Messages

I have a low Nvidia card running GPUGRID tasks, and sometimes estimated time > deadline
I would like to have an option to check every e.g. 15 minutes, if estimated time > deadline + X hours, abort the task automatically
So that I do not do this manually, nor do I complete the task but not receive any credit.
This option could be added in the Warnings tab
Also an email could be sent via Cloud when this happens as a notification

Add extra columns:
GPU [%]
CPU [Freq]
GPU [Freq]

I am running GPUGRID and I do get the feeling that the GPU is not fully utilized.
I would prefer if I can remotely see the % and the actual freq on the GPU.
Also it would help in case of throttling to see the level of frequencies being used.

https://github.com/BOINC/boinc/issues/1699

Why not using nativeBOINC, as it is actively maintained?

From their main page:
NOTE: Development of this project was suspended over two years ago. Currently, the NativeBOINC doesn't works correctly on Android 5.0 or later. We recommend to use an official BOINC client for Android.

Also on github there is no activity since a few years

[True, it's the NativeBOINC. I've already installed also the official version, but not yet tried to monitor it. What's the problem there?

There is no way to access the data folder.
So no password and IP allowed to change.

Can you please head over to github, and give insights how would modifying boinc would make discovery on BOINCTasks possible/easier?
https://github.com/BOINC/boinc/issues/1699

Right now I am using Boray POWer (although old it is still workin on Windows 10) to achieve what I want.
I would have wanted to have an option in boinctasks, smth like.
When idle, activate Power Saver profile or High Performance.

The same was done with ParkControl, and since they were talking to opensource part of their code, maybe you could implement it in boinctasks

I would prefer to have a more precise control of how BOINC is using electricity

I would like to have the following option:

If following application are not running ...X
    If computer has been idle for ... Y seconds
         Activate power plan Z(if too difficult to select one, set default Power Saver)

If no longer idle, then return to previous power plan (e.g. Balanced)

If this is possible to implement, then I will put it on all my computers to run when they are really idling, since in the power plan I can set the Processor freq really low, to be just above the idle power draw. Basically it should consume 10-20% higher than idle, but at reduced freq (e.g. around 800MHz)

This could also solve the issue about people or companies complaining about high power bills

I am still reluctant to run boinc on my laptop due to noise, and if cpu/gpu gets throttle, some of the applications do not fully continue where they left off.

Can you not just bypass the boinc remote access?
For me is sufficient to know:
1. Which smartphones are connected to the same local wifi
2. What is the CPU% of all connected Smartphones on the same wifi network
If I know this info, I don't care which projects have deadline or whatever, since the official boinc app is doing a pretty good job of managing this stuff, plus I can also use bam to manage indirectly.

NativeBOINC is giving me the error "Unable to get access password" on Android 5.* and 6.*
On Android 4.* it is still working, but there were other issues with NativeBOINC.
So for me the only option is official app.

Could a simpler version be made for android devices?
I would be more interested in seeing if at least are online and what is the cpu utilization.
I already have more than 5 android devices, and it is kind of boring each day to check if they are still working or not.

If the GPU% setting is to difficult to implement, would the next possibilities be feasible?

1. Limit the GPU overall frequency to a certain value
           E.g 700Mhz instead of the stock 1000Mhz.
2. Limit the number of Compute Units to be used for the boinc project.
           E.g. Use only the last 4 out of a  total of 16 CU.
3. Combination of 1&2

I do not have 2 video cards inside my laptop, I have just 2 Compute Units .
I was thinking if it was possible to "assign" the CUs within a gpu card.
I think I will have to wait also for this option maybe a few more years .

I now that there is no boinc option, but I was thinking that you might find a workaround
My 6490M has 2 CU. What I would like to have do is allocate the first CU to the OS, and the others CUs for BOINC(in my case the 2nd CU).

original thread from boinc forums:
http://boinc.berkeley.edu/dev/forum_thread.php?id=7939

The problem I am facing with HCC GPU app, is that if I add them to exclude list, boinc will suspend them.
When it resumes then, all the progress is lost due to HCC lack of checkpointing for HCC GPU app.
But from what I see, instead of using the "exclude" option in boinc, if I use your tthrottle option with low GPU temps, it will not reset the progress, since it is "keeping alive, but throttles" the GPU WU.

Is it possbile to implement Leave Application in Memory for GPU tasks in BoincTasks?