Messages

Ok I found out how to do this. Issue Solved.

Within BoincTasks I had an Account Manager: BoincStats selected.

Once I removed this I then had full control and could Detach from projects.

We have a CPU% so can we please have a GPU%

Reason: when doing other activities like gaming then you know how much GPU resource is being used by a Boinc GPU task.

Thanks

I have detached from projects before so am happy that this works fine.

I want to detach from GoofyxGrid but I command the detach and its still there listed on the project page.

Is there another way to remove a project or maybe there is a setting somewhere preventing this?

I suppose I could delete directories on my PC or maybe I need to close BoincTasks and try and remove it using Boinc Manager.

Update: After checking in Boinc Manager I see this project listed and also the "Remove" button is greyed out.

Hi,
when I am gaming the first GPU is mostly being used for that and would prefer to not have a Boinc GPU task running on it at the same time.

The only way I have found to do this is,

1. suspend pending GPU tasks

2. allow 2 GPU tasks to run and then suspend the one on (d0) and allow the one on (d1) to continue (d1 is set as PhysX GPU).

Maybe there is a way to control this in the BoincTasks settings and if so please point me in the right direction.

Thanks

Thanks for the reply Fred. I guessed as much but just wondered if I was missing something.

I will just manually manage some CPU load balancing to get these WUs running again.

My example for this is the WU's from climateprediction.net. Specifically sub project Weather at Home 2 (wah2)".

The deadline for these tasks is over 330 days. The runtime on my computer is shown at about 12 days and I have 2 of these WU's with 3 days each runtime left.

Both these tasks have been bumped into Waiting to Run. Without manually suspending other project work to free up 2 CPU cores is there a way to either set them into a state where they run and do not get bumped into the waiting queue.

I was thinking to be able to somehow flag them as "High Priority" but maybe you have other thoughts.

I did look at the rules setting but could not see anything there that seemed like it would do what I am asking.

1. yes i did use the instructions from your website.
2. I do not need to ping the laptop, it is already recognised by windows on my host PC "Mary"



There were no errors in the Toolbox on the laptop. I even enabled  the debug so more info could be shown.

I have been trying for about 2 days now but my host PC running BoincTasks just does not find my laptop.

Both use Windows 10 64bit are on my home network and in windows settings I selected private network and confirm discovery and file sharing is active.

I even used the following link to be sure the services and windows features needed were running/installed on both.

https://answers.microsoft.com/en-us/windows/forum/windows_10-networking/connecting-pcs-and-users-for-sharing/fb92e439-5dbc-4367-9857-1ef360e9ffdf

When I open a file explorer and choose network then I see both my PC and Laptop and correctly they have the Public folders. This is the same on PC and laptop.

On the Laptop TThrottle, Boinc Manager and Boinc Toolbox 64 is running with all 6 boxes are ticked as per your website instructions.

The laptop firewall is set so Boinc Client and Boinc Toolbox has access for private network.

On my PC there is no firewall entry for Boinc anything. Maybe this is the problem?

I had been running both independently and had downloaded BoincTasks onto both so the only part of your instruction I did not follow was to copy the Toolbox from my PC to my Laptop.

Also I left the Toolbox settings to use the default Dynamic IP.

Thanks Fred for the reply.

Isn't it easier to set the Project to one CPU task, in the project website setting

I will look for this setting on the project website but I have not seen it so far.

You should know I am not really sure what I am doing and by just running tasks without adjusting BoincTasks settings has probably added to my problem.

The way I am doing it now is to just run 1 of these Primegrid tasks on a GPU with nothing running on the second GPU. While this is very much a workaround and really does not make best use of the 2 separate GPU's it is all I can do atm.

I will however look into your suggestions even though I do not fully understand what I would need to do.

Can I set a rule to suspend a single task at the next checkpoint if the Throttle goes above a certain threshold?

I did look in the BoincTasks Settings, Rules but there did not seem to be an option to select based on Throttle value.

For instance I have 2 GPU tasks and when they both run the Throttle is 99% which makes it totally impractical to run them.

If I run one of them then the Throttle value is reduced to 50% which (for me) is ok.

Here is the example of the 2 tasks, thanks.



Here is the image showing when 1 (large) Primegrid task runs the Throttle is 50% (it is the same, 50% Throttle, when I choose either of the Primegrid tasks shown in the first image).



Is it perhaps the Throttle is not working correctly for this situation?

I have 2 Nvidia 980 cards so if any one of these 2 GPU task runs then the Throttle is 50%

But when I run both together then the Throttle goes up very quickly to 99%.

What you might be able to do is set a venue (in the project website setting) so that certain computers are handled differently on the project side.
Next add the computers to a specific venue, like work or home.

Thanks for the pointer I have all set generic atm.

I was considering say my PC with Home profile and choose a different one for my laptop.

I guess this just moved up my todo list

As it can't throttle the actual GPU it slows down the feeder CPU process.

Thanks Fred, I understand better now

Hi,
I believe my configuration is correct but the same throttle value is applied to both my GPU cards even though the temperatures are different. I was expecting the throttle values to be independent as in TThrottle I identified 2 cards by input of the 2 temperatures and it recognises the workload on both of them separately. See images.

Is there something I missed in the setup? Thanks.

Hi,

I had all GPU options ticked on the Collatz Project preferences and the CPU option was unchecked.

I found that my PC correctly received Nvidia GPU tasks "0.998C + 1NV" (only).

On my Laptop I receive AMD GPU tasks "0.971 + 1 AMD/ATI GPU" (correct) and Intel GPU tasks "0.829C + 1INT" (not correct).

Now I have unchecked the Intel GPU tasks option in project preferences so I should not receive the Intel tasks in the future.

I have adjusted some settings in BoincTasks App and also in the Boinc Manager App and put this in the questions thread as I am not sure if I created this situation by adjusting settings or if it may be a bug in the Boinc Manager or BoincTasks App.

One additional piece of info is when on my laptop an Intel GPU task runs as a solo GPU task then the BoincTasks temperature value is permanently ?? and the TThrottle on-screen temperature bar only shows CPU temps.

If it runs together with an AMD GPU task then both GPU tasks have the same GPU temp value and TThrottle on-screen temp bar shows both CPU and GPU temps.

If it helps then from the Computer tab here is my laptop info,



Regards.

Hi,
out of interest I clicked on BioncTask Settings, Tasks tab, Checkpoint to see what is would show.

Not only did it show the time from the last checkpoint but it also adds an additional option into the "right click when mouse over task" list and then you have the normal "Suspend" but also an additional "Suspend at Checkpoint".

https://efmer.com/boinctasks_settings/#tasks does not describe this

I see no need for more and more detail to be put into the help. This is a very well organised program (TThrottle too) and I like that most aspects are intuitive.

So I was thinking that without changing your design maybe you could always have the "Suspend at Checkpoint" in the Tasks right click list but when the "Checkpoint" column is not selected then this right click list item is just disabled (greyed out).

Users can then always see the option and should be smart enough to work out what is needed to enable it.

I just raise this as it potentially saves already processed data and with a save at checkpoint (in my case checkpoints are set to 60 seconds although I see in real time some go out to nearly 3 minutes) you are able to preserve your progress.

The only reason I see for suspending some WUs is that while I am using the PC  they are either super intensive CPU/GPU users or require TThrottle to ramp up on the throttle factor to maintain my settings for CPU/GPU temperature limits.

These suspended WUs will then be set to resume while I am not directly using my PC.

Cheers.