Folding Forum

There is no real difference in performance in this project number (13415) and 13409 that myself and other high-end AMD GPU users have complained about in this thread I have gotten 13415 WUs exclusively since Friday night. It's killed my PPD but it's also helped me to cope with not being so obsessed with my points. My WU count has jumped, so there's that I guess.

FAH has been working with "standard users" for over a decade. It relies on simple installation (which it is for most people), and no monitoring, fine tuning or controlling the donors' systems. I don't understand why you think that has to change?

It doesn't have to change, but it does have a real chance to get to a new level. Hardware and software evolve constantly, and so should FAH.

ajm wrote:It doesn't have to change, but it does have a real chance to get to a new level. Hardware and software evolve constantly, and so should FAH.

Change has to be for a beneficial reason that can justify the opportunity cost. I would love to see a solution that aligns work perfectly with the capacity of each donor but FAH cannot know for sure that a particular piece of hardware will be used 24/7 dedicated excusively to FAH. This has a huge impact on the value of the donor's hardware.

Indeed. But this aspect can be handled with a couple check boxes somewhere and then maybe an algorithm based on the observed behavior of the slot. There are countless such problems to solve of course and, yes, it would be a very ambitious endeavor. But then FAH is the most powerful supercomputer on Earth, after all. Huge resources have been leveraged these last months, by individuals and entreprises - this shows that such a project can gather energies and goodwill.

As for the deployment solution, many people came to this forum to ask for it, because they had the hardware but not enough time on their hands for configuring hundreds of kits one after the other, and then monitoring them efficiently.

One good question important to ask is how long those big companies will be interested in folding? Will it last after SARS-Co-V2 is in control some day?

How much money and effort is reasonable to spend hoping they will stay?

The truth is that we don't know. What we do know is that they did come. FAH asked for donors and they came. Then FAH asked for servers and they gave them.
Now, if FAH asks for programming manpower, with a couple interesting forward-looking projects, the chances are good that they come.

ajm wrote:The truth is that we don't know. What we do know is that they did come. FAH asked for donors and they came. Then FAH asked for servers and they gave them.
Now, if FAH asks for programming manpower, with a couple interesting forward-looking projects, the chances are good that they come.

A lot of server and computing capacity has been made available by companies because it had become available due to shut downs which let hardware run idle. Others have donated resources in hope that it will shorten and lessen the economical impact. I expect that as soon as things do return to normal, a lot of resources will be withdrawn from FAH again because they will be required to generate profit.

Yes, that is highly probable. But we can argue that the virus will come back, this one or another one, and that we'd better be really prepared, then. Another good reason to shift the focus on programming solutions that will make FAH more efficient and reactive.

Me neither am not believing in Santa Claus no more.

Well, I'd say that there are quite a few Santas here, on this forum...

Besides, large companies have budgets for this kind of things. Call it marketing. They may want to be associated with FAH and the general idea of fighting diseases and protecting the population. Be it with lending hardware, bandwidth or manpower.

Marketing? Thats pretty much where I see a problem here. Where people loose their interest there's no marketing value any more...

Yes, there has been a variety of resources provided the F@h, over and above prior support given. But that support has to be coordinated by a relatively small group, and any changes, enhancements, or other modifications still need testing before being put out to an active client/server system. From posts here it should be fairly obvious that small changes can have major impact on how the system works. Much of what is being requested is in the area of large changes, those are going to take longer, and many will be on hold for deploying so as not to cause major disruptions during this period.

Depending on how that outside support stays around after the COVID-19 push, F2hmay have to pick which to put out as well based on what they can support.

Finally, in one way COVID-19 came along at a poor time for F@h. It is in the middle of a transition from being mostly based out of Dr Pande's lab at Stanford to a more distributed model. Work was in progress to make the software more open source so that others could contribute easily, that is mostly on hold while working to meet the current needs for servers and getting projects ready.. There is more to come out in the near future, but probably not on the server side of things. These things will be announced as they are ready to be tested and then put into production once validated. Some may not make it out at all.

I had a power failure and when the power came back all 6 units that were interrupted went faulty with "Forces are blowing up!" when processing resumed.
Now I have 7 slots with the 13415 projects. Going to be like those horse races to see which one finishes first

After Standby/Hibernate and resume on Windows the GPU FahCore_22 0.0.10 shows an error and resumes from last checkpoint. But after several Standby/Hibernate the work unit gets dumped as BAD because of max error retries reached. Standby/Hibernate should not count as error.
https://github.com/FoldingAtHome/fah-issues/issues/1529