a charity idea

[kyky-pipi]

Well, hello everyone. It's my first post here and probably the last so pls be generous.
I also didn't want to spam in the other topics so I thought here it would be acceptable

Want to say some words about the average users wanting to contribute.. Well, of course there are many addicts out there doinge 100K ppd's every day, clocking their brand new bought cards to climb up the efficiency charts showing muscles, but in my oppinion all the other thousands average users are making this global network project work. More or less I think myself as an average and by the expreience I was making with the FAH in the last week I come up here with some (hopefully) useful ideas, for example: I've got only one desktop computer with a 96 shaders GPU, I love silence while sleeping so I clocked it under at the lowest clocks to work 24/7 dissipating least possible heat to keep the fan (safely and manually) quiet and make me not care about my electricity bill at all.

So here is the whole charity idea:
When you buy a brand new video card today it is 90% certain it could run FAH. We know that there is also tones of other useless jink inside telling you what a video card is. With not many efforts and some noble support from the manufacturer there could be also a small sized childrens' book, with a UNICEF like picture on the front and a FAH cd at the back, which explains to the newcommer what a piece of powerful machinery he has bought and how great appreciated his help for this noble cause would be.
This will work in the long run only if some very important demands to the client are being fulfiled:
-background software form the type "agree and forget" - it forces the driver at its lowest speeds when doing exceptionally FAH work to keep the card (and the bill) quiet (fan control only as long the card doesn't overheat) - it uses the core !only! if not other 3d intensive applications are being run (a.k.a. - cpu usage doesn't rise to full - monitoring the directx exclusive access is not an option because of the fancy GUIs nowadays) - when addressed it always mentions the newest successes of this analisys to bring up the importance of its work

Well more or less that was it. I'm sure some of you have the right connections there to try it, after all a small 2-3 page coloured cd book will add some cents to the price (if planned for big quantities, of course) and could be affordable even to the guys who are ready to spend some buck for the idea. If not, well - a simple banner on the Manifacturer drivers' webpages, on the Catalyst and Forceware webpages will also be effective. And pls dont forget - as charity orinted "agree and forget" - able software as possible. Then there will be less average users against as possible.
Good luck

[John Naylor]

I think it's a good idea but pointed in the wrong direction... almost all of my computer using friends are pros at using them but don't know their Core 2 from their PCI-E... you may be better off trying to convince the likes of Dell or HP to ship their systems with a flyer in the box... but then you have to ask if they would risk a higher RMA rate in case of failure due to overheating from running F@H (I know it shouldn't happen but it does, especially with low grade OEM parts). If you can convince one of those companies to ship with a flyer in the box you would have a much bigger audience

[kyky-pipi]

It seems I've caused some confusion by naming the target audience "average users".
Who I meant were the people who go to the store and buy new graphics to upgrade their video system and look insde the retailer box. They definetly know better what a GPU is and what risks are to considerate by runnng such a software. And I'm sure there are many users of this type out there - ATI sold already more than 2million chips from their 4xxx series, with 40% market share this makes about 2 million more NVIDIA 8xxx, 9xxx and 2xx chips and I Seriously doubt that more of 20% of these cards went to the end users by ready-sold configurations. And even so - when you spend much more money for powerful graphics (not some integrated junk), you know in most cases why exactly.
But John Naylor , if you're right about the fear of RMAs, than some charity banners on the ATI's, NVIDIA's and other driver support&download pages would be cheaper and more effective. After all the people who visit them know what we're talking about.
LG

[codysluder]

I hate to disollusion you and I certainly do not disagree with your fundamental idea, but ----- running GPU software for science is still very much a cutting-edge activity. You're assuming that the technology is mature and the software is both stable and bug-free. Changes to both the drivers and to the FAH software are distributed frequently, making essential improvements. We're just not there yet.

The software that runs on the GPU today is absolutely not a background activity, nor it is a "agree and forget" type of activity. First, there is presently no concept of foreground/background when it comes to work assignments for a GPU. With every update there are changes to the number of people who say it makes their system feel sluggish but most agree that game-playing requires you to stop FAH and resume FAH when you're done.

Second, although the software is evolving rapidly, my guess is that it will be another year (others may disagree) before it reaches the developmental stage that you'll be able to trust it to keep working while you ignore it.

The software that uses a single CPU-core has reached the level of maturity that you're expecting, so it could be distributed as you suggest. Yes, it produces more limited results, but it is truly a background activity that you can set and forget. (I won't slow down your other activities or crash and probably the only way you'd notice it is because it keeps the CPU at 100% constantly. In fact, several years ago the FAH cpu client was distributed with the Google Toolbar and the only user interface was a Yes/No selection about whether you wanted to contribute unused resources to science.

