I discovered SiDock yesterday, after looking for a project to participate in because World Community Grid is offline for maintenance for 2 months. I have however found that running this project is infeasible on my computers, because the option “Use at most xx% of CPU time” is not (well) observed.

On one of my computers (Intel Core i3-4030U, Win7 x64, BOINC v7.16.20, Use at most 50% of CPU time), when all 4 cores are used by SiDock it results in an almost continuous load of 100% without any real oscillation, see image:
Imagelink:

The difference between normal operation and using SiDock is clear when I snooze BOINC:

This issue is especially visible when using all available cores. If just 3 of the 4 cores are used by BOINC/SiDock, the picture is somewhat different:
In this case some oscillation can be observed, but it does not seem like it is a level of oscillation that corresponds with 50% CPU time.

On another, more powerful computer (Intel Core i7-7700, Win10 x64, BOINC v7.16.20, Use at most 40% of CPU time) just a very slight oscillation can be observed when all 8 cores are used. However, CPU usage is still too high for 40% of CPU time:
Again the difference with BOINC snoozing is big:

Also, when using 1 core less than the maximum (i.e. 7), the oscillation is much more present:

In all the examples above no other project was running other than SiDock.

It seems that when SiDock uses all cores something is wrong, when the option “Use at most xx% of CPU time” is set. I have never seen this with other projects, like World Community Grid or Einstein@home. With those projects the oscillation was clearly present and therefore heat generation was much less. For me this is a no-go at this time for this project. Unfortunately, because I really would like to run this project, that is why I took the time to sort this out and post this on the forum. Therefore I hope it is something that can be solved.

EDIT: unfortunately I apparently cannot insert images, even though I can use BBCode. Maybe someone can explain to me how to insert images so that my post becomes clear? Thank you in advance!

EDIT 2: Here are the links for the images:
1: https://pasteboard.co/Lwgz2ZSJiH2y.png
2: https://pasteboard.co/JIeE3qoP6tYS.png
3: https://pasteboard.co/97oAtaWUCs7k.png
4: https://pasteboard.co/7FhoHthK06VP.png
5: https://pasteboard.co/7uk7vghdWT0w.png
6: https://pasteboard.co/YArBvWdc32Go.png

Are you setting low CPU time because you have high temps?
Personally i prefer to set the maximum concurrent running task by reducing the % o CPU and leaving CPU time at 100%

For example on your i3-4030u (4t) you can set % of CPUs between 50 and 74% to run 2wus, 75 to 99% to run 3wu

I also suggest you to check your runnig background app, because on the I3 you're using 6.9GB of RAM while on the I7 is 5.8GB with doble of active wus

Hello! Regarding the images, you insert them correctly. The problem seems to be in the hosting you are using.

...
Personally i prefer to set the maximum concurrent running task by reducing the % o CPU and leaving CPU time at 100%
For example on your i3-4030u (4t) you can set % of CPUs between 50 and 74% to run 2wus, 75 to 99% to run 3wu
...

Yes, reducing number of concurrent tasks may be better solution than percent of CPU time.

...
Personally i prefer to set the maximum concurrent running task by reducing the % o CPU and leaving CPU time at 100%
For example on your i3-4030u (4t) you can set % of CPUs between 50 and 74% to run 2wus, 75 to 99% to run 3wu
...

Yes, reducing number of concurrent tasks may be better solution than percent of CPU time.

Indeed - this is not a Sidock problem, it happens with many projects. Boinc put this option in which doesn't work, or relies on something most projects don't use.

Are you setting low CPU time because you have high temps?
Personally i prefer to set the maximum concurrent running task by reducing the % o CPU and leaving CPU time at 100%

For example on your i3-4030u (4t) you can set % of CPUs between 50 and 74% to run 2wus, 75 to 99% to run 3wu

I also suggest you to check your runnig background app, because on the I3 you're using 6.9GB of RAM while on the I7 is 5.8GB with doble of active wus

For high temps, TThrottle is excellent. You simply set the max temperature you desire, and Boinc will be slowed down to not exceed it. Works with any project, even virtual box ones.

...
Personally i prefer to set the maximum concurrent running task by reducing the % o CPU and leaving CPU time at 100%
For example on your i3-4030u (4t) you can set % of CPUs between 50 and 74% to run 2wus, 75 to 99% to run 3wu
...

Yes, reducing number of concurrent tasks may be better solution than percent of CPU time.

Indeed - this is not a Sidock problem, it happens with many projects. Boinc put this option in which doesn't work, or relies on something most projects don't use.

The new total 64bit version of Boinc was supposed to have better control of that kind of stuff but with the Developers being 100% volunteers now due to their loss of funding, Boinc proved it's original grant funding theorem many years ago, the Developers are working to keep things running not implementing any new features. The existing version of Boinc is a hodgepodge of both 32 and 64 bit stuff, one reason it can't see more than 4gb of ram on a gpu, but it does work for the most part.

The new total 64bit version of Boinc was supposed to have better control of that kind of stuff but with the Developers being 100% volunteers now due to their loss of funding, Boinc proved it's original grant funding theorem many years ago, the Developers are working to keep things running not implementing any new features. The existing version of Boinc is a hodgepodge of both 32 and 64 bit stuff, one reason it can't see more than 4gb of ram on a gpu, but it does work for the most part.

Who was funding Boinc development and why did they stop?

Thank you all for responding. I have indeed discovered that this issue is not just related to SiDock. I also tried Universe@home, which apparently has the same issue. I thought it was just SiDock because the other projects I participate(d) in never had this issue.

