going @home

No, I'm not leaving, but there are these distributed @home projects where you donate computer power to some purpose. Among the most well known of these are SETI@home looking for extra-terrestials, Rosetta@home and Folding@home which work on 3d folding of proteins. And there's LHC@home which among others tasks simulate proton collisions at the LHC. And many, many others.

Thousands of people take part in these projects, resulting in impressing processing capacity. Folding@home, for instance, is one of the fastest computer systems in the world with speeds approaching 100 petaFlops (june 2019).

I consider joining one or two of them, using my Linux computer. Does any of you have experience with any of them? Most projects run on the BOINC platform, but Folding@home does not. Does that matter?

pibbur who many years ago, before he was pibbur, contributed to SETI@home.

OK, now I have some experience with these things. My Linux machine decided to crash, so I've temporarily installed the software (BOINC+project specific SW) on my Win10 floortop machine. I've set it to run all the time, but to suspend activity whenever CPU use for non-BOINC tasks exceed 25%.

I've restricted BOINC to only use 4 cores, and 75% CPU time. The latter setting forces the projects to take a break every 4th second to prevent excessive heat. At these settings CPU temperature varies between 60 and 65 degrees Celsius. According to WWW that should be quite safe. Any comments?

Currently I'm running jobs for the LHC and the World Community Grid (engaged in several medical research projects). For some reason the LHC seems to claim a much larger part of the resources I've made available than WCG, most of the time I see only LHC jobs running. I have to find out what I can do about that.

EDIT: It doesn't create much more heat if I let it run 100% on 4 cores, so I updated my settings accordingly.

pibbur who finally thinks he's actually contributing to something useful.

I tried the folding one for a while with my old computer and it was burning it so I was forced to stop.

Yes. Which is why careful editing of options is needed.

ATM, I've set up my Liinux machine to run continuously, on 2 hyperthreaded cores out of 8 (Intel I7 49-something) limited to 70% CPU-time, and to run on GPU (2 Nvidia 980's) as needed. Keeps the temperature below 70. I had expected that I could run it at higher settings (CPU is water-cooled). I will run this 24/7, except when leaving the premises.

On my main floortop Win10 machine I've set it to run occasionally, whenever I choose to, on 4 hyperthreaded cores out of 16 (Ryzen 2700x), and to only involve GPU when I'm not using the computer. Unlike the Linux machine I will turn this one off at night.

Currently I'm running the following projects:

Linux machine:

1. World Community Grid (healthcare projects).
2. Einstein@home (looking for pulsars and gravitation waves.

Win10 machine:

1. Rosetta@home (protein folding)
2. Climateprediction.net (Climate models).

pibbur who runs himself most of the time

I avoid these things - too much concern over security.

Originally Posted by you
I avoid these things - too much concern over security.

That is of course a valid point. However, at the Berkely site they claim - which you may choose to believe or not - that there hasn't been any serious security breaks, and that they have implemented several steps to ensure that things can be trusted.

Additionally, I think that the work is important, so I'm willing to take the risk.

pibbur who nonetheless will, based on @your input, examine security aspects more closely.

Oh, I've been doing SETI@home for… jeez…. about 25 years now maybe? Back at the start, it took a week to do a single job. They had a progress bar that was amazing! I've been doing Milkyway@home for a long time, too.

I did GPUGRID for a while. Those used the NVIDIA card instead of the CPU but the jobs were really big and didn't have a lot of time to do them - but I got some done. They don't seem to be generating work anymore.

For stats I just set it to use 75% CPU. I used to have it start with my PC startup then close it whenever I was going to start playing a game but lately I just start it up when I think about it.

I wonder if running the seti@home software in a VM is safer - i know there was a bug in VM a while ago (at the chip level I think) that allowed the VM to be escaped but perhaps it has been patched.

Another report.

It turned out that GPU computing did not work on my Linux machine (which I have called sinux), the boinc-client couldn't detect my 980 as a "usable GPU". First of all, I had to install NVidia drivers, so I had to find out how to do that. Boinc still didn't detect my 980. I discovered other components was needed as well. I'm working on that. Fortunately there are good (but not necessarily systematic) web resources.

Part of this is due to insufficient Linux knowledge, but I have to say: It's much easier under Windows, where everything worked out of the box. I certainly can understand that people find Linux confusing and frustrating. I do not. BECAUSE THIS IS FUN!!!!!

pibbux who considers changing his name.

Originally Posted by you
I wonder if running the seti@home software in a VM is safer - i know there was a bug in VM a while ago (at the chip level I think) that allowed the VM to be escaped but perhaps it has been patched.

I've read a bit about that, and the general opinion is that using the Oracle VM box is safer. LHC@home use it a lot. And there are installation files combining boinc and that VM thing.

pibbur who

An update.

These are the projects I run now:

Linux machine, running 24/7.
This one is equipped with two 980's and I run projects taking advantage of GPU computing. I've enabled 2 CPU cores and both GPU's for these projects, and the jobs run very fast.

• Einstein@home: Looking for pulsars
• Primegrid@home: Looking for laaarge prime numbers. This project has the most demanding jobs and some of them take a very long time - one job will finish in 3 days if running continuously (which it won't, as active projects switch every 60 minutes).

Windows 10 machine. running approximately 12/7.
Six (of 16 hyperthread cores) made available, no GPU, therefore always CPU-only tasks. The jobs suspend themselves whenever unrelated CPU use exceeds 15%, so ordinary use of the machine is not affected.

• World Community Grid: Several medical projects.
• Einstein@home: Looking for gravity waves, connected to the LIGO project.

I have several times tried running jobs for the LHC, but these are a bit complicated, and quite often the jobs fail. Too bad.

pibbur who thinks that this selfless charity work may have enough WAF (and PAF) to justify (excuse) buying a Ryzen 3900x (doesn't need new MB/RAM). And liquid cooling.

Guess I'll have to find the aliens myself, then. Note I am totally taking credit if they are found. (And, if they decide to destroy us all, we'll be too dead to blame anybody. I can only win! ;)

Originally Posted by Zloth
Guess I'll have to find the aliens myself, then. Note I am totally taking credit if they are found. (And, if they decide to destroy us all, we'll be too dead to blame anybody. I can only win!

And if they decide to destroy only some of us, I'm coming for you!!! (https://www.youtube.com/watch?v=zfvTZF2ZBfo).

BTW, I've switched PrimeGrid for LHC. CPU ran a bit too hot while processing primes and I finally found out what was the problem with LHC jobs failing.

pibbur who loves particles, especially the charming ones (and those that are strange).

PS. I have a CPU question, but I will post that in another thread. DS

I decided to run some SETI@home too. To get things done. The performance of a 'puter operated by a sloth must of course leave a lot to be desired.

pibbur who compares himself to a cheetah. An old cheetah, obviously, but still a cheetah. Or a raptor.

Turtles and hares, sloths and who's… same story.

Originally Posted by Zloth
Turtles and hares, sloths and who's… same story.

I didn't say I (he) was as fast as a cheetah, only that I he (I) compared myself (himself) to one.

pibbur who observes that a cheetah would run the 100m sprint distance in about a quarter of a time he (pibbur) would need.