BOINC Alpha Testing for 6.8
Howdy Folks,

GridRepublic has been working on a Facebook application for use with Volunteer Computer via BOINC. While the Facebook application isn't quite complete yet, the 6.8 client will be focusing on features and functionality geared toward that effort. We are looking for the client software to be finalized and finished by mid-June.

To that end the only new features that are going to be included with this client version is the new screensaver framework and the ability to attach to an account manager via a web browser cookie.

A new BOINC-based client will be making its entrance next week, we will be focused on that until it is ready to go. After the new client is wrapped up we will follow up with an updated GridRepublic client followed by a BOINC client. The new client has a hard deadline of mid-June.

GridRepublic has updated their server-side software to put cookies on your system during the sign-up process to help test the new cookie attach mechanism. We have packaged up an experimental client with the new features to begin testing and we will be following up with the new client software as soon as I get the new skin and artwork.

In order to test the cookie mechanism you'll need to create a new GridRepublic account at http://www.gridrepublic.org/. When it gives you the link to download the new client click on it and then abort the download. Launch the experimental client. It should automatically attach to GridRepublic and then attach to whichever projects you have signed up too. The supported browsers are Internet Explorer 5.x+, Safari 3.x+, Firefox 2.x+. If you use a different browser go ahead and give it a try, let us know if it works.

To see the new screensaver you'll need to suspend the client before letting the system start the screensaver. Other ways include letting the client go idle, or the only active tasks do not support graphics. This new screensaver component is OpenGL based, at some point we hope to incorporate project news via RSS feeds and the like. For now it displays the basics.

I'll be updating the BOINC Alpha website with the instructions for testing the cookie attach process and new screensaver framework after this email goes out.

Here are the links to the experimental clients:
X86: http://boinc.berkeley.edu/dl/gridrepublic_6.7.5_windows_intelx86.exe
X64: http://boinc.berkeley.edu/dl/gridrepublic_6.7.5_windows_x86_64.exe
____________

Does this new unofficial client address any of the current bugs or do anything useful at all?

Does this new unofficial client address any of the current bugs or do anything useful at all?

Facebook? Ummmmm,......Facebook. What the hell does that have to do with anything? Sheesh. Things are not running well enough to try to attract people that have no clue what-so-ever. Am I missing something? Facebook? Really?

Facebook? Ummmmm,......Facebook. What the hell does that have to do with anything? Sheesh. Things are not running well enough to try to attract people that have no clue what-so-ever. Am I missing something? Facebook? Really?

From face value, this is all about trying to open untapped markets. It's probably connected to funding based on new sign ups per month. All Boinc needs to do is set up some Facebook groups under the usual banners of "save the world from disease", "save the world from global warming", "save the world from insert text" and people will sign up whether they understand it fully or not.

I'm really hoping they are not going to try and make it install on any machine they log on to, otherwise it is going on my ban list at work.

Rob

If there is a design intent, I have not seen it, none was published for comment so I don't have a good idea of what they are thinking (if they are thinking) ...

But, that seems to be the idea ... connect to facebook and magical things will happen ...

Personally I think what we will mostly do is create a whole new group of people that get a bad impression of BOINC...

For a user like myself it is not that big a deal ... but BOINC is no way ready for the non-technical person ... and everything that has been tried to make is so has been relegated to the sidelines by the UCB people...

So, we may get a few people that come in that way ... but I suspect that far more are going to come in, have problems, and leave ... soured on BOINC for all time ...

There are loads of SaH Classic people that never made the move to BOINC (for a variety of reasons, not the least of which that UCB did not pay attention to their concerns, and oh-by-the-way they were only interested in one project...)

If there is a design intent, I have not seen it, none was published for comment so I don't have a good idea of what they are thinking (if they are thinking) ...

But, that seems to be the idea ... connect to facebook and magical things will happen ...

Personally I think what we will mostly do is create a whole new group of people that get a bad impression of BOINC...

For a user like myself it is not that big a deal ... but BOINC is no way ready for the non-technical person ... and everything that has been tried to make is so has been relegated to the sidelines by the UCB people...

So, we may get a few people that come in that way ... but I suspect that far more are going to come in, have problems, and leave ... soured on BOINC for all time ...

There are loads of SaH Classic people that never made the move to BOINC (for a variety of reasons, not the least of which that UCB did not pay attention to their concerns, and oh-by-the-way they were only interested in one project...)

Some how I don't know If any users of Facebook will want to do Boinc(Dreamers and their dreams), Seems kinda pointless for them to do this, But We shall see.

Yeah and We don't need that, Of course at least We do have Boinc, F@H is just an app like S@H was(cpu and gpu), All manual control and such.

That's too bad about the people who didn't migrate to Boinc, They liked I gather something that required almost no setup to run.

Therefor they should set the default load to 50% or something. Usually the last 30% load pushes cooling systems towards their limits. I used to run boinc on a laptop with 65% CPU load (on all cores). The fan was audible, but still a lot quieter than at full throttle...

But I guess you could find just as many people who'd be upset if the default setting was not 100%.

MrS

It seems to me that the 6x batch of BOINC pretty much all have issues (many of which would rationally characterize all the releases as beta).

For me, unless the workstation is running CUDA (or the Milkyway optimized client for ATI GPU's - apparently anathema to the BOINC client development team), or one is running Windows Vista or Windows 7, the 5.4.5 client still seems a better choice.

That older client doesn't have near the issues getting work that seem to exist on all the 6.x clients if one has multiple projects and works just fine on notebooks - which don't have much CUDA to offer anyway.

I've not had problems running BOINC on notebooks, and tend to run them on 'fire and forget' projects like Climate.

In any event, I just wish that the 6x clients didn't punish folks in exchange for the ability to start up automatically on Vista/Win 7, and the ability to run GPU projects. As it is, in many respects all the 6x clients seem to have had (for me at any rate) an 'alpha/beta' feel to them which is seriously unfortunate.

Barry

As to the earlier comment on BOINC being fire and forget on an single or handful of projects ... well, yes and no ... One bug Richard is chasing is an "initialization" bug where in effect he has only two projects eligible for work fetch and then the one that actually should be selected is not selected because of a late disqualification ... well, then that means that the first project is scanned and SHOULD have been replaced by the more eligible project (which is disqualified late) but it is not ... so the wrong project is selected for work fetch ...

In that this is a "timing" and special case bug (though I suspect it is actually more common than is commonly known) the last fix has not yet been proven to be a miss ...

Anyway, the point is that as has been noted in another thread the developers have the mind set that they know more than the operators and mostly resist giving us the tools to better control the behavior of the client. Fundamentally this is one of those reasons the participants from SETI Classic did not move up with us... they liked only running SaH and having total control over how, when, where, why tasks were run using the available tools... so that the science application was run the way they wanted it run ... not how UCB wanted it run ...