Folding Forum

Hello I was reading on Wikipedia as well as some older news post that world Folding at home used to be at 40 petaflops, now all I see we are around 20. Is there a particular reason we lost half of our FLOPS, like new way of calculating them? It just seems very peculiar and I have never seen anyone mention why.

I doubt there's a new way of calculating them. The reports have always been based on a very conservative estimate, so the real throughput is at least that much, but that wasn't your question.

I can think of a couple possible reasons, and there may be more.

1) One family of projects (called "bigadv") which was assignable only to server-class computers with a multitude of CPUs was discontinued. (Those projects were very lucrative in the points they awarded.) Many of the donors who had access to that class of hardware moved away from FAH projects.

2) Updates to the analysis software ("FahCore_xx") that was available for GPUs (where xx = 15, 17, 18), while more productive that the FahCore_a4 for CPUs, have been undergoing changes and now are being gradually replaced by FahCore_21. I hope these changes will re-grow FAH's 40 PFLOPS and will even exceed that high-water-mark.

3) The popularity of volunteer based projects, much like that of political candidates, are influenced by many factors. FAH has done very little in the way of advertising other than to claim superiority of the scientific research and doesn't really compete with other projects that have a more game-like look-and-feel.

These are just my opinion, not anything based on hard data.

One data point supporting unfortunate point 3) are the numbers of computers as shown on the homepage. Down from 300k to 100k; if my memory serve me well in the last two years.

I think the higher marks of FLOPs were achieved when the PS3 client was still active ... it's not the case anymore, so that could explain why the numbers went down ...

And the economic situation around the world might not be favourable to donating electricity to a scientific project ...

WMCheerman wrote:Hello I was reading on Wikipedia as well as some older news post that world Folding at home used to be at 40 petaflops, now all I see we are around 20. Is there a particular reason we lost half of our FLOPS, like new way of calculating them? It just seems very peculiar and I have never seen anyone mention why.

You could probably provide as much insight was anyone else, being a fah donor.

How has your folding changed in the last few years? Upgraded any PCs lately? Or moved to a mobile device instead? Any negative economic effects? Have any children move out so their computers no longer fold? Or maybe retired and had to cut back a little? There are many things that affect how people make donations, including changes to people and changes in this project. For example, a laptop spends much more time moving around than sitting and folding 24/7 like a PC. And the mobile device is also less powerful than a PC.

What is your fah user name? Curious to see how your donations have changed over the last year as compared to the rest of us.

It is also common knowledge that computer contributions are down across all distributed computing projects, not just Folding@home. Here are two articles about that...

http://www.nature.com/news/computer-sha ... um-1.14666
https://www.inverse.com/article/6592-th ... ure-cancer

Unfortunately, there are no black and white answers here, but decline is the general trend (so you should go and recruit some new donors). Also note that Folding@home has corporate donation projects that come and go, and many times those donations are not publicized because of Non-Disclosure Agreements. Other DC projects might get jealous.

Would be still good to separate report on "@home" community donors vs corporate donors. Don't get me wrong: corporate contribution is good for the science; great. But for the community building the contribution of corporate donors are less relevant and should be separated to avoid confusion on what the community achieved itself.

ChristianVirtual wrote:Would be still good to separate report on "@home" community donors vs corporate donors. Don't get me wrong: corporate contribution is good for the science; great. But for the community building the contribution of corporate donors are less relevant and should be separated to avoid confusion on what the community achieved itself.

Sorry, my community takes a more inclusive approach. As the project has always said, "All donations are welcome."

And your comment is off topic because it does not address the OP.

ChristianVirtual wrote:Would be still good to separate report on "@home" community donors vs corporate donors.

Whether this is a good idea or a bad one, I don't think it's feasible.

When a new installation of FAH first contacts the server, it's given a "UserID" automatically which keeps track of that particular installation. The servers can't tell a corporate installation from the installation by an individual Donor.

"And your comment is off topic because it does not address the OP."

As you mentioned in your previous post:

"Also note that Folding@home has corporate donation projects that come and go"

The high of 46 PLOPS seen a year ago compared to the current production is very likely due to this sudden decline.

mdk777 wrote:The high of 46 PLOPS seen a year ago compared to the current production is very likely due to this sudden decline.

That may be true and it may not be, but it would be based on an unsupported guess. Nobody knows when there are corporate donations and when FAH is waiting for the next wave of donations.

Corporate donations are not announced and FAH does not provide free advertising or other corporate benefits in exchange for those donations.

Here are the numbers from 2 years ago and it were 26 Petaflops.
http://fah-web.stanford.edu/cgi-bin/mai ... e=osstats3

So the 40 PLOPS was partly a corporate donation with high probability.
Because home user donors count will not change so quickly about this amount.

Are the corporate donations captured under the "Anonymous" user name in the "Default" team?

PS3EdOlkkola wrote:Are the corporate donations captured under the "Anonymous" user name in the "Default" team?

Most likely, but that, too, would be a guess. So are the donations from hundreds of other anonymous individual donors. Tough to answer when under an NDA.

Using HP as one possible example, what is the difference between 10 computers in a small research lab inside of Hewlett Packard, and 10 powerful computers in the HP Cloud? Or the 10 computers in a Job Training lab at a church? Or the 10 computers in a lab at the local college? Or 10 computers in my basement that I paid for while working for a large corporation?

Maybe BigAdv folders should NOT be considered @home folders? Those giant server class computers would never have been purchased by an average @home consumer for their home use. Who really buys more than one or two GPUs for a computer to use at home or who needs more than a dozen threads for @home use?

But again, this is not the topic of discussion. Examples of why donor numbers can and have fluctuated have been offered. Topic asked and answered. The only party that knows the full answer is Pande Group and even they cannot disclose the full answer.

This is a support forum and how does this discussion provide useful support to people who are having troubles setting up their system?

Topic closed.

Here's an update on the subject : https://folding.stanford.edu/home/closi ... petaflops/

I have noticed in the last couple of weeks a steady decline in FLOPS, from a high of about 106 petaflops through today it is around 88 petaFLOPS. Does anyone have any thoughts on this sudden decline? I noticed also the users have drop of has well from just over 100k to in the low 90sK. Is this related to some of the servers going offline? I have also wondered maybe people are switching to NVIDIAs new 1080 & 1070 graphics cards and those cards having issues with F@H? This drop just seems so odd especially after we just got to 100 petaFLOPS.