Need a help

Moderators: Site Moderators, FAHC Science Team

Post Reply
MaxP
Posts: 1
Joined: Sat Jul 27, 2019 7:56 pm

Need a help

Post by MaxP »

Hi!

I realize this might be unlikely because of the nature of FAH and the simulations, but I thought I would ask just in case. Sorry if the following is naive.
So I would like to figure out how much computation a project needs, or better yet, how much a full protein simulation would need. Ideally in x86 FLOPs. If that is unestimatable for an arbitrary protein, what about the amyloid beta protein?

It looks like different work units, actually require a different amount of computation? Is there any standard, average, or equation that can be used to summarize or predict the amount of computation needed per work unit or a sequence of work units for a project?

So if I had this information about the computation needed for a project, protein, or the amyloid beta protein, I can make estimates that relate to x amount of Amazon instances and what not, which would be the meat of the blog post.

Hopefully a post like this would let more people know about folding at home, and maybe even encourage some tech companies to run FAH on their employee computers during non-work (idle) hours, or even better yet, having Amazon AWS running it on unused/unallocated instances.

Thanks!
bruce
Posts: 20910
Joined: Thu Nov 29, 2007 10:13 pm
Location: So. Cal.

Re: Need a help

Post by bruce »

I don't know a way to answer that question.

Before FAH was started, calculating the motions of N protein atoms for K nanoseconds could be estimated as proportional to K and to N**2. The numbers were so daunting that the concept was considered impossible. Then FAH's research team cam up with more and more scientific methodology that could accelerate the process which skipped increasing percentages of unnecessary simulation segments while actually calculating the essential parts. FAH's reseaerchers continue to develop improvements the overall efficiency of the calculation processes.

Of course the hardware developers have also improved the speed of the calculations over the same period of years. Originally, the increasing the number of concurrent calculations was a big hardware factor as CPUs went from single processors to 2, 4, 8, 16 ... concurrent floating point calculations. Then GPUs came along with dozens, hundreds, thousands, ... of concurrent FP operations and software development continued so that the problems presented could use those increasing numbers of GFLOPS effectively.

The problems being solved have also evolved ... mostly by needing to analyze more complex proteins ... although there still are a lot of smaller proteins that need work.
foldy
Posts: 2061
Joined: Sat Dec 01, 2012 3:43 pm
Hardware configuration: Folding@Home Client 7.6.13 (1 GPU slots)
Windows 7 64bit
Intel Core i5 2500k@4Ghz
Nvidia gtx 1080ti driver 441

Re: Need a help

Post by foldy »

I don't know what is needed to complete a whole project but a work unit gets FAH points for completion. CPU or GPU performance is then measured in points per day PPD. A nvidia gtx 1080ti gets 1 mill PPD and current GPU work units have 50k points in average which is 20 work units per day. An 8 core CPU may do 100k PPD and current CPU work units have 10k points in average which is 10 work units per day. So you can say for an Amazon AWS instance according to its CPU and GPU configuration how many PPD it gets and work units per day it can do. By the way Amazon AWS is expensive there are more cheap alternatives for folding like https://vast.ai/console/create/
debs3759
Posts: 138
Joined: Fri Oct 07, 2011 3:29 am

Re: Need a help

Post by debs3759 »

I don't think it can be broken down that easily. Different jobs take different amounts of time on a reference system, and are awarded base points accordingly. Then, if you have a passkey, you gain extra points for how fast you complete the work. Less time = more points. Different work units will require a different amount of processing power to earn the same points as well. So to work out how many ppd/$ for rented server time is probably beyond the forum to give a formula for.
Post Reply