S39L Observation Thread

Digitalis
Digitalis
Joined: 9 Feb 05
Posts: 18
Credit: 90764
RAC: 0

I visited the message boards

I visited the message boards yesterday for the first time in a few months and was intrigued to see that there was an optimised client for Einstein. After reading through the various threads I replaced the standard Albert4.37 app with S39L on a P4 Prescott @ 3.2Ghz. Processing times have dropped from an average of 30k secs/wu to about 6.5k secs/wu. Some of the newer ones have been validated already so all looks well. Amazing indeed!

Nothing But Idle Time
Nothing But Idl...
Joined: 24 Aug 05
Posts: 158
Credit: 289204
RAC: 0

RE: I visited the message

Message 26414 in response to message 26413

Quote:
I visited the message boards yesterday for the first time in a few months and was intrigued to see that there was an optimised client for Einstein. After reading through the various threads I replaced the standard Albert4.37 app with S39L on a P4 Prescott @ 3.2Ghz. Processing times have dropped from an average of 30k secs/wu to about 6.5k secs/wu. Some of the newer ones have been validated already so all looks well. Amazing indeed!


And this should be strong encouragement to read the msg boards more often; you never know what transpires and how it effects you.

Darth Vader
Darth Vader
Joined: 11 Feb 05
Posts: 7
Credit: 51643
RAC: 0

RE: Well, I examined some

Message 26415 in response to message 26351

Quote:
Well, I examined some project applications some months ago. But my favourite is Einstein@Home.


Few things are still occupying my mind.
1) Did you (also) take a closer look at LHC@H regarding optimization?
2) When will there be a SSE2 E@H app.?

I hope I'm not too demanding. ;)


Reuben Gathright
Reuben Gathright
Joined: 15 Feb 06
Posts: 23
Credit: 6851215
RAC: 0

RE: RE: Well, I examined

Message 26416 in response to message 26415

Quote:
Quote:
Well, I examined some project applications some months ago. But my favourite is Einstein@Home.

Few things are still occupying my mind.
1) Did you (also) take a closer look at LHC@H regarding optimization?
2) When will there be a SSE2 E@H app.?

I hope I'm not too demanding. ;)


Oh not all dear, to make SSE2 work, he just needs to pony up $$$ to pay for a compiler.

From your post on the SETI forums about donations... I do not think you are willing to help either. >:( http://setiathome.berkeley.edu/forum_thread.php?id=29103&nowrap=true

Darth Vader
Darth Vader
Joined: 11 Feb 05
Posts: 7
Credit: 51643
RAC: 0

[way off topic] RE: From

Message 26417 in response to message 26416

[way off topic]

Quote:
From your post on the SETI forums about donations... I do not think you are willing to help either. >:(


To me there is no difference between people who own some green badge and people who don't. Everybody contributes in a way of their ability. Period.
And if you feel comfortable by minusing, well, have fun.
[/way off topic]


Misfit
Misfit
Joined: 11 Feb 05
Posts: 470
Credit: 100000
RAC: 0

RE: I visited the message

Message 26418 in response to message 26413

Quote:
I visited the message boards yesterday for the first time in a few months and was intrigued to see that there was an optimised client for Einstein. After reading through the various threads I replaced the standard Albert4.37 app with S39L on a P4 Prescott @ 3.2Ghz. Processing times have dropped from an average of 30k secs/wu to about 6.5k secs/wu. Some of the newer ones have been validated already so all looks well. Amazing indeed!


Yes, you're Prescott will love 39L :)

me-[at]-rescam.org

Akos Fekete
Akos Fekete
Joined: 13 Nov 05
Posts: 561
Credit: 4527270
RAC: 0

RE: RE: Well, I examined

Message 26419 in response to message 26415

Quote:
Quote:
Well, I examined some project applications some months ago. But my favourite is Einstein@Home.
1) Did you (also) take a closer look at LHC@H regarding optimization?
2) When will there be a SSE2 E@H app.?

1, Not yet.
2, I don't know. SSE2 would be good for a 1-2% faster code. :-)

JardaM
JardaM
Joined: 21 Mar 05
Posts: 9
Credit: 2555930
RAC: 0

RE: RE: 1) Did you (also)

Message 26420 in response to message 26419

Quote:
Quote:
1) Did you (also) take a closer look at LHC@H regarding optimization?
2) When will there be a SSE2 E@H app.?
1, Not yet.
2, I don't know. SSE2 would be good for a 1-2% faster code. :-)


Unfortunately, they won't allow you to. I posted the question two weeks ago on L@H crunchers corner. Considering tremendous amount of data that LHC will produce this is the right time to consider ways to make processing effective.

The answer was that the program is written in Fortran and that it would be difficult to make any comparison between their official and an optimised version.

Actually, just now I came to the idea that they didn't understand my question. I was not interested in optimizing L@H (as they probably understood) but in new application that must come to process data from Collider. L@H has a limited lifetime but the application I have in mind would/ will be a real challenge.

Digitalis
Digitalis
Joined: 9 Feb 05
Posts: 18
Credit: 90764
RAC: 0

RE: RE: I visited the

Message 26421 in response to message 26418

Quote:
Quote:
I visited the message boards yesterday for the first time in a few months and was intrigued to see that there was an optimised client for Einstein. After reading through the various threads I replaced the standard Albert4.37 app with S39L on a P4 Prescott @ 3.2Ghz. Processing times have dropped from an average of 30k secs/wu to about 6.5k secs/wu. Some of the newer ones have been validated already so all looks well. Amazing indeed!

Yes, you're Prescott will love 39L :)

For clarity I should say that since its a HT processor thats 2 wu's every 6.5k secs , so a real average would be 3.25k secs/wu.
If I was one of the developers of Einstein I'm not sure if I would be pleased about this development or embarrassed. From other posts it seems that this client will be adopted as the official one so it must be more of the former and not the latter. Kudos to the Einstein team.
On a related note is BOINC capable of distributing different version of the science app based on the client cpu type?

Raistmer*
Raistmer*
Joined: 20 Feb 05
Posts: 208
Credit: 177636077
RAC: 220514

RE: 00109404: 8B 4C DE 80

Message 26422 in response to message 26406

Quote:
00109404: 8B 4C DE 80 83 EB 10 75 F3 89 EA 90 ->
00109404: 0F 18 44 DE 80 83 EB 10 75 F2 89 EA
SETI_K78b


Ok, it seems I returned too late to download this update so I applied patch by hands. Patched version finished current WU successfully but all others finished with error immediately. So this version is not for Athlon XP.

Quote:
Well, I modified, but I'm not interested in SETI.

Very sad news for SETI crunchers :( Your SETI_K78 outperforms both Crunch3r's SSE2&3 versions on all Athlon 64/ Pentium 4 I have to install it.
It's amazing, thank you very much!
Maybe you would consider possibility of creation 3DNow! optimized client for SETI too? Your D-40 for Einstein is great....
Thank you again.

Comment viewing options

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