What would happen if Google used all their servers and workstations for Folding@Home for 24h?

[Real_Smoky]

Let's imagine Google stopped offering the search engine, mail service and all of their other services for one day, just straight up paused it, and used all of their available hardware to run F@H for 24h? How big would the impact for F@H be? Could F@Hs infrastructure even work with the amount of processing power that Google has?
 

[Needfuldoer]

The Internet would grind to a halt almost immediately, and F@H's servers would be overwhelmed soon after.

[Hairless Monkey Boy]

And markets around the world would crash, induced by the fear of the entire internet grinding to a halt.

[igormp]

Let's not forget that google not only have their own core services, but also their cloud offering (GCP), which is responsible for ~10% of the cloud market, so you'd see almost 1/10th of the internet services going down along with it.

 

I 'd dare to say that AWS would have a bigger impact lol

[Zando_]

24 minutes ago, Needfuldoer said:

The Internet would grind to a halt almost immediately, and F@H's servers would be overwhelmed soon after.

^ This. F@H has had a hard time keeping up with work unit requests during big folding events put on by the community, they wouldn't be able to serve even a small percentage of the compute power Google has. 

[Eigenvektor]

38 minutes ago, Real_Smoky said:

How big would the impact for F@H be?

Aside from their public facing servers, I would imagine they have a ton of internal compute power. For example machine learning for automated driving, future versions of bard and so on. If we assume the F@H client is compatible with all of these… but since we don't really know what type of resources they have, my best guess would be "huge".

[Alex Atkin UK]

58 minutes ago, Eigenvektor said:

Aside from their public facing servers, I would imagine they have a ton of internal compute power. For example machine learning for automated driving, future versions of bard and so on. If we assume the F@H client is compatible with all of these… but since we don't really know what type of resources they have, my best guess would be "huge".

As already mentioned they struggle to keep up as it is, so I don't think it would make any appreciable difference other than making our participation no longer necessary.  The bottleneck is more and more down to funding and staffing at their-end.  They can't just magically ramp up the WU generation.

[Poinkachu]

This :

https://youtube.com/clip/Ugkxrb0bHT8WaLRBRQ-15CsEinpdCSz8jmf3?si=k2esV_K6s6sF_NWm

[Alex Atkin UK]

6 hours ago, W1zarDdd said:

If Google decided to use all of its servers and workstations for Folding@Home for 24 hours, it would result in a significant increase in the computing resources available for the project. This could speed up the research and data analysis carried out within Folding@Home and possibly help achieve new scientific discoveries or improve our knowledge of biochemistry and medicine.

Unfortunately completing the WUs quicker, doesn't mean they can process them at their-end quicker.  They frequently run out of WUs when a lot of people are running F@H as its not like the people doing the actual work at the other end are sitting on their hands waiting for our results, they have finite resources to process our results and issue new WUs.

[Eigenvektor]

9 hours ago, Alex Atkin UK said:

As already mentioned they struggle to keep up as it is, so I don't think it would make any appreciable difference other than making our participation no longer necessary.  The bottleneck is more and more down to funding and staffing at their-end.  They can't just magically ramp up the WU generation.

So you're saying completely overwhelming F@H's servers is not a huge impact?

[leadeater]

13 hours ago, Real_Smoky said:

Could F@Hs infrastructure even work with the amount of processing power that Google has?

No, it couldn't cope a few years back with the massive surge so we actually already know the answer.

 

13 hours ago, Real_Smoky said:

How big would the impact for F@H be?

Depending on any reliance with Google services F@H itself could stop... so probably not a good idea to stop everything 

[William Payne]

It sounds good but would be highly impractical. 

 

I have a few issues with folding. I do find the folding at home movement quite commendable but as someone with a large interest in science I can't say I like the idea of doing it myself. 

 

But only because I want to be the one doing the science stuff. Folding at home is equivalent to lending a physicist my calculator. I am still not the one doing the work. 

[Alex Atkin UK]

9 hours ago, Eigenvektor said:

So you're saying completely overwhelming F@H's servers is not a huge impact?

I guess the fact were able to do that ourselves on Folding month makes me not really consider that so huge.

 

I wonder if I'm undervaluing my own impact on the project given I can't see that impact other than hitting the top 256 contributors? (I've dropped ranks in the summer though)

[SansVarnic]

7 hours ago, William Payne said:

Folding at home is equivalent to lending a physicist my calculator.

Not a very good example. It is much more complicated.

