My Linux boxes say a Good Bye to Einstein@Home.

gnaag
gnaag
Joined: 19 Apr 05
Posts: 3
Credit: 2,704,450
RAC: 0

@smith: I joined this

@smith:

I joined this project, because I'm truly interested in the scientific outcomes. That's the main reason, why I'm still crunching E@H WUs on my Windows boxes and will also do it again with my Linux box, when there's an optimized Linux app.

But I'm also interested in some other DC projects. akosfs magic optimizations of the Windows app raised the question what's the Linux albert is actually doing on my box. It seems, that it's doing 25% science and 75% heat, entropy, whatever. I think, that's not very efficient. I'm not angry or something like that, I'm just trying to produce the most science per compute cycle with my computers.

There are other projects out there, like the mentioned QMC or LHC@Home and so on, which are able to use my Linux box more efficiently IMHO than Einstein@Home and which are also important and interesting. I'm switching my boxes to the projects, where I think my boxes can be the most help and can do the most science. I'm not switching them off, I'm switching projects. ;-)

All I can say is, that I'm very proud that I can attend to this great project. It's one of my favorites and will ever be. But the Linux application seems to have a poor performance and that's why I'm switching my Linux box to somewhere else, where the applications squeek more performance and science out of my hardware. There's no bad blood, it's just economics. ;-)

gravitysmith
gravitysmith
Joined: 8 Nov 04
Posts: 55
Credit: 77,933,025
RAC: 11,366

RE: But I'm also interested

Message 27509 in response to message 27508

Quote:
But I'm also interested in some other DC projects. akosfs magic optimizations of the Windows app raised the question what's the Linux albert is actually doing on my box. It seems, that it's doing 25% science and 75% heat, entropy, whatever. I think, that's not very efficient. I'm not angry or something like that, I'm just trying to produce the most science per compute cycle with my computers.

Okay. I think I am getting the picture now (correct me if I am wrong). We probably all join a project with an assumption that the science app is making the best possible use out of our computers. With Akosf's optimizations that assumption is shown not to be true. Akosf is a very clever guy, and pointed out some things the devs probably never thought of. Since it is likely an honest oversight, you give the devs the benefit of the doubt and give them a chance to catch up. I guess the complaint now is that while the science app has been improved by the devs, they have not released it yet.

Since you are interested in other projects, you are likely thinking that in terms of absolute number of work units completed for E@H, the ones you are not doing now will be more than made up for after the new app is released. Or in other words, you could crunch for 4 months at the current rate, or wait until the new app is released and crunch 1 month to do the same amount of science.

Of course the "most" units would be crunched for this particular project by keeping your computers connected to E@H, but given that you are involved with other projects, some sacrifices have to be made.

Thank you for the explanation! I think my preferences/sacrifices are different from yours (probably the reason I was having trouble understanding your motives), but at least it makes sense to me now now. Happy crunching!

Ziran
Ziran
Joined: 26 Nov 04
Posts: 194
Credit: 60,800
RAC: 0

Hear is how I se

Hear is how I se things.
Imagine you have a small farm of 3 equal spec computers 1 Mac, 1 Linux and 1 windows machine. You want to share your resources equal with 3 projects. So you install BOINC on them and give the 3 projects 1/3 of the resources form each machine. This is the textbook example of how to install BOINC. But then you start thinking if you somehow can make those computers do more science. So you compare the standard crunch time per project on each computer. If you then find that the MAC is relatively best suited to crunch project A, Linux project B and Windows project C, wouldn’t it then be better to let the computers do the project that they was best suited for 99.9% of the time? This way you still would be giving each project 1/3 of your resources, but all projects would get more work done with those resources.

Until Akosf released his optimized application, the Mac’s was best suited to work on Einstein, so many farmers set a high resource share on there Mac’s and removed it from there windows machines.

BOINC was made to make it possible to donate resources to different projects. But you should maybe also think of how you donate you resources over time. Ether you can for the next 2 months do 2 projects at the same time, or you could do one project at the time for one month each.

Lets take a look at the current situation. Akosf’s optimized application is 4x faster then the standard application. A new standard application is ready, but isn’t distributed because the servers isn’t capable of handling the anticipated load. So at the moment, this project have more resources then it can optimally use. So if you are contributing or thinking of contributing to other projects, this is the time to do so.

Then you're really interested in a subject, there is no way to avoid it. You have to read the Manual.

B52
B52
Joined: 19 Feb 05
Posts: 45
Credit: 273,899
RAC: 0

RE: RE: Hi

Message 27511 in response to message 27497

Quote:
Quote:

Hi Michael

Did'nt wanna cut in all off this, but as I see your objections, then its not at all about a turboalbert, but about the Boinc CC ??

It's both, waste of energy on my Linux boxes and of cause cc is a problem. When I, some months ago, brought my first Linux box in the project (A64), I was a bit ashamed about the low cc, which sometimes pulled down other users. Compiling a 64Bit Boinc Client without any tricks nor any optimizations raised the cc from 3x to about 100. The WUs have been longer at that time and the app listend to the name einstein. The high cc was caused by a rather bad Linux app., with the launch of albert the cc almost normalzed to about 60.
Now my XP@2200MHz(Win) needs about 50min and the A64(Linux)@2400 about 4h for one of the long WUs. That's just dissapointing.
And yes, I'm in a top team and crunching is also a lot of fun for us.
Quote:

Meaning that in your opinion, then the CC should give equality to all.

