Folding Forum

Hello
Can anyone hear that the wrapper of F@H is created for BOINC? Is it true.

AFAIK, BOINC doesn't have the required resources (I guess the programming?) that is needed by F@H. Although, there was an attempt to get F@H on BOINC but that never passed the beta stage due to numerous problems and was abandoned.

TomaszPawel May 11, 2010 - 19:29:

This is normal, the project has periods of drought ...

P.S.

There is a group of recruiting in our forums to Folfing @ Home ...

Guests have bad luck, because not enough that soon you see a BOINC wrapper ---> F @ H as the RC5-72 -> dnetc.net is still our team is also present at F @ H.

So by BOINC, you can count nabijając F @ H points here and there:)

And what is it?
There is a link http://forum.purepc.pl/SetiHome-UD-inne ... ml&st=2370

filu wrote:Hello
Can anyone hear that the wrapper of F@H is created for BOINC? Is it true.

Not likely, no. Sorry.

From the FAQ. FAH on boinc

7im wrote:Not likely, no. Sorry.

I do not want to fold into BOINC. Team Poland is looking for volunteers to fold. And each member of BOINC tells us that it makes no sense, because soon the F @ H will be a wrapper.

If a fah client for BOINC has been on the shelf for the last 4 years, it's probably not coming down off the shelf any time soon. There is no wrapper from Stanford, and any 3rd wrapper would be unsupported, potentially even blocked if it broke the End User License Agreement.

If they want to fold, they should just run the fah client.

filu wrote:I do not want to fold into BOINC. Team Poland is looking for volunteers to fold. And each member of BOINC tells us that it makes no sense, because soon the F @ H will be a wrapper.

Please ask those BOINC members to find the source of this rumor and provide corrected information to those who are spreading it. It's simply not true.

There was a development project to build a wrapper about 4 years ago, as 7im said. That project determined that there were serious incompatibility between FAH and BOINC which could not be resolved with a wrapper and further efforts were abandoned. There are no plans to re-open that sort of development effort.

Anyone is free to choose to run either FAH or BOINC. An inaccurate rumor of a phantom future development is not a valid reason to delay making that choice. Although it's possible, with some trickery, to run both at the same time, we do not recommend that option. Make a choice and stick with it.

Link to the topic in our forum
http://www.forum.zwijaj.pl/viewtopic.php?f=3&t=616

""Project of wrapper FAH/BOINC is not the work of Stanford.
It is created by team, who recently did distibuted.net BOINC wrapper.
Proteinfolding @ home / Folding @ home / is already an advanced stage of production..."" http://www.boincatpoland.org/smf/polski ... #msg103087

7im wrote:There is no wrapper from Stanford, and any 3rd wrapper would be unsupported, potentially even blocked if it broke the End User License Agreement.

Don't want to flame here, but I cannot believe it.
Basically you are saying that, if there is somebody who would put some effort into this, it will be blocked due to EULA.
But on the other hand, FAH says it doesn't have the resources/will to do it by itself.
Well, it's so.... closed-minded.
distributed.net also had a grudge against boinc if I remember, but right now there are 2 boinc wrappers for it.
dnetc has become the 9th overall boinc project, and the 2nd in recent statistics.
In the past months, fah stats page http://fah-web.stanford.edu/cgi-bin/mai ... pe=osstats has flat out, even diminished.
I cannot see a solid argument why fah would not benefit from a boinc wrapper.
How many iterations had the fah client, and how many did boinc had.
Regarding the question of data security, I can understand your point of view, but I think the wcg had the same issue towards boinc, and now they have fully figure it out (probably it's better than yours).
Not trying to flame here, but, if fah would be more open-minded, maybe it will benefit more:
security issues - take a look at wcg
molecular dynamics - take a look at gpugrid
multicore utilization - take a look at aqua, also regarding opencl
optimized applications - take a look at seti beta

If there will be no boinc wrapper, not the end of the world, but I don't see fah passing 10 native PFLOPS in the next 2 years.

From my personal point of view, If there was a boinc wrapper, I would run fah and as back-up(yes, you can set it as back-up) collatz, seti or gpugrid (when they have an ati app)

Till this day I don't know what happens if you find a cure for aids or cancer? Are you going to market it to corporations or are you going to make them public?

cristipurdel wrote:Till this day I don't know what happens if you find a cure for aids or cancer? Are you going to market it to corporations or are you going to make them public?

I don't know enough about the advantages/disadvantages of the latest versions of BOINC to comment on that, but any results that this project gets from research are always released to the public.

There's one feature I really like about BOINC and that's the multi-brand GPU support. I have a HD3870 as my primary graphics card (currently RMAing the HD4850 I had) and a 9800GT as my secondary graphics card (dedicated to F@H). To run both of them at the moment, I have to use the GPU2 client for the HD3870 and the GPU3 client for the 9800GT.

I certainly hope F@H v7 supports multi-brand GPU configurations.

Sahkolihaa wrote:There's one feature I really like about BOINC and that's the multi-brand GPU support. . . .

I certainly hope F@H v7 supports multi-brand GPU configurations.

A wrapper that would allow FAH to run inside of BOINC would do nothing to provide support for other GPUs. FAH would still run only on ATI or nVidia GPUs of recent generations and only with certain versions of the drivers.

Our hopes for multi-gpu support depend on whether OpenCL lives up to its expectations. From what little I understand about it, at the present time, if one unit of work is presented to a nVidia GPU through CUDA, it takes, say, one unit of time. If the same unit of work is presented to an ATI GPU through Brook/CAL, it takes somewhat longer. If the same unit of work were presented to an arbitrary GPU through OpenCL, it would take much, much, much longer to process. There's no way to know when the OpenCL consortium and the manufacturers of drivers for brand-X GPUs will be able to develop code that is sufficiently optimized to be truly productive, but at present, OpenCL is more of an advertising tool than a productive computing platform.

A wrapper would only make sense for FAH's classic CPU client, and the work assignment and result uploading processes would not work through the BOINC APIs but rather happen directly to Stanford while the WU is "running" Because of those issues, BOINC would simply start the existing FAH client and the FAH points would only show on the FAH statistics. That sort of non-integrated processing would not make a FAH shell act like it was a part of BOINC. You can essentially do the same thing by downloading and running FAH directly.

This topic diverged into a discussion of OpenCL. I've split that into a new topic: viewtopic.php?f=16&t=15152

bruce wrote: You can essentially do the same thing by downloading and running FAH directly.

That is the reason I quit folding. As much as I respect the research being done, I simply hate the fah client.
I will start again only if there was a wrapper, so that I can also manage other projects.
I know I'm not representing a large group, but that's the way I feel.

cristipurdel wrote:

bruce wrote: You can essentially do the same thing by downloading and running FAH directly.

That is the reason I quit folding. As much as I respect the research being done, I simply hate the fah client.
I will start again only if there was a wrapper, so that I can also manage other projects.
I know I'm not representing a large group, but that's the way I feel.

Maybe just simply exposing some API would be enough to integrate into BOINC even though it would be a 3rd party thing and unsupported by PG