The worst software I have ever seen.

[v00d00]

Its down to experimentation. For example. if you download the latest client then you get version 7.5.1. However if you download the Linux server version you get 7.4.4. So I have manually installed 7.5.1 on the Linux servers. No instruction to do this but I figured it out myself. Now I could have got on here and moaned as to why they don't fix their website. Well, actually I think they are better off trying to fix "Us" with their protein research, which I have absolutely no clue about. I do have a clue about Linux and I can get my machines to their highest performance, with a little help from the community. Some people need to start thinking about "how to fix" rather than "Who to blame".

A few things:
1. I'm running on linux too, what's the difference between 7.5.1 and 7.4.4? I'm currently running 7.4.4
2. How do I get 7.5.1?
3. Yes I agree, I am also thinking about "how to fix" which is why I would love to help with the dev side of things, to contribute in more ways other than being a compute donor.

Difference between the versions isnt massively relevant to Linux Users. I use 7.4.4 and have done since it came out. I launch and run it from a custom method that everyone says is wrong, but it works well for me and gives me slightly better PPD than using the accepted method. i keep to myself for the most part rather than butt heads with people, which ive done in the past.

On the question of growing the community, I dont think they want to. What they like is a good set of people, who dont complain massively, are willing to take on things that crash occasionally, sometimes you dont get points for things or it all goes wrong, but the science is good and papers get published from time to time. its been like this for the last 20 years. Hasnt changed a great deal since Vijay started it up back then. Management has, but the core way it runs and things get done is much the same as it was long ago. They generate workunits, we process them, send them back, they give us new workunits and the cycle continues. From time to time someone proposes some new expansion like the Playstation Initiative, which runs for a while but eventually whoever sponsored it quits or decides not to continue updating it. GPU seemed to be like that, but it somehow survived, I think because QRB was implemented and it became attractive to folders. But since then a few more white elephants seemed to have surfaced, I honestly dont keep tabs on that side of it. i know what works for me and i continue to do what works for me. As long as that continues to happen i wont bother people. Dont fix whats not broken and if it is broken ask yourself why and then decide whether its worth fixing.

Another thought I had was they could have split the project off. Forked some of it to deal with Covid and left a core set of resources for everyone who really doesnt care about the hype and just wants things to continue like they used to. People whose folding prefernce is for other projects, like the more long term projects. All the people who are into the hype and want to fold covid projects could be running out of a separate fork that caters only for that, has all the sponsored hardware which will disappear the second a cure for covid is found and upon completion of the covid task that side could be shutdown and we could get back to normal. I want stability, thats all im bothered about. Workunits to fold, servers that work and the same like-minded people who have been doing this for a long time.

[iceman1992]

Difference between the versions isnt massively relevant to Linux Users. I use 7.4.4 and have done since it came out. I launch and run it from a custom method that everyone says is wrong, but it works well for me and gives me slightly better PPD than using the accepted method. i keep to myself for the most part rather than butt heads with people, which ive done in the past.

Mind sharing your method?

On the question of growing the community, I dont think they want to. What they like is a good set of people, who dont complain massively, are willing to take on things that crash occasionally, sometimes you dont get points for things or it all goes wrong, but the science is good and papers get published from time to time. its been like this for the last 20 years.

Fair point, but I just drool on the idea of 2.3 (and possibly 3+ now) exaflops working on the long term projects.

[Joe_H]

1. I'm running on linux too, what's the difference between 7.5.1 and 7.4.4? I'm currently running 7.4.4

As the current version, and prior ones were released to public Beta tests here on the forum, Release Notes included change logs if you are interested in that level of detail.

But to summarize the major changes, the current client has updated connection links to reflect some of the changes involved with moving from being based out of folding.stanford.edu to foldingathome.org. For systems doing CPU folding, it has additional code to handle WU requests for CPU thread settings not currently available from a WS. Support for FAHViewer to display for an older CPU folding core, A4, was removed. Core_A4 has since gone out of use. The code to detect a failed upload or download was improved a bit from 7.4.4 to 7.5.1, but still misses some.

There are some other changes, but these are the major ones I recall.

