Topics

I noticed during the discussion of glenaxl's Shift+click bug: you said

select the first item, press shift and select the last,

which of course is exactly right  The thing is, I've noticed that on the Task, Transfers, Messages or History tabs, you can block select rows either from the top down or bottom up.  But on the Projects tab you can only select a block of rows from the top down--bottom up will not work.  But on any tab you can Ctrl+click (extended select) in either direction.

I'm posing this as a question rather than a bug, because I'm just curious--what is different about the Projects list from the other tabs that makes it behave this way?  I've always assumed that BT uses a simple spreadsheet- or flat file database-type table structure, to allow for the various datatypes you have to display and manipulate.  (Don't feel obligated to disclose any "trade secrets")

Best wishes.

Better late than never, regarding the fix in v0.48:

(Changed) Tasks tab: When status is downloading and % = 100, then status is download error with the error color.

I finally got another download failure and captured the error messages

4/13/2010 05:13:12   rosetta@home   [error] expected 99a107259744c9b3fcf681adc56a0101, got 5a94ffcc45843810067b753d9e05fdff
4/13/2010 05:13:12   rosetta@home   [error] Checksum or signature error for rb_04_12_48_119_t0350_d1_aat000_03_05.200_v1_3.gz

The Status displayed in BOINCMgr was "Download failed".  BT correctly displayed this task in red (error color) with Status "Download error".

This is just an advisory for BT users with NVIDIA GPU hosts.

GPUGRID is betatesting a new ACEMD application for Windows (v6.22), the first I've seen compiled using CUDA 3.0.  The buzz on their forum is that many WUs are crashing on various GPUs; a few are getting through to completion.  The only beta WU I've gotten so far crashed immediately (actually 2.39 sec ET), after crashing on 3 other hosts before me  I've seen nothing to indicate whether CUDA 3.0 is a problem, or whether BOINC 6.10.x has a problem with it.  GPUGRID's new code may have issues not related to the new CUDA release

In any case I'm watching the other GPU projects I crunch (Einstein, PrimeGrid, Seti, Seti Beta) for any indication they will be upgrading to a CUDA 3.0-based application.  Eternal vigilance

Request you add the BOINC website homepage to the BT Help pulldown menu.

Regarding:

(Fixed) Some projects with low CPU use the CPU% was not correct. (FreeHAL, v0.46 PrimeGrid's AP26 Search (cuda23) v1.01)

I can confirm that with Primegrid's AP26 Search (cuda23) v1.01, BT 0.47 does indeed show CPU % during the run in the range 0.05x~0.07x%   There is still no () CPU time in the ET field.  But that's understandable since you appear to be set up to display time as xxd,HH:MM:SS, and not fractions of a second.  This is also consistent with BOINCManager, which doesn't display fractional CPU time either.  They probably never thought it would be necessary.

Once again Fred, thanks for the attention to detail (and while on vacation to boot)

(Changed) CPU%: if CPU % < 1% add 1 digit, like 0.001 and round of properly. 0.000 should be shown as empty.

Sorry, but in v0.46 PrimeGrid's AP26 Search (cuda23) v1.01 is still showing no CPU activity (CPU % = null field, no CPU time in parentheses) through its entire run   According to PrimeGrid's database, my most recent result reported an actual CPU time of 0.64sec on a run time of 628.64sec, or 0.102%.  That's on an Intel Q9400 at 2.66GHZ, with an EVGA GTS250 moderately overclocked.

This still isn't a big deal for me, but I predict that as other projects optimize their CUDA (& possibly ATI) apps, you'll be seeing more tiny CPU:GPU ratios like this (depending, of course, on what kind of data they are crunching).

Best wishes.

• When you implement the keyboard shortcuts already on The List:
  

  - Extra selection options with the keyboard, up/down, shift up/down, Ctlr A, Shift Home, Shift End, Page up, Page dwn.

  request you let spawned windows (e.g., Abort confirmation, Project properties) respond to the Esc key as a cancel/close window command.
  


• When you implement the History Tab filters already on The List:
  

  - History view: Filters for specific time periods.
  - History view: Filters for specific states.
  

  request you add a filter for specific project (i.e., show project on the highlighted row only).

Best wishes.

I've noticed that PrimeGrid's AP26 Search (cuda23) application (their only GPU app), through its entire run, consistently shows a 'CPU %' of 0.00 with the 'CPU % Long time average' option unchecked, and a null field with it checked.

Now the latest version of AP26/cuda23, (v1.01, deployed 26Feb2010), has a really low CPU:GPU ratio.  According to my Task history on a quad core/GTS250 host, <=0.5 CPUsec on an Elapsed Time ~500 sec is typical.  (Which is only ~0.1% or less, but still nonzero.)  Their previous version (v1.00) had a more typical CPU load of ~10%.

Does BT literally round low CPU usage down (as indicated by CPU %=[0.00] in this case)?  Or is this possibly a side effect of the change mentioned in the release notes for BT v0.43:

-(FIXED) Task view: Sometimes with progress below 1% the progress was not shown properly. Now the alternative way is only used when the progress % is below 0.01%. This is only uses in the first phase of GPU Tasks.

?  I'm posting this as a question rather than a bug, since either way it's not the end of the world  However, it might not be a bad idea for the 'CPU %--Long time average' to at least show 0.00 consistent with the short time option, rather than a null field. (Just a thought )

Best wishes.

I can confirm that the Tasks tab is much cleaner & easier to interpret without the alternating stripes...bravo to whoever suggested that  Thanks for implementing it.

Best wishes.

When tasks in state "Waiting to run" are filtered, it appears that the Elapsed Time & Progress % displayed are the lowest among members of the group.  Is this intentional?  It seems more informative to show the highest ET & Prog% of a Waiting group rather than the lowest

1) The ability to call up a Task property sheet similar to your Project property sheet (two very handy features of BOINCMgr).
2) The ability to snooze the remotes in addition to the localhost. (Maybe a rt-click context menu on the sidebar computer tree?) Possibly including a "Snooze All" option? 

Maybe that's actually three requests.  Best wishes

After several weeks running fine I noticed that BoincTasks stopped displaying core temps for one remote host.  Checking the local Tthrottle control panel showed that core temps were blank, & the OSD read 0.  Restarting Tthrottle had no effect, nor did rebooting the machine.  Removing & reinstalling Tthrottle restored normal function.

Tthrottle v1.73, running on Vista SP2 32bit.  The CPU is Intel T5450 Core2 Duo.

Is it possible to get the combine filter to not group seti@home & seti@home beta tasks together?  Or are we stuck as long as both projects are using the same application & rev. number?

I'm sure this is a standard feature of whatever tool you're using to build the tables, but could you add an option to turn off the alternating-row gray stripes?  They're helpful in a one-color table as in BOINC Manager.  But they just add to the confusion in the Task table where there are so many possible states and you're trying to use colors & shades to sort out which row is in what state

Thanks and Best Wishes.