Do I get a prize? I just got a long one and a short one sent to me. I've suspended all my other WU's and I'll let you know how long they take.

__________________
Old laser physicists never die, they just become incoherent.
These days, every Tom, Dick, and Harry thinks he knows what a photon is, but he is wrong. - Albert Einstein

OK, the new S5 short WU took about the same amount of time as the "long" S4 WU's took with akos's application (~49 minutes on an overclocked AMD 64 4000+). Since they are supposed to do twice as much work as the "short" S4 WU, they are clearly not quite as optimized as the old akos application (the "short" S4 WU's take ~15 minutes on the same machine), but they are an impovement on the old albert 4.37 application.

The new S5 long WU's look to be very long indeed. Taking the current CPU time and dividing by the current progress, it looks like that WU will take almost 8 hours (7 hours, 48 minutes), or nearly 9 times as long as the old "long" WU's.

Indeed. My fastest machine would not run out of work even if it only was sent "short" S5 WU's. However, I hope that they optimize the app a little more in the coming weeks.

__________________
Old laser physicists never die, they just become incoherent.
These days, every Tom, Dick, and Harry thinks he knows what a photon is, but he is wrong. - Albert Einstein

Was going to offer this assistant professorship I have lying around, but you don't seem to need it now...

Thanks for the great info on the timings; the only downside is that the advice I've been giving that now even slower machines can run E@H (using akosf) will have to be revised/withdrawn.

I don't know if the servers will send only short master files to slow machines, but if not, a ~500 MHz machine would need a good long while to do a long WU. It probably could make it under deadline if E@H is the only project on that machine?? But since I am now fond of running multiple projects, I've suspended E@H on my 550, but will increase the other boxes' proportion of E@H when they get S5 WUs.

The other issue of course is those of us used to akosf speed must be prepared for a decline in our RAC due to the new credit system. Some in the thread cited are saying a big decline.

But hey, this is all about S5 and the best chance to find gravity waves! And congrats to Laser Jock for being the first BAUTer to begin the search!

__________________
Ken

Visit the BAUT Team Page
BOINCview: Network control
and a superior UI!

Yes, that appears to be the case. Just extrapolating from that one WU (claimed and recieved 12.8 credits), I'll only get ~381 credits a day out of my fastest machine (compared to 580+ now). However all of the other teams RAC will go down accordingly. I also think the fixed credits/WU is a good thing. If you look at some the top computers you can see that many of them are using akosf's app and something else to boost their claimed credit. This should even the playing field for everyone.

__________________
Old laser physicists never die, they just become incoherent.
These days, every Tom, Dick, and Harry thinks he knows what a photon is, but he is wrong. - Albert Einstein

I'll be removing E@H from the 266Mhz Pentium 2. Akosf's app made it possible to run E@H on that machine. I'll go back to running LHC@H when I can get work. Anybody have any info on which projects are slow machine friendly?

__________________
"To excel in physics is to embrace doubt while walking the winding road to clarity." - Brian Greene

I believe Rosetta@H is slow machine friendly, since it runs repeated stochastic tests on the same assignment, it tunes the number of tests and complexity of the assignment to the speed of the machine and the number of hours per assignment you specify.

Ok, I'll partially retract that. They list 500MHz and 512MB as recommended minimums.

__________________
And the "driving on the freeway on a scooter" analogy still holds true because the pilots are sitting in 7 to 30 ton aircraft o' doom and you are running around them in your very own Meatbody, Mark I. Beep, beep.
Trying to make sense of computers, The Error Log.

The "long" S5 WU took 7 hours, 33 minutes on my overclocked AMD 4000+ or about 9 times as long as the average "long" S4 WU. It claimed 121.34 credits. As I said before, I hope they optimize the application some more.
SZTAKI@Home may be a good choice. We have a BAUT team now, and I have found that the WU take less time (and sometimes much less time) than the "long" S4 WU (with akosf's app) of E@H. However, the deadline is just a week.

__________________
Old laser physicists never die, they just become incoherent.
These days, every Tom, Dick, and Harry thinks he knows what a photon is, but he is wrong. - Albert Einstein

hmmm. I switched to the 3rd party app. Should I delete it and let boinc update E@H to the standard version? Or will it still update when a new albert is released?

__________________
http://boincwapstats.sourceforge.net.../style:2/p.png

You shouldn't have to do a thing (I haven't). You are automatically sent a new app for the new WU's (it's actually called einstein_S5R1 4.02). The old WU's still run on akosf's app, and both can be on your computer at the same time.

__________________
Old laser physicists never die, they just become incoherent.
These days, every Tom, Dick, and Harry thinks he knows what a photon is, but he is wrong. - Albert Einstein

Klaus, In addition to Henrik's and LJ's good suggestions, MalariaControl has pretty consistently short WUs, but account creation is closed ATM.