[v00d00]

Isolated cores is the method. One core per fahclient. Isolated at boot using isolcpus flag, then taskset the client to an isolated core. It works well on 1 and 3 gpu systems, less so on 2 and 4 (unless headless). Basically on my FX 8 core I isolated cores 6 and 7 (isolcpus=6,7), the remaining 6 cores are left to the cpu_scheduler. i run CPU client on the first 6 (taskset -c 00,01,02,03,04,05 ./fahclient-smp &). I run GPU on the 7th (taskset -c 06 ./fahclient-gpu1 &). I run any and all programs like firefox, mplayer, etc on the 8th (append taskset -c 07 programname to all shortcuts).

On higher core counts it gets much easier. On my old Xeon 2u I had 32 cores, so ran 2x 12, 1x 6 and isolated 2 cores for 2x GPU, but the whole server was headless. But for an 8 core or 6 core system with a single GPU that you want to use normally, isolating gives you good performance for folding and also leaves the system capable of being used for most things (general browsing, listening to music, watching movies on mplayer, etc), but not anything hard on rendering like youtube (frame drops due to the gpu already being maxed out). If all you do is read mail, view forums and listen to music, you can do all of that from your free cpu core, if you want maximise even further, you can run X also from that core by assigning the X process to core 8 (or 07). its actually easier to do it a different way by getting rid of GDM or whatever you use and starting in text mode, logging in and adding an alias for startx to your bashrc that adds taskset -c 07 to it. X tends to use about 15% constant cpu time when GPU is running. if it runs on its own core, its obviously not eating into your cycles that could be used by folding.

Code: Select all

alias startx='taskset -c 07 startx '

The above is an advanced users only config. it requires editing your bootloader config, startup scripts/systemd, general faffing on the console. For people without a good 6 months of constant usage and experience its probably going to be too hard.

Fair point, but I just drool on the idea of 2.3 (and possibly 3+ now) exaflops working on the long term projects.

Its all fine reaching that level of power, but if the infrastructure wont handle 2 exaFLOPS with any reliability, you would just end with an even worse situation where probably 70-80% of the clients cant get work. Better to have 1 exaFLOPS worth of clients that have work 100% of the time, than 3 exaFLOPS and no workunits for most of the people folding. The buck will always end with the number of scientists generating work. You can add millions of new servers, but unless you have the skilled people to deal with the workunits, its all meaningless. The number of people who fully understand the folding process to being useful is very low, that wont change massively. Developers can add more servers, make new software, but they dont have PhD's in BioChem and cant deal with the actual work.

[HaloJones]

Its all fine reaching that level of power, but if the infrastructure wont handle 2 exaFLOPS with any reliability, you would just end with an even worse situation where probably 70-80% of the clients cant get work. Better to have 1 exaFLOPS worth of clients that have work 100% of the time, than 3 exaFLOPS and no workunits for most of the people folding. The buck will always end with the number of scientists generating work. You can add millions of new servers, but unless you have the skilled people to deal with the workunits, its all meaningless. The number of people who fully understand the folding process to being useful is very low, that wont change massively. Developers can add more servers, make new software, but they dont have PhD's in BioChem and cant deal with the actual work.

I agree with this wholeheartedly. I'm happy to see the interest in CV-19 but also miffed that it wasn't there for Cancer or Alzheimer's. When (hopefully) a vaccine is found for CV-19 will all these new folders and the new hardware melt away? I expect so. But will this current situation of failed uploads, delays in getting work, lost points, etc. have deterred some of the dedicated folders? I expect so.

The likes of GamersNexus, Linus Tech Tips, Nvidia and Intel, encouraging their users to fold need to also encourage their users to continue when CV-19 is dealt with. Along with money to fund the actual research generated from the results of folding.

[iceman1992]

I agree with this wholeheartedly. I'm happy to see the interest in CV-19 but also miffed that it wasn't there for Cancer or Alzheimer's. When (hopefully) a vaccine is found for CV-19 will all these new folders and the new hardware melt away? I expect so. But will this current situation of failed uploads, delays in getting work, lost points, etc. have deterred some of the dedicated folders? I expect so.

The likes of GamersNexus, Linus Tech Tips, Nvidia and Intel, encouraging their users to fold need to also encourage their users to continue when CV-19 is dealt with. Along with money to fund the actual research generated from the results of folding.

Exactly my thoughts. In the long run Cancer or Alzheimer's will kill many many more people than Covid19, we all should encourage people to continue afterwards.

