Topics

[1: Scanning the BOINC folders for files, check if they are running and throttle when needed.2: Check with the BOINC client what programs are running an throttle when needed.Method 2: Should mean less overhead. Presents only the program that need actual throttling. Solves the problem when both CPU and GPU work and both need throttling.Method 1: Is still available and is used as a backup. The BOINC tab check "connect with the BOINC client" is a way to fall back on method 1, when the check is removed.]

This version departs from the previous method, of acquiring the BOINC programs to throttle.

1: Scanning the BOINC folders for files, check if they are running and throttle when needed.
2: Check with the BOINC client what programs are running an throttle when needed.

Method 2: Should mean less overhead. Presents only the program that need actual throttling. Solves the problem when both CPU and GPU work and both need throttling.

Method 1: Is still available and is used as a backup. The BOINC tab check "connect with the BOINC client" is a way to fall back on method 1, when the check is removed.

Add: Read the information about running tasks,  directly from the BOINC client.
Change: Allow programs to be throttled on both CPU and GPU.  +_abp2cuda will include the exe in the CPU and GPU throttling.

Add: Send the temperature graph to BoincTasks.

This topic has been moved to Questions.

http://www.efmer.eu/forum_tt/index.php?topic=547.0

To setup a custom time format, select Extra->BoincTasks settings->View.
At Time Format select: user. In the box right of the selection, enter the time format string.

Some BoincTasks predefined formats:

24 hour: %d-%m-%Y %H:%M
System: %x %X

Accepted formats:

%a Abbreviated weekday name
%A Full weekday name
%b Abbreviated month name
%B Full month name
%c Date and time representation appropriate for locale
%d Day of month as decimal number (01 – 31)
%H Hour in 24-hour format (00 – 23)
%I Hour in 12-hour format (01 – 12)
%j Day of year as decimal number (001 – 366)
%m Month as decimal number (01 – 12)
%M Minute as decimal number (00 – 59)
%p Current locale's A.M./P.M. indicator for 12-hour clock
%S Second as decimal number (00 – 59)
%U Week of year as decimal number, with Sunday as first day of week (00 – 53)
%w Weekday as decimal number (0 – 6; Sunday is 0)
%W Week of year as decimal number, with Monday as first day of week (00 – 53)
%x Date representation for current locale
%X Time representation for current locale
%y Year without century, as decimal number (00 – 99)
%Y Year with century, as decimal number
%z, %Z Either the time-zone name or time zone abbreviation, depending on registry settings; no characters if time zone is unknown

Implemented as of BoincTasks version 0.78.

More info about BoincTasks settings dialog.

Is there anyone who has problems aborting tasks?

V 2.2

Changed: Only 4 GPU temperatures where send to BoincTasks, now the number is 6.
Add: Config.xml <GPU_SETUP> option to set the maximum number of ATI GPU's detected.
Add: Config.xml <GPU_SETUP> options to set the minimum and maximum regulation for the GPU.

V 2.10 1

Changed: Installer V 1.31, some text updated. Installer now runs as administrator only.

Not really a TThrottle update, but now the installer, can only be run as administrator.

Changed: Now tells BoincTasks how many NVIDIA and ATI cards are present, for correct temperature assignment.
Fixed: Some CPU are looping in the temperature detection. Don't use the detector on known CPU's. Take known default values for: 1/2 cores Intel = 1/2 sensors, 1/2 cores Amd  = 1/1 sensor.
Fixed: The ATI Gpu temperature was not read properly.

Changed: Updated NVIDIA libraries.
Changed: Drastically reduced TThrottle overhead, on some systems with NVIDIA cards and WIN 7.
Fixed: The Run time bar wasn't working properly.

As of V 0.70.

Checkpoint warning can be set to any time / project / application, by adding <checkpoint> entries to the config.xml file.

<project> & <application> may be a partial name, or leave it blank for a general match.
<seconds> checkpoint time as displayed in seconds.
<red><green><blue> warning color in RGB.

This example always shows the checkpoint cells green. Red when the checkpoint time goes past 1 hour.

Disabling on a specific project.

The last project collatz conjecture always shows green, in effect disabling showing warnings on this project.
Make sure you place the projects you want to disable last.

Sequence:
The checkpoint colors are check in sequence, the last one in the list matching the project / application, is the one that's used.

<config>
   <checkpoint>
      <project>SETI</project>
      <application>enhanced</application>
      <seconds>0</seconds>
      <red>0</red>
      <green>255</green>
      <blue>0</blue>
   </checkpoint>
   <checkpoint>
      <project>SETI</project>
      <application>enhanced</application>
      <seconds>3600</seconds>
      <red>255</red>
      <green>0</green>
      <blue>0</blue>
   </checkpoint>
   <checkpoint>
      <project>collatz%20conjecture</project>
      <application></application>
      <seconds>0</seconds>
      <red>0</red>
      <green>255</green>
      <blue>0</blue>
   </checkpoint>
</config>

Warning:

Spaces are removed from the xml file.
Use %20 instead of a space.
So WORLD COMMUNITY GRID should be WORLD%20COMMUNITY%20GRID

The following characters are NOT allowed to be present in a text: < / >.