Also coming soon, hopefully, will be the beta version of Leiden Classical, a very nice project, IMO. Account creation is closed for now, but I started running the alphas some time ago and have had no problems; WU's always under 2 hours on my 550 MHz. So this will be a good "backup" project, especially for those who want to run mainly physical science apps. I will lobby for a team once it's officially beta.

And what Laser Jock said: the new S5 app will automatically install itself without conflict with akosf.

__________________
Ken

Visit the BAUT Team Page
BOINCview: Network control
and a superior UI!

ok, thanks guys.

Just got my first S5 (I think). Wow. Looks like it will take 22.5 hours on my P4 2.8 Ò_ó!

__________________
http://boincwapstats.sourceforge.net.../style:2/p.png

I'm seeing little difference in time on a Pentium 4 1.60GHz.
9778417 is a z1_xxxxx__xx_S4R2a: 8841.21 seconds
9885445 is an h1_xxxxxx_S5R1__xxxxx_S5R1a: 9002.58 seconds

Note: links only active for 2 weeks.

__________________
"To excel in physics is to embrace doubt while walking the winding road to clarity." - Brian Greene

Thanks for the info Klaus.

Note that that particular S5 WU is one of the short WUs:

"The short workunits have XXXX.X less than or equal to 0400.0"

Also for your slow box, in case you may have suspended it, LHC is back up with 63000 WUs to crunch, so you might want to set that machine's cache high to get a week's worth of work... That is, if LHC is still its only project.
___

I've been feeling bad 'cuz I'm still stuck crunching S4s, but according to the E@H status page, S5 is only 0.195 % done, so they'll be plenty of work for everyone.

__________________
Ken

Visit the BAUT Team Page
BOINCview: Network control
and a superior UI!

__________________
www.csphysmath.com
BAUT Team Stats

Congrats to azazul and all who are now crunching S5 ; both my machines now are, the 2.7 GHz has short ones & the 3.3 was just sent long ones. And thanks for posting your times...
___

Easy way to update Preferences:

azazul's Table of Links on the BAUT Team Page uses the fact that the URL for each project's "Your Account" is a PHP script, and so if you've allowed a cookie to automatically log you on, the same URL works for everyone, for example,

http://einstein.phys.uwm.edu/home.php

takes everyone directly to their E@H account.

It has taken me a long time to recognize/realize* that in fact most BOINC URLs work this way, and so quick shortcuts are available, most neatly for General Preferences:

http://einstein.phys.uwm.edu/prefs_e...?subset=global

will take everyone directly to the edit page of their general prefs!

vBull elides the above on my screen, though the URLs work, so the full address is:
Since GPs are indeed global, I mean, the same across all projects, and since most of the BAUT team crunch Einstein@Home, that one URL is all you ever need to change your General Preferences. (If you don't crunch E@H, just change the part before the slash to your primary project's Home Page.)

This may not seem like much, but if you ever have had to explain to someone how to do it...
Now you can just say, Go here.

You can elaborate endlessly: if the venue you want to change is Home:

http://einstein.phys.uwm.edu/prefs_e...bal&venue=home
(To see these changes immediately, I think you have to use as a base URL a project that actually uses the 'Home' venue, and manually Update from a computer that project is running on, though changes propagate to all your projects in a day or so.)

To edit Resource shares, append '&subset=project' instead of 'global,' like:

http://einstein.phys.uwm.edu/prefs_e...subset=project
Of course, for this one you certainly must use as the base address the URL of the project whose share you are changing.

___
*I only just realized this thanks to a post by CPDN project leader carl. Of course, it has been before my eyes, in the address bar, every time I've changed prefs, for well over a year --but never noticed it til now. My apologies to all who've known this all along...

__________________
Ken

Visit the BAUT Team Page
BOINCview: Network control
and a superior UI!

I just noticed that my 'unattend' running machine is working on S5 WUs also...
It's an 'AMD Sempron(tm) Processor 2600+' system with following times:
akosf S4 short: 400/900 sec (there are 2 types of short WUs)
akosf S4 long: 3000 sec
new S5 short: 4650 sec
new S5 long: 43800 sec!

seems that my 'daily quota problem' (the mentioned computer had only 4h of work with 32 short WUs) is over...
this computer is customized to only fetch 1 WU per time (0.05 days for 'connect every' parameter) and is crunching mixed WUs, remaining S4 and new S5 WUs...

__________________
Udo
Udos BOINC Stats

I still only have S5 on the one computer. If times are consistant with Laser Jock's observations, then a long S5 will take about 24 hours to finish.

__________________
"To excel in physics is to embrace doubt while walking the winding road to clarity." - Brian Greene

So if I decide SETI has had enough of my time and decide to switch over to something else, what do you think is the best "cause" to support? (I think SETI is cool, but I doubt they're going to find anything the way they're looking.)

__________________
Everything I need to know I learned through Googling.