[v00d00]

Covid is the current hyped up disease. But like all viruses, a vaccine or treatment will appear, sooner or later. Cancer, Huntingtons, Alzheimer's, HIV (and other similar diseases) are more deadly and kill more people overall than this will. The main difference is they dont do it quickly. But if no cure is ever discovered, how many people will die from Alzheimers in the next 30 years globally, or Cancer?

[offtopic]
One thing im still wondering which I hope the research will give answers to is why CV-19 seems to kill more Asian, Black and other Minority groups, than dare I say it, White people. The stats in many country seem to point to it almost targeting those groups over others. In the UK some stats were published yesterday suggesting that out of the total population the number of deaths in certain sets of people were proportionally higher than others. The media seems to push the view its down to poverty, but im not 100% sure of that. Hopefully the research will be able to prove why and what is causing it. Or maybe it is poverty. Either way an answer (or theory) would be nice.

[HaloJones]

[off topic]
ethnic minorities generally have less disposable income, are less likely to have health insurance, are less likely to be able to work from home, less likely to be able to have the time for exercise, likely to be in less healthy living conditions, less able to eat healthily, etc.

lots of contributing factors to shift the survivability odds

[BobWilliams757]

Its all fine reaching that level of power, but if the infrastructure wont handle 2 exaFLOPS with any reliability, you would just end with an even worse situation where probably 70-80% of the clients cant get work. Better to have 1 exaFLOPS worth of clients that have work 100% of the time, than 3 exaFLOPS and no workunits for most of the people folding. The buck will always end with the number of scientists generating work. You can add millions of new servers, but unless you have the skilled people to deal with the workunits, its all meaningless. The number of people who fully understand the folding process to being useful is very low, that wont change massively. Developers can add more servers, make new software, but they dont have PhD's in BioChem and cant deal with the actual work.

I agree with this wholeheartedly. I'm happy to see the interest in CV-19 but also miffed that it wasn't there for Cancer or Alzheimer's. When (hopefully) a vaccine is found for CV-19 will all these new folders and the new hardware melt away? I expect so. But will this current situation of failed uploads, delays in getting work, lost points, etc. have deterred some of the dedicated folders? I expect so.

The likes of GamersNexus, Linus Tech Tips, Nvidia and Intel, encouraging their users to fold need to also encourage their users to continue when CV-19 is dealt with. Along with money to fund the actual research generated from the results of folding.

I'll be the first to admit that though I had done some folding in the past, I got away from it due to other life priorities. And seeing some links and feeds about folding got me back interested. I really didn't expect that the systems would be impacted as greatly as they have been, but figured any contribution would help. I completely agree that sources should urge people to continue folding long after we have forgotten about COVID-19, assuming that it is gone before other needed research is solved.



And I can completely understand the frustration of long term folders, as they have devoted more time, money, and hardware resources towards folding. I'd be fine if "new" users got no points for a period of time, were placed on lower priority WU's, or whatever.



But in the long run, won't this sudden surge in interest help the project become more robust and ready for the next cause? Though I was out of the loop when the PS3 surge hit, I can imagine it created some of the same issues. But I would have to think that if a surge in participation actually slowed the research any, they would at some level "pull the plug" and not allow people to sign up, or somehow limit participation so as not to cause issues.

I would hope that the long term folders would simply stay the path, as well as at least some of the newer folders staying on long term.





As for "the worst software" sentiment of the OP... not even close in my opinion. I've found folding to be easy both back in the CPU only days and now. IIRC my last folding system was powered by Cyrix. Granted being around computers long enough to remember setting interupts via dip switches on cards might be a factor in where I find software "bad". And the background in overclocking, power systems, etc was enough to understand stress tests and making sure a system is rock solid before tossing tougher tasks at it.

[empleat]

I agree with this wholeheartedly. I'm happy to see the interest in CV-19 but also miffed that it wasn't there for Cancer or Alzheimer's. When (hopefully) a vaccine is found for CV-19 will all these new folders and the new hardware melt away? I expect so. But will this current situation of failed uploads, delays in getting work, lost points, etc. have deterred some of the dedicated folders? I expect so.

The likes of GamersNexus, Linus Tech Tips, Nvidia and Intel, encouraging their users to fold need to also encourage their users to continue when CV-19 is dealt with. Along with money to fund the actual research generated from the results of folding.

