I had no idea, that is very cool.

E@H is my favorite, so I am happy the team does so well in it. I am waiting on word from a graduate school where if I get accepted, I intend to study gravitational waves. I would never have gotten into it had I not looked more into E@H.

I look forward to seeing our team and the projects themselves continue to grow in the future.

__________________
www.csphysmath.com
BAUT Team Stats

By my count we have averaged ~6,900 credits/day on E@H.

I had it marked on my calender to make note of this day, but you beat me to it.

__________________
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

Wow, a year onboard. Happy crunchiversary.

Instructions for using an other computer as 'network station' (Windows tested):

Remove all BOINC activity from 'network station' (NS):
- uninstall BOINC client (if previously installed)
- delete BOINC folder from filesystem / harddisk
Install same BOINC client as on 'stand-alone computer (SAC)', use same parameters as on SAC (in my case, 'run as service')
Do *not* attach to any project.
Stop all BOINC activity on NC (stop the service if necessary)
Copy the BOINC folder (c:\program files\BOINC) to a safe place (we will call this folder 'BOINC-initial')

Stop BOINC activity on SAC:
- Commands, SUSPEND
- Commands, Network activity suspended
- stop service / program and BOINC Client
Copy BOINC program folder from SAC to NS, do not copy files boinc* (4 files), *.dll (8 files with boinc.dll counted both times) and master.html
start BOINC program on NS (start service or by starting the BOINC client)
Connect NS to the internet
- commands, run based on preferences
- commands, Network activity based on preferences

all outstanding up-/downloads are performed on NS

Stop BOINC activity on NS:
- Commands, SUSPEND
- Commands, Network activity suspended
- stop service / program and BOINC Client

Now you can copy the files back to the SAC and reactivate computing. If you have to copy several SAC, just use 'BOINC-install' folder on NC and add the neccessary files from the SAC.
It's important to stop the service/program on both sides before copying the files.

As a result the Computer name is set to the name of the NS, which can only be seen by yourself. But if you like, you can change the name of your NS before you connect to the internet (reboot required).
In my case you can see that all office hosts are now Windows XP (which is my private notebooks operating system at home).
I did this procedure to deliver all my outstanding WUs but it is to costly to continue with all computers in this way.
I think I will do it for my office PC I'm working with (217888) and for one powerful fileserver (470358).

Udo

p.s. in this way you could cheet: download lot of WUs and then distribute them to different computers.

__________________
Udo
Udos BOINC Stats

Thanks for the information Udo, if anyone has a problem similar to yours, I know where to point them. I am glad that you found a way to continue crunching at work with at least some computers.

__________________
www.csphysmath.com
BAUT Team Stats

"LIGO Kicks into High Gear for Gravitational-Wave Search with 18-Month Observation Run"
http://www.spaceref.com/news/viewpr.html?pid=19142

"Now that the LIGO is sensitive enough to detect changes in distance a mere thousandth the diameter of a proton, Marx adds, the science return should be even greater. Recent results from the Swift satellite pinpointing the location of short gamma-ray bursts (GRBs) have also heightened astronomers' interest in the results from LIGO's current observational run.

The current 18-month science run could lead to even more important discoveries, and if nature is very kind, to the first direct detection of gravitational radiation since Albert Einstein predicted the phenomenon's existence in 1916."

__________________
jwj

It's ok not to know. We should try harder to find out.

I just made a post in General Science about the newly started MalariControl.net project. If you are interested in this very worthwhile project, act fast as account creation will close at 500 for the beta, and they are already over 200 at time of writing.


Jerry,

Sorry so long to reply thanks very much for the post and the link. . I do want to relate this to the Einstein@Home part of LIGO. The quote refers to the main LIGO S5 data collection now underway, which is at the design sensitivity of LIGO.

In our pulsar search, Einstein@Home crunchers are still working on S4 data ATM. As Wolverine posted earlier, there were about 200,000 WUs to crunch from that run. The server status page now indicates we are 35% of the way through this data set. Presumably we will then get to crunch the S5.

I haven't double-checked this, but if memory serves, S5 will be essentially the last data until LIGO is (possibly) upgraded in (possibly) 2010.

And we must hope for gravity wave detection at least by by then, hopefully with S5, since the LISA satellites mission, as all future NASA science, seems to have an uncertain future, if I (possibly) remember correctly.
__________

Also, many thanks to azazul, the links table on the BAUTstats site has been updated to include all the current BAUT projects, so it is your one stop for links to stats, account preferences, etc.

__________________
Ken

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

This is from an older post over at LHC@home:
(emph added)

Dr Klem also contributes other info of interest about the progress of the building of the LHC in the short thread, which is a very good read.

It's my personal opinion that right now LHC@home is the BOINC project providing the best scientific "bang for the crunch." I mean, the Large Hadron Collider is a major facility and you and I are helping design it.

This doesn't mean necessarily that more people should join the project, it has enough crunchers to easily go through the studies they are doing, so much so that it is often out of work.

Einstein@Home is probably still the place where more resources are current most needed, since it is still grinding somewhat slowly through the S4 data, as mentioned by Wolverine previously.