%3C = <
%2F =  /
%3E =  >

Upper and lower case are not used so: "WORLD" is exactly the same as "world".

The config.xml file should be located in the same folder and the exe.

Changed: Translation updates.
Fixed: AMD processors logged one temperature core less, that actually present. This Processor has (4) 0 Cores with a temperature sensor. Should be 1 instead of 0.

I want to detect if a ATI or NVIDIA gpu is running a tasks.
So I'm only interested in GPU tasks.

1) NVIDIA do all tasks show + x.00 NVIDIA GPUs = xNV in the Use column. I mean on all projects, or are there projects that show something else.
2) ATI what do these tasks show in the Use column. Are they all the same, on all projects.

V 1.90

A lot of changes, deep in the program, so be careful.

Changed: Show NVIDIA and AMD/ATI cards in mixed systems.
Changed: The max number of GPU cards is 6.
Changed: Add __ati as default to the GPU list, when empty.
Changed: The mini window width is better calculated.
Changed: Revised driver V 2.0 that can handle more than 8 cores.
Changed: Adaptable temperature read from driver, reading only the cores needed.
Changed: The maximum number of temperature cores is 8, the maximum number of virtual cores is 16.
Fixed: RNA World puts play.exe in a slot. TThrottle added everything with play in it, like wmplayer.exe to the CPU list.

This is a pre release version.

Place the exe in an empty folder without any dll's, this is an English only version.

http://www.efmer.eu/download/boinc/tthrottle/TThrottle_184_pre.zip

Changes can be found here: http://www.efmer.eu/forum_tt/index.php?topic=50.0

[Warning this part is not 100% tested so be careful!]

Warning this part is not 100% tested so be careful!

The Astropulse example seems to be working correctly.

To rescheduling work from other BOINC projects, fill in the project url and application name.
The project must of course be set up to accept CPU and GPU work for that application.


For example, Astropulse for the ATI gpu.
As of V 2.6 there is a separate Seti Astropulse tab.

Application astropulse_v505, Version: 505
Application astropulse_v505, Version: 506 , Plan class: ati13ati

Go to the other tab and fill in the following:

Project url: "http://setiathome.berkeley.edu/"
Application: "astropulse_v505"

More information: http://www.efmer.eu/forum_tt/index.php?topic=428.0

For example, Einstein.

Make a copy of client_state.xml and search in the copy for:

<project>
<master_url>http://einstein.phys.uwm.edu/</master_url>

Find <app_version> directly thereafter.
Look for the right application, fill in that application in the rescheduler.

[Add rule]

Some examples

If project X with application Y is active for more than 3 hours an event is triggered.
If project X with application Y generates 10 errors within 10 minutes an event is triggered.
If any task has a progress of less than 0.01% within 10 minutes an event is triggered.
If project X with application Y runs below 56C for xx time it will trigger an event.
If 10 errors appear within xx time an event is triggered.
If computer X lost the connection for xx time an event is triggered. If the computer is reconnected within the xx time the rules isn't triggered.
If project Y on computer X has more work than 1 day set "No more work".
If project Y on computer X has less work than 12 hours set "Allow new work".



It is possible to add a rule from the Tasks view. Click on the task you want add a rule for and press the right mouse key.
Select Add rule
This fills in all the names relevant for this task.

Rule name: Must be unique name that describes the rule.
Computer: Optional name of the computer, leave this empty if the rule applies to all computers.
Project: Optional name of the project, leave this empty if the rule applies to all projects.
Application: Optional name of the application, leave this empty if the rule applies to all applications. As of V 1.28, wildcard may be used like t?s? = test, test* = test1whatever, test* = test2

* the above fields may hold a part of the name: Like 3.02, instead of the whole name.

Type:

Elapsed Time The time the task has been running since the first start.
Cpu% The percentage between wall clock the actual run time of the CPU.
Progress % The progress percentage of the task.
Time left The time the task has left.
Progress / min % As the Progress % but now the progress that has been made in one minute.
Use The number of CPU and GPU % that are planned for this task. E.g. 0.04 Cpu + 1 GPU or 0.025Cpu.
Temperature The CPU or GPU temperature. This rules must have a valid computer name.
Wallclock In a time interval. This rules must have a valid computer name.
Deadline Deadline =  (Deadline time - System time) - Time Left. So a deadline that is 1 day from now and time left of 2d gives -1d, that means it will go 1 day beyond the deadline. < -14d,00:00:00 triggers if 14 days or more past the deadline. < 4d,00:00:00, triggers if 4 days or less, before the deadline.
Time Left Project Cpu and Gpu time left on a specific project.

The 3 rules have to be ALL valid, at the same time for the event to trigger.
You may leave unused rules empty.