I got my information from the BOINC explanation of preferences: https://boinc.berkeley.edu/wiki/Preferences where “Use at most N % of the CPUs” is linked to keeping some CPUs free for other applications, whereas “Use at most N % CPU time” is linked to heat and energy usage.

I have changed my settings to using less cores, and at the same time increased CPU-time to 100%. It does run better now. I will try whether I should reduce this 100% a little bit, because occasionally my computer did not respond immediately after I was away, when BOINC pauses its work. Besides heat that is another reason for having had CPU-time set at a lower value than 100%.

I will also look into Tthrottle.

Thanks again for your help.

P.S. I tried to post this reply several times last weekend, but was unable to, because my browsers several times and at various points showed the message that the website took to long to respond.

I wish it wasn't % of CPUs. I have 24 cores. And I can never remember if it rounds up or down. The number of cores I want is often a recurring decimal when expressed as a percentage.

I have mine running all the time, even when I'm using it. I just have Boinc stop when I play games. Nothing else experiences a problem. The only exception is Virtualbox tasks, too many of those and an AMD gets very sluggish, so I just limit how many of those run.

I have the same problem. I have a core i7 with 14 cores that are hyperthreaded. I keep marching the allowed number of CPUs lower and lower and lower. I am currently at 40% of cpus used, and this project is still spiking to 100% CPU usage. I believe that the process is creating multiple threads, each of which may take an entire CPU.

And, no, it is not because of high CPU temperatures, it is because my mouse freezes up.

I have verified it is this project. I set it to no new work, and the problem is gone once the work for this project was completed.

I will have to leave this project set to no new work until such a time as the problem is fixed. I need to be able to use my computer.

The project processes should be running in low priority so i don't know why it would cause your Intel Alder lake system to bog down. You indicated temperature is not a problem that it is not throttling so i stumped.
Could be new architecture issues still needing to be patched.

I have this project running on 2nd gen thru 8th gen Intels using all cores 100% of the time and they all stay very responsive.

I believe that the process is creating multiple threads, each of which may take an entire CPU.

This is not true at all.
Here you can see how much % of CPU each SiDock WU get.
3% = 1 of 32 thread CPU (16 core x SMT)

I will have to leave this project set to no new work until such a time as the problem is fixed. I need to be able to use my computer.

There are no any issues in general with this project.
I also crunching this project on many systems: from Nehalem arch.(Intel Xeon X5550 / 2009 y.) to Zen+ arch. (AMD Ryzen Threadripper X2950 / 2018 y.) without any issues.

Your case is some kind of rare confluence of some factors.
For example:
- CPU - problems with the new architecture (big.LITTLE) Intel have already been noticed in some situations/applications.
- OS - very raw Windows 11
- Hardware problems (in combination) with other devices
& etc

So if you really want to help this project (instead of just slam the door and leave with discontent) - give a full description of hardware components of your computer, OS version and installed drivers.

Windows 11 Home version 21H2 build 22000.856 Windows Feature Experience Pack 1000.22000.856.0
Processor: 12th Gen Intel(R) Core(TM) i7-12700H 2.30 GHz
RAM: 16.0 GB (15.7 GB usable)
System Type: 64-bit operating system, x64-based processor
There are 20 instances of the driver for CPUs listed. So, 20 cores?

There are dozens of drivers. Anything in particular you are looking for?

There are 20 instances of the driver for CPUs listed. So, 20 cores?

20 vCPU (Virtual), because Hyper-Threading also add "cores", but they are no "real" cores.
And news Intel CPU now have 2 class cores: "Performance-cores" and "Efficient-cores".
And HT supported only on "Perf" cores.
So: (6p x 2) + 8e = 20 virtual cores (e.g. treads).

Judging by your low power/mobile CPU, that is laptop, right?

Such CPUs, as well as laptops of such class(not gaming) in general are not designed for constant / long-term high load.
Accordingly, the cooling systems(in such laptops) are not designed for efficient cooling of CPU under heavy load ("Turbo Boost" mode).
Based on its Thermal Design Power(TDP) and current temperature, such processors under prolonged/high load will quickly lower frequencies of all cores.
I can assume that this is what causes the problems with responsiveness of your system.
I recommend you trying to reduce the number of threads available BOINC, to 6 (according to the number of real Performance class cores in this CPU).
"Options" >> "Computing preferences" >> "Usage limits" >> "Use at most 30% of the CPUs"

You should also (at least for the first time) carefully monitor the temperature of all cores, the CPU as a whole, as well as the current frequency of cores (which will show how much and how often the processor uses throttling under such a load).

Low end gaming laptop.

Let's recap what I saw in task manager. With BOINC CPU usage set at 30%, the CPU usage was spiking at 100%, BOINC would then suspend all the BOINC processes. The CPU usage would drop to less than 10%. BOINC would then unsuspend all the BOINC processes. The CPU usage would then spike to 100%. Rinse, repeat.

Removing this project from the mix solved the problem. It may be an interaction between this project and windows, but in the end, it is this project that was causing the problem.

I will help fix it, but blaming it on this hardware is a non-starter for me.

Hi jm7!

BOINC has two settings:
- Use at most n % of processors
- Use at most m % of CPU time

Which option you use?

Both were set. Use at most 30% of CPUs and 80% of CPU time if I recall correctly.

Task manager was reporting 100% of all CPUs used. which is what led me to believe that some project was starting more than 1 thread somehow.