But of course, no project has yet found a G-wave, cured cancer or malaria, calculated an orbit, or happened on an alien. It may indeed take one of the above to give Distributed Computing a lift from the doldrums it seems to be in.

But it is nice to hear that at least the LHC@home project has contributed in a concrete way to scientific progress.

__________________
Ken

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

And congratulations to PeterFab for passing the 100k E@H milestone!

I add my congrats to PeterFab and steal Nereid's text to say:

And congratulations to Laser Jock for passing the 100k E@H milestone!

__________________
Ken

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

Thanks. It looks like you'll be the next one to cross that mark.

__________________
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

For about two weeks now I've been running a 3rd party optimized Einstein@Home Albert application program, the C37 - optimised x86 compatible windows executable, made by akosf.

As it says, this version is for Windows and Intel only -- but see end of this post for AMD.

On my machines, it has more than halved my computation time! (For the same master file.) Although credit claimed per work unit is likewise halved, your overall credit granted will most likely increase because each WU is being verified by slower boxes -- likely now much slower.

For example, my E@H RAC has increased from 165 before, to 273 now. This with no change in resource allocation; no computation errors of any kind; no lock-ups etc; temperature of machines the same as before.

Now the thought of a "3rd party" "optimized" app is a little scary, but the thread on the E@H board where the app was announced is prominently stickified, and it is certain it would have been removed by the E@H brass if there were problems with these apps.

Of course, you may prefer to wait until an improved app is officially released (if one is). But then, benefits to you personally and the BAUT team would be nil, since if everyone sees the same increase, credit for all will be the same as before.

In any case, the amount of science done by using optimized apps is greatly increased: you are simply returning the WUs much faster. Although probably the biggest benefit will be for newer computers, the speed increase offered by C37 may make it feasible for some older machines, or ones that can't be run 24/7, to run Einstein@Home now. Deadlines need no longer be a problem!

Here's all you have to do:

C37 - optimised x86 compatible windows executable

Install:

1) download the C37 file, here.
2) stop BOINC -- be sure the tray icon is exited too.
Optional: rename the file 'albert_4.37_windows_intelx86.exe' in case of a problem
3) unzip file to ../BOINC/projects/einstein.phys.uwm.edu/ folder
4) restart BOINC -- answer yes if your firewall asks if it's OK.

There is no need to wait for your WUs to finish: running WUs resume just fine after the change.

It really is that simple, and I'm sure the results will please you & greatly help the team!!
_________

As you can see in the above referenced thread, there is now an akosf-optimized app for AMD: the D40.
I don't have an AMD, would appreciate a report from anyone who tries this.

I'm currently trying out the S39L, which uses Intel's SSE instructions; will let you know results. If not, there's the C40, which promises.... Verily, I am going down the darkening path.

Whatever, enjoy your crunching!

__________________
Ken

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

I'm still running my same old XP2200+ ... and I'm feeling adventurous today, so I'll give it a whirl.

I installed the add-on and it seems to be working perfectly. Talk about an improvement...

Average CPU time of my 5 WUs prior to install: 17442.68 sec. (4.85 hours)
CPU time of the first WU completed after install: 5,044.53 (1.4 hours)

Yowzah! Thanks so much for the heads-up on this one, Ken! I'll post an update tomorrow.

After Wolverine's success, I think I will give it a try as well, later today. Thanks for the information Ken.

__________________
www.csphysmath.com
BAUT Team Stats

You are both very welcome!

That kind of improvement on AMD stuns me actually, as we've all along heard that E@H was FPU intensive and since AMD is known to have a much better FPU than Intel, I would have thought there'd be less room for improvement. But figures don't lie, and Wolverine's latest WUs are running at ~4350 sec, so nearly a quadrupling of speed. Awesome.

Just a minor cautionary note, pending his update: all his recent results are still pending, so there's a remote possibility of invalid results (I have still had no invalid ones using the Intel optapps.) Indeed, your pending credit will likely balloon at first if you try one of these, since the other machines in your quorum will be literally struggling to keep up. And to repeat what I said earlier, the complexity of the credit granting process will mean the ultimate improvement in your credit will not nearly match the speed increase.

But mine has been pretty good, now 295 RAC vs the original 165; and this before I changed from the original C37 to the S39L, the one optimized for newer Intels using the SSE.

The S39L has made for an additional 25% or so improvement over the C37; it is hard to compare directly as both my machines have moved to different master files (how could they not, doing ~20 WU/day?) and so results now aren't directly comparable with my original times. But over both my machines, it is about triple the speed. Several results with this newer app have now validated, and like the earlier C37, S39L has worked flawlessly on both my machines.

For convenience I'll put here the direct links to download the newer akosf apps:

For AMD: D40

For Intel: C40 (10% faster than original C37, though I have not tested it.)

For Intel with SSE: S39L (~25% faster than C37; works great for me. As I undersand it, even Pentium IIIs have a form of SSE, so this can be used by most of us...)

There are also now several mirror sites, for example, here .

Have fun doing more science & I look forward to more results.

__________________
Ken

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

__________________
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