Time:

The time the rule has to be valid to produce an event. Blank means 0 seconds.
Don't use a short time, as this may result in a lot of false positives.

Event:

A rule may have one internal an one external event at the same time.
E.g. Suspend a task and start an external program.

A rule may have 0 events, so it only shows a color.

Run program Runs an external program or batch file.
The field below must hold the full name of the program. E.g. beep.bat
The folder is the default BoincTasks folder: C:\Users\user name\AppData\Roaming\eFMer\BoincTasks\

So beep.bat is the same as C:\Users\user name\AppData\Roaming\eFMer\BoincTasks\beep.bat

An ENTER (key) may be used to keep the text in the box.

You may add parameters: E.g. beep.bat, parameter
The "," is used to separate the program name from the parameter.

Suspend project

Suspends the project of the task that triggered the rule.

Suspend task

Suspends the task. When more tasks are triggered at the same time, only one is suspended. The other tasks are suspended, one every following second.

No more work

Sets the project to "No more work". E.g. Deadline < 2d,00:00:00 No more work.

Allow new work

Sets the project to "Allow new work". E.g. Deadline > 2d,00:00:00 Allow new work.

Snooze

Snoozes the entire computer for a time set in the right corner window, like 30:00. (30 minutes).
To allow the computer to run again, set a time of 00:00 (0 second), this will remove the snooze condition.

"show logging" check

When checked, BoincTasks will show the rules logging window, when an event is triggered.

Resetting a rule:

Rules are reset when you:

Edit any rule.
A tasks is no longer running.


Some inside information:

The rules are checked, using the history information.
This means, the rules are checked, at the history interval time.
CPU % and Progress / min are measured, over a period of 2 History cycles.

Progress / min:

Cycle 2: Progress 61   1 Progress / min
Cycle 4: Progress 64   3 Progress / min
Cycle 6: Progress 66   2 Progress / min
Cycle 8: Progress 66   0 Progress / min
Cycle 10: Progress 66   0 Progress / min

Rules can be used on a specific computer, or on all computers, unless stated otherwise.
Leaving a rule field empty, is the same as a all.
So a blank computer = all computers, a blank project = all projects, a blank program = all programs.

The rules below are specials.

Status rules:

Status rules can trigger on a number of tasks in a specific status.

The syntax for the value is: status, count 3,4
A rule can only have status types, all others types will be ignored.
A status rule must have a valid computer and project name. An Application name will be ignored.

Examples:

A rule Status > 3,4 Time 10:00 is active when there are 4 tasks in the error status in the last 10 minutes.
A rule Status = 5,0 Time 60:00 is active when there where no completed tasks the last 60 minutes.


Status codes:

3 Computation error
4 Uploading
5 Ready
6 Aborted


Wall clock rules:

As of V 1.52.


Wallclock time rules can:
Suspend and resume work fetching.
Suspend and resume specific projects.
Snooze and resume CPU and or GPU on a computer.

Select "Wallclock" in the Type column.

Click on the schedule and modify it.

The shift key lets you select a range.
Together with the Ctrl key a range can be deselected.

Connection rules:

You can setup a rule, that goes active, when a connection is lost or never made, after xx seconds.
The Project, Application and Value fields aren't uses.
A blank computer name means all computers.
=<> A lost connection
<> No connection

Connection =<> 00:05:00 (Type, Operator, Time) When a connection is lost for more than 5 minutes the rule triggers.

The only Event that can be used is "Run program" or "Show" logging.

Time left on a project rules:

Make sure you set a valid computer and a project, the application name isn't used.

E.g.
Time Left Project > 1d,00:00:00 will trigger if the project has more than 1 day work. Typically "No more work" is used as Event.
Time Left Project < 0d,12:00:00 will trigger if the project has less than 12 hours work. Typically "Allow new work" is used as Event.

Warning use a time of 2:00 on the  "Allow new work" rule, to make sure it isn't triggered at start up, when the time left is 0, for a short while.

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

The rules are stored in C:\Users\username\AppData\Roaming\eFMer\BoincTasks
File: rules.xml

itime = trigger time in seconds
ivalue0 if time than in seconds

Interesting reading:
http://boinc.berkeley.edu/trac/wiki/RuntimeEstimation

Yes you can. You can set a lower and upper limit on the duration correction factor and BoincRescheduler will adjust the factor, when it goes out of this range.

http://www.efmer.eu/forum_tt/index.php?topic=428.0 Read the config.xml part.

V 0.2

Add: Error text, instead of numbers, for the file copy errors.
Fixed: rsc_fpops_est and rsc_fpops_bound were calculated incorrectly.

Fixed: The VLAR and VHAR count after rescheduling wasn't always correct, after scheduling from the GPU to the CPU. This didn't effect the actual scheduling. (V 2.4)

Add: Astropulse tab.  (V 2.6  )
Changed: Optimized internal list handling -> quicker.  (V 2.6  )