 

F@H is a way to give a hand in running the various scenarios for the discovery and proof part of the scientific process. Each WU is not meant to result in the same result it is meant to reveal how a different result, results. Why do the proteins change and what causes it to change? To accomplish this, having a chain of computers run the same thing hundreds to thousands of times helps along the process.

 

22 hours ago, Real_Smoky said:

Let's imagine Google stopped offering the search engine, mail service and all of their other services for one day, just straight up paused it, and used all of their available hardware to run F@H for 24h? How big would the impact for F@H be? Could F@Hs infrastructure even work with the amount of processing power that Google has?

Beside the others answers given, F@h in the long run is probably better to yield results rather than using super computers or the cloud. Independent pc helps achieve the goal of independent verification of the results.

[Srius1]

Waste of time and your energy $ to do this nonsense.

 

someone show me ONE THING THAT fold has accomplished other than raising the heat and wasting energy

[Alex Atkin UK]

4 hours ago, Srius1 said:

Waste of time and your energy $ to do this nonsense.

 

someone show me ONE THING THAT fold has accomplished other than raising the heat and wasting energy

You do know science is based on multiple peer-reviewed studies, right?  Its not like Folding will immediately provide all the answers, its a small part of getting a bigger picture that eventually, hopefully, results in something we can actually work with.  It provides data set to work with, that still needs a ton of research on the results to actually achieve anything, but without that data, the research can't happen.

 

https://foldingathome.org/papers-results/?lng=en

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7337393/

 

Quote

The project has provided insight into diverse topics, ranging from signaling mechanisms.13–15 to the connection between phenotype and genotype.16–18 Translational applications have included new means to combat antimicrobial resistance, Ebola virus, and SFTS virus.19–21

Also, I only Fold when its cold so while its more expensive than the gas heating, its serving a dual-purpose so well worth the extra cost.

[William Payne]

10 hours ago, SansVarnic said:

Not a very good example. It is much more complicated.

 

F@H is a way to give a hand in running the various scenarios for the discovery and proof part of the scientific process. Each WU is not meant to result in the same result it is meant to reveal how a different result, results. Why do the proteins change and what causes it to change? To accomplish this, having a chain of computers run the same thing hundreds to thousands of times helps along the process.

 

Beside the others answers given, F@h in the long run is probably better to yield results rather than using super computers or the cloud. Independent pc helps achieve the goal of independent verification of the results.

 

I know exactly what it is for. What I was saying was if I am going to have my computer do scientific work I want it to be my own work.

[SansVarnic]

11 hours ago, Srius1 said:

Waste of time and your energy $ to do this nonsense.

 

someone show me ONE THING THAT fold has accomplished other than raising the heat and wasting energy

Producing results to help understand viruses and cancer to make treatments. I cant say off hand which one atm but I am aware this program has done much in producing viable information to the production of treatments for various things over the years.

[cmndr]

Let's reduce the scope of this to: 
"what would happen if Google took all their non-critical compute and redirected it at F@H?"

This would look like the laptops and desktops of 300k employees and contractors, a bunch of research systems, etc.

1. A small drop in the bucket for F@H in terms of improvement, assuming their servers can keep up. 
2. You'd be diverting efforts from research.

Realistically F@H is a brute force effort. The big win for protein folding is going to come from smarter algorithms, NOT from more compute. The right algorithm can result in a computer doing 1 billion times the work in the same amount of time. 

[Mark Kaine]

For anyone saying they wouldn't have enough processing power because of more processing power: no. these things can fairly certain be scaled up.

 

for anyone saying this would lead to the almost immediate collapse of the internet and other services/ business with it: Bingo! 

[Alex Atkin UK]

On 9/23/2023 at 9:01 AM, Mark Kaine said:

For anyone saying they wouldn't have enough processing power because of more processing power: no. these things can fairly certain be scaled up.

The computing can only scale up if the human beings dealing with the results can be scaled up, which needs funding and people qualified to do so.

 

We've run out of WUs before during LTT folding events due to this.

[TimedPing]

On 9/22/2023 at 1:30 AM, Eigenvektor said:

Aside from their public facing servers, I would imagine they have a ton of internal compute power. For example machine learning for automated driving, future versions of bard and so on. If we assume the F@H client is compatible with all of these… but since we don't really know what type of resources they have, my best guess would be "huge".

Or repurposed hardware from Stadia, the GPU alone is already very spicy per rack