Managing Preferences[edit | edit source]
(For the BOINC PrimeGrid Applications)
Managing Preferences can be simple or complex, depending on how granular you want to get in managing your systems. Primegrid uses a “locations” metaphor in which you place systems. Each of those locations can have its own individual settings for what type of tasks are allowed, automatic prime reporting, and what projects are allowed (and on what hardware). These can also control computing preferences (such as CPU time and work cache).
The Basics:[edit | edit source]
- When you first create your PrimeGrid account, you will have a default location, and three others visible, “Home”, “School”, and “Work” (browse to “Your account ”, and “PrimeGrid preferences ” for project preferences and “Computing preferences ” for, well, computing preferences). There are ten additional locations (Sun, Mercury...Pluto) you can configure if you need additional granularity. These preferences permit you to distinguish between AMD and Nvidia GPUs, but not between individual models. So if you have one machine with a Titan Black that you want to run GFN22 exclusively on its GPU, and another machine with a GTX 770 that you wanted to run PPS-Sieve exclusively on its GPU, you would need two locations to accomplish that.
- You assign a computer to a location, and on its next update, it will download that location’s configuration, and proceed accordingly. (“Your account ”, and “Computers on this account ” will bring up the list of your computers which have successfully attached to PrimeGrid).
- Certain preferences are only available in the “Primary (default) preferences”, specifically whether PrimeGrid can email you, whether PrimeGrid can display your computers, and a default location for new computers attaching to PrimeGrid.
Other Considerations:[edit | edit source]
- You can’t assign a computer to a location until it has registered with PrimeGrid. You may want to take this into consideration when selecting projects for your default location (which is your default preferences unless you configure a specific location as default). You may want these to be short, such as PPSE or SGS for CPU and GFN15 or 16 for GPU. SoB and DYFL are often bad choices for default projects.
- The (primary) preferences for a given location govern on what hardware PrimeGrid’s applications will run. These override project preferences, so if you deselect “Use CPU”, computers will not fetch CPU tasks, even if you have CPU projects selected.
- If you want attribution for any primes you find, strongly consider setting prime reporting permission to yes and providing your real name. If you do not provide permission to report primes under your name, you will need to respond to any notifications you get. If this isn’t timely, a double-checker on that prime will get attribution. If you don’t permit email to you and permit automated prime reporting email, you will likely never know you found a prime, and will lose out on any attribution. PrimeGrid does not provide your email to anyone, including other PrimeGrid users. Your real name will appear with the notifications associated with any primes you find, and will be used in the event you find a prime which qualifies for the Top 5000 Primes database.
- If you don’t permit PrimeGrid to display your computers, it will make it difficult or impossible for community members to assist you with issues if you have any.
- Read the details that are available for each project when you are on the “Edit PrimeGrid preferences” page for a given location. In particular, the deadline for tasks for a particular project and recent average time (broken down by CPU and GPU where relevant) are important. So long as a computer is actively working on a workunit for a given task, and reporting back to PrimeGrid, the initial deadline can extend out to the maximum deadline. If you start a task, turn that computer off (or otherwise stop the BOINC client from running and reporting), and turn it on again after the initial deadline is passed, PrimeGrid will likely have sent that workunit out again, and which wastes computing effort (and electricity).
- Consider carefully your computers’ processing capabilities when choosing projects. Older (non-AVX capable) and slower CPUs are still useful, but may have trouble completing longer tasks by even the maximum deadline. This means additional unnecessary workunits being sent out and processed which wastes effort and electricity. Also consider that AMD’s implementation of AVX does not perform as well as Intel’s with PrimeGrid (and that AMD CPUs tend to have less floating point units as processor cores). Sieving or the smaller GFN units may be the best alternative for systems that struggle with calculating for other projects.
- Consider carefully your computers’ thermal dissipation capabilities when choosing projects. PrimeGrid projects will stress your computers’ hardware, and this in turn generates heat. Heat kills, and can reduce the lifespan of electronics, sometimes significantly. Some equipment is better able to dissipate heat than others. Systems such as iMacs, all-in-ones, NUCs, laptops, mobile devices and other small form factor systems tend to be far worse at heat dissipation. You can damage computers or their components running PrimeGrid, and are strongly discouraged from running it on systems which cannot handle it. You are strongly encouraged to install thermal monitoring tools on systems running PrimeGrid, thus ensuring that your hardware stays within its designed operating parameters.