Focusing on one project should not actually be more beneficial?

in #gridcoin7 years ago (edited)

People often say that focusing on one (CPU or GPU) project is most profitable. According to how I understand the calculation of MAG, this should not be the case.

The magnitude for a project is calculated as

MAG = ((Researcher_RAC/Team_RAC)/Number_Whitelist_Projects) * 115000

Number of whitelisted projects and the multiplier 115 000 can be defined to be constants, so the only deciding factor is the proportion of your RAC to the Team RAC. Of course one needs to notice here, that Team RAC also includes your own RAC. So we have

Your_RAC/(RAC of others in team + Your RAC)

For easier denomination, lets use

W = your own RAC
A = RAC of others in the team.

Ultimately if your own RAC is very small compared to the team RAC (W<<A), you can approximate

A+W ≈ A

So your mag is proportional to

W/(A+W) = W/A.

This mean that if you multiply your RAC by 8 (e.g. number of cores in your CPU), your MAG is also multiplied by 8. Easy peasy, right? In this situation the rewards grow linearly, so it doesnt really matter if you focus on the project or not.
But lets look at the situation when you are a large contributor for the team in the project. Lets assume that your one core produces one tenth of the RAC of others in the team, i.e. W=0.1A
This leads to your mag to be proportional to

0.1/(1+0.1)=0.0909… ≈ 0.091 = 9.1%

So what happens if you use all 8 cores on one project?

0.8/(1+0.8)= 0.444… ≈ 0.444 = 44.4%

If we divide the bigger contribution by the smaller one, we get

0.4444/0.090909 = 4.888…≈ 4.9 = 490%

So, by using 8 times the cores, we see only a 4.9 times increase in magnitude. This is because the mag for a project is capped, you are essentially competing for magnitude against yourself. Of course if an other projects has much higher Team_RAC, it might not be worthwhile to change.

Mathematically, focusing on one project should not be benefical.

So what gives? Why do people say that you should focus on one project for max benefit? Honestly, I don’t know the exact answer. Either people just feel that they are getting more by focusing (bigger single mag number to look at and confused by the effects varying RAC of yourself and team), or there is something in RAC or decay calculations that gives an exponential growth to your RAC based on credits done.

If someone can point me where I've gone wrong, please tell me in comments or in a PM. I've been confused about this for a while and would appreciate some clarification.

If on the other hand my calculations are correct, people should be able to mix and match projects that are close in Mag/work, assuming your work units dont get old because of it. You should still keep CPU and GPU projects separate, to not waste your CPU computing power. Bigger players with huge mining rigs could at times benefit a lot from mix & matching. Of course the amount of GRC per day is constant, so all these converge to the same total in the end. But I think many people would like the freedom to contribute to and try out multiple projects.

Sort:  

"Why do people say that you should focus on one project for max benefit"

its the safest and easiest path/advice to give. might not be the most optimal. not everyone has a rig like dutch

I've also heard on multiple occasions that focusing on one project gives the best mag and GRC. Maybe people have mixed it up somehow?

I admit giving that advise. but in reality. I don't practice it.

Its the best advice you can give for people with a few computers and also for new comers.

Having a few projects ensures that your rigs have something to do if a source of work units dry up or are in maintenance.

I think fortex is correct in saying, "This is because the mag for a project is capped, you are essentially competing for magnitude against yourself." I'm new to Gridcoin, but I thought about this too. If someone has several powerful rigs and they put all of their resources into one project, their own RAC begins to work against them in the magnitude calculation.

One other observation regarding spreading around your computational resources. Unless you're spending a lot of time tracking who is coming into and out of a project and your RAC relative to everyone else, it would be better to have more than one project. That way if someone with a lot of RAC potential jumps into one of your projects (and you didn't notice), you aren't as negatively impacted because you've got other projects to sustain your overall account magnitude. It's akin to not putting all of your money into a single stock unless you're closely watching it each day. It's better to spread around your risk with multiple stocks (or projects in gridcoin) and play for the best average return.

As someone new to Bionc/GridCoin, thank you for the great information.

Generally your observation is correct. It shouldn't matter if you work multiple projects. However, not all tasks are well coded. On certain projects the check-pointing is not handled correctly.

When the manager decides to switch projects it pauses the task it is currently working on. When you resume that task, you go back to your most recent checkpoint. You can lose progress on your task when this happens, if the project doesn't write checkpoints as often as they are supposed to (YAFU and probably others).

The best setup for picking projects is to choose two. A primary at 100% resource share and a backup at 0% resource share. With this setup you run the primary constantly unless it runs out of tasks, then it switches to the backup.

Congratulations @fortex! You received a personal award!

Happy Birthday! - You are on the Steem blockchain for 1 year!

Click here to view your Board

Support SteemitBoard's project! Vote for its witness and get one more award!

Congratulations @fortex! You received a personal award!

Happy Birthday! - You are on the Steem blockchain for 2 years!

You can view your badges on your Steem Board and compare to others on the Steem Ranking

Vote for @Steemitboard as a witness to get one more award and increased upvotes!