasks being run in the first in first run from any particular project. However, if there is a task that is in danger of being returned late by this method, the CPU scheduler switches to an Earliest Deadline First mode (the details of how it selects the task to run in this mode vary from version to version - the latest development versions use a modified form of EDF where projects with earlier deadlines can be ignored temporarily while a project that has a deadline problem gets attention - even if the task run has a later deadline than some work from another project).

[edit]

Item #5. I believe that S@H is looking for narrow band pulses, the sort of thing that would indicate the possible brosdcast of information.

AstroPulse is looking for wide band pulses.

---

BOINC WIKI