Yes, afaik that is planned anyway on the base of counting the fp-ops. I think that's a good idea, because _if_ there is a virtual payback system, it has to be fair.
And from the ecological view it doesn't make sense to lure every old CPU 'that can't climb the next tree fast enough'*), into the project.

cu,
Michael

*) free translation of a German joke. ;)

So its still a Boinc CC matter

bonnyscott
bonnyscott
Joined: 4 Feb 06
Posts: 6
Credit: 5,373,362
RAC: 0

RE: When you joined E@H

Message 27512 in response to message 27506

Quote:
When you joined E@H (before Akosf's new apps?), presumably you thought the science was worth the electricity, costs, etc. With the updated apps, people running the Linux boxes (and us users waiting for the official version) are still doing the exact same work as before. If it was worthwhile before, why isn't it now?
Quote:

Ok, I'm trying to keep it short.

Of course the science was, and is, still worthwile. No matter.

But I'm a *frustrated* Linux- user, as well.

"That little genius", named akosf, has brought the chance to the project to do the examination of the collected data *within a quarter* of the estimated time.

And Dr. Allen has promised to implement akosfs improvements "asap" into the "official" app.

Shortly after that: "Oops. Probably our servers wouldn't stand the traffic. I'll be back soon".

Since, *not the slightest* "huff" or "moo", or whatever, from the "official" part.

As far as I know the "mills of bureaucracity" at universities, and elsewhere, the "guys in charge" are "frantically" trying to get the Ok for hardware upgrades, and someone to pay the bills for that, and are surely getting nowhere.

"Mere politics".

Maybe, E@H- participiants should start "mailbombing" the uwm- elders. Or the congressman in charge. Or whom ever. Even "Homeland Security", if it helps, in any way.

From the point of view of the Linux- user, it currently feels like: Every cpu- cycle for E@H is a "planet heater", to quote akosf again, because 4 cycles are needed, where a "bloody Win- box" needs one.

In regard of the use of "elsewhen idle" cpu- cycles slightly ridiculous, of course.

But, not to forget: The "sportsman"- angle of view, btw.

The threadstarter is the teamleader of one of the "main contributors".
Just to drop mention. ;)

And in regard of *all the other* scientific projects, which are "literally scrambling" for some additional computing power, there remains the "bitter taste of waste", at the moment, to the Linux / Unix user.

And none of the "officials" even bothered to drop a line, yet. Afaik.

"Even if" he has nothing else to say as: We're trying to upgrade our machines, but can't afford.

Would give users "the chance" to mailbomb someone. Or start a collection. Or whatelse. At least, Linux and other Unices are a good 10 % contributors here.

Well, "there is" a risk to the career of the official, who speaks out loudly.

Maybe *that's* the point.

However. Frustration remains.

Nonetheless, happy crunching, all folks.

Bonnyscott

ps: Ahrrg!

Again a "complete new way" to format forum- postings!
What the heck?
Not now!

bonnyscott
bonnyscott
Joined: 4 Feb 06
Posts: 6
Credit: 5,373,362
RAC: 0

RE: So its still a Boinc

Message 27513 in response to message 27511

Quote:

So its still a Boinc CC matter

Not at all.

It's a question of shere computation- cycle and energy- economy.

Is that "too hard to get" for an US- citizen?

BTW: "Bomber", or "Hairstyle"?

Europeans might just react "touchy" to your nick!

Regards, bonnyscott

Pooh Bear 27
Pooh Bear 27
Joined: 20 Mar 05
Posts: 1,376
Credit: 20,312,671
RAC: 0

RE: [BTW: "Bomber", or

Message 27514 in response to message 27513

Quote:
[BTW: "Bomber", or "Hairstyle"?


When I see B52, I think of the band.

Erik
Erik
Joined: 14 Feb 06
Posts: 2,815
Credit: 2,645,600
RAC: 0

RE: Is that "too hard to

Message 27515 in response to message 27513

Quote:

Is that "too hard to get" for an US- citizen?
BTW: "Bomber", or "Hairstyle"?

That was callous and untoward. Btw he/she's from Denmark.

(Congrats Zaran for the most articulate post so far)

bonnyscott
bonnyscott
Joined: 4 Feb 06
Posts: 6
Credit: 5,373,362
RAC: 0

RE: RE: Is that "too

Message 27517 in response to message 27515

Quote:
Quote:

Is that "too hard to get" for an US- citizen?
BTW: "Bomber", or "Hairstyle"?

That was callous and untoward. Btw he/she's from Denmark.

Ok. Done wrong. Ashes on my head. Accepted.

But that was "btw".

Still waiting for some "official anouncements" here.

Regards, bonnyscott

M. Schmitt
M. Schmitt
Joined: 27 Jun 05
Posts: 478
Credit: 15,872,262
RAC: 0

RE: So its still a Boinc CC

Message 27518 in response to message 27511

Quote:
So its still a Boinc CC matter


50/50

http://einsteinathome.org/goto/comment/26173

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.