[teigertoes4]

WANT TO SET UP A COMPUTER SPECIFICALLY ORIENTED TO 24/7 USE ON FAH.

I might use it a little as a server for access to stored files, relatively infrequently. I gather that the graphics cards processors are now quite useful--so am looking at not the hottest CPU (??), on board that has what it takes to let FAH access resources effectively (faster PCI-16-??), and how much RAM, overall balanced with the speed of the Comcast line, but not "overkill", and put the expenditure emphasis on the graphics card/processor. (???) I don't even mind the FAH processing to tell me "sorry, your server request will have to wait a few minutes."

I know enough to put a good heatsink on it and go in and set BIOS values, as I can, and enough to build it physically. I would prefer to run an inexpensive AMD as the main processor if it is fast enough to the internal shuffling required--like the 2.2 Ghz one.--or, someone tell me what the feasible low-end regular CPU would be for this.

So I will run this machine for "charity" PARTICULARLY as my son has familial amyloidosis and I have an incentive. But someone else needs to do the "charity" for ME of providing a high-level motherboard and components design for the machine. A shot at around $300 and around $500 would be a good starting place--I have cases, fans, etc., a management monitor (off most of the time) and assume I need a new power supply. My low-energy server idea is to run a bunch of read DVD drives, so I would need regular slots for the cards to connect those. Doesn't have to be compact--will hang in a crawlspace box with a trap door for access--for keeping temps low.

[4ndr3w]

This is kinda relevant, but I'd like to see some advertising of f@h. if standford and pandegroup began to advertise, i'm sure they'd see and influx of new users.
with this in mind, it should not only be stanford and pandegroup advertising. SPREAD THE WORD!

Also, if they do begin to advertise, i seriously suggest making the download page a lot simpler. how about a "basic" version and "advanced" version.
in fact, i know there are methods of figuring out your hardware set-up without the user having to know. so, if my grandparents went out and bought a new computer with a mid-range graphics card, and they decide to run f@h, all they would have to do is go to the website, the site would scan the computer, and select the appropriate software to install, and then would do so. this could be so simple it wouldn't even be funny.

[4ndr3w]

teigertoes, look at my setup (in my signature at the bottom of my posts).
THAT is a pretty-powerful pc that won't break your budget.

that setup only cost me $600. If you bought a cheaper case and smaller powersupply and a nvidia gts250, you could save probably $150 off that. If you know how to put it together, you might want to consider pm-ing me your email, and i'll send a link to a newegg shopping cart. just give your final budget.

[John Naylor]

This is kinda relevant, but I'd like to see some advertising of f@h. if standford and pandegroup began to advertise, i'm sure they'd see and influx of new users.
with this in mind, it should not only be stanford and pandegroup advertising. SPREAD THE WORD!

Also, if they do begin to advertise, i seriously suggest making the download page a lot simpler. how about a "basic" version and "advanced" version.
in fact, i know there are methods of figuring out your hardware set-up without the user having to know. so, if my grandparents went out and bought a new computer with a mid-range graphics card, and they decide to run f@h, all they would have to do is go to the website, the site would scan the computer, and select the appropriate software to install, and then would do so. this could be so simple it wouldn't even be funny.

Folding@home only has limited funds so between upgrading servers, paying for any third party coding (e.g. the new server code for distributing work), backup of results on long-term storage (running to tens and probably hundreds of terabytes a year now) and other running costs of the project, there is very limited (if any) funding available for advertising.

As for the downloading, the eventual aim is to have a universal installer (one for each of Windows, Mac OS X and Linux) which is downloaded to the PC and then selects, downloads and installs the relevant clients based on hardware present and contributor preferences. However the Pande Group want all the versions of the software to be stable before they implement this, as the SMP client on Windows and the GPU client are still in beta, and as such are not recommended for new users (though most people can get the GPU client up and running with no prior knowledge, it's still not as stable as the Pande Group would like).

[bruce]

with this in mind, it should not only be stanford and pandegroup advertising. SPREAD THE WORD!

The best advertising is word-of-mouth. Please tell your friends about FAH. Make a video and post it on-line. Twitter about it. etc. Wherever you are, that's a good place for YOU to advertise FAH.

As a general rule, advertising at a University is mostly limited to attracting new students, not making a big push for a project like FAH. (They're more likely to advertise scientific achievements in the publications for Alumni where they're looking for bequests and other large donations -- you know, the kind of donations that puts your name on a new science building.)

NOTE: I'm not saying that FAH won't advertise at some time in the future, I'm just saying that IMHO it's unlikely.)