Exactly my thoughts. In the long run Cancer or Alzheimer's will kill many many more people than Covid19, we all should encourage people to continue afterwards.

If someone cares about points, i don't know what to say, but people do... If you wanted to make FAH more popular - give people free PC games, for each x points, people like getting things for free. And so many gamers would join and share that with their friends, or advertise that on stream. Or give random wheel of fortune, people will crunch just like that, can't believe so many people pay for chance to win something... Which is like... Or better PR and advertisement. I read it on The Verge, which is popular online mag, but still not sure how many people will see one article. Btw i knew boinc for years and i am a lot on my computer, so i know things... Or hype it on social media, there is a study, if people get excited about something, there is very high likelyhood they will share it with friends. HAHA FAH could use of troll farms that would be funny.

And thing is - FAH needs to be more user friendly and warn people to check their cooling and remove dust and do simple stress test on their computers. Make program install only after you click link and read all of this. But explain it is really simple and doesn't require any knowledge so people just don't leave... And make tutorial for that !!! And gpu usage limiter, single pc users can use only cpu. Nvidia are greedy bastards, they won't help and nvidia dev says it is up to FAH to come with gpu usage limiter, not sure who has the truth. But it is hard to believe it can't be done.

[iceman1992]

If someone cares about points, i don't know what to say, but people do... If you wanted to make FAH more popular - give people free PC games, for each x points, people like getting things for free. And so many gamers would join and share that with their friends, or advertise that on stream. Or give random wheel of fortune, people will crunch just like that, can't believe so many people pay for chance to win something... Which is like... Or better PR and advertisement. I read it on The Verge, which is popular online mag, but still not sure how many people will see one article. Btw i knew boinc for years and i am a lot on my computer, so i know things... Or hype it on social media, there is a study, if people get excited about something, there is very high likelyhood they will share it with friends. HAHA FAH could use of troll farms that would be funny.

Good idea, and would work if it happens, but I think it would require sponsorship/partnership from those game companies.

And thing is - FAH needs to be more user friendly and warn people to check their cooling and remove dust and do simple stress test on their computers. Make program install only after you click link and read all of this. But explain it is really simple and doesn't require any knowledge so people just don't leave... And make tutorial for that !!! And gpu usage limiter, single pc users can use only cpu. Nvidia are greedy bastards, they won't help and nvidia dev says it is up to FAH to come with gpu usage limiter, not sure who has the truth. But it is hard to believe it can't be done.

I would say as a first step for making it more user friendly, remove the passkey requirement for bonuses, that just confuses so many newcomers looking from the number of posts with missing passkeys.

[HugoNotte]

I would say as a first step for making it more user friendly, remove the passkey requirement for bonuses, that just confuses so many newcomers looking from the number of posts with missing passkeys.

The passkey separates those who can read from those who can't...
Really, the information is available during installation of the FAH Client software, where it points out that a passkey is required for bonus points.

[PantherX]

...And gpu usage limiter, single pc users can use only cpu. Nvidia are greedy bastards, they won't help and nvidia dev says it is up to FAH to come with gpu usage limiter, not sure who has the truth. But it is hard to believe it can't be done.

Technically, F@H Team can't come up with a GPU limiter since the drivers/firmware can't support it. I am aware that Nvidia implemented concurrent execution of floating point and integer instructions in Turning GPUs but that isn't what we can use to limit the GPU usage.

[iceman1992]

I would say as a first step for making it more user friendly, remove the passkey requirement for bonuses, that just confuses so many newcomers looking from the number of posts with missing passkeys.

The passkey separates those who can read from those who can't...
Really, the information is available during installation of the FAH Client software, where it points out that a passkey is required for bonus points.

It's not really essential for the science, is it? Let people get bonus points without passkeys, but keep passkeys available for people who want to be sure that they're only seeing their own points.

[PantherX]

...It's not really essential for the science, is it? Let people get bonus points without passkeys, but keep passkeys available for people who want to be sure that they're only seeing their own points.

Passkey is required for security purposes and it also has the flexibility to add new Stats feature in the future: https://foldingathome.org/support/faq/points/passkey/

I do know that there was a case were it was confirmed that a donor was illegally installing F@H on systems and folding. After an investigation, and confirmation of that behavior, all the points were zeroed out under that username's passkey. While not frequent, it has happened so that's why a passkey is needed... to protect your identity.