S39L Observation Thread

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

RE: 1. I stopped boinc 2. I

Message 26383 in response to message 26381

Quote:

1. I stopped boinc
2. I made a backup of crunch3rs version, including the .pdb file.
3. I deleted crunch3rs files
4. I copied your version to the folder
5. I started boinc

Maybe I should have left the .pdb file in place?

Why so important PDB?
I did three things:
1. stop BOINC
2. copy files to project directory
3. start BOINC
It was enough to me. I think it would be enough to you too.

Santas little helper
Santas little helper
Joined: 11 Feb 05
Posts: 36
Credit: 4699218
RAC: 0

RE: (I have a full list of

Message 26384 in response to message 26380

Quote:
(I have a full list of humanitarian DC projects here)

According to physics non-BOINC projects the MUON1 Distributed Particle Accelerator Design is quite interesting and I didn't find it at your list. :-)

Greetings, Santas little helper

Pav Lucistnik
Pav Lucistnik
Joined: 7 Mar 06
Posts: 136
Credit: 853388
RAC: 0

Can anyone post S39L client?

Can anyone post S39L client? Akosf's website is dead right now...

[Edit] Ok, got it, website is alive again.

archae86
archae86
Joined: 6 Dec 05
Posts: 3145
Credit: 7023854931
RAC: 1804616

Regarding akosf's SETI ap as

Message 26386 in response to message 26379

Regarding akosf's SETI ap as found at:
SETI K7-08

Quote:
Quote:
akosf, I'd love to try this.
So, installation process:
1. stop BOINC
2. copy seti...exe and app_info.xml to the project directory
3. start BOINC

OK, I've installed your SETI science ap on two machines previously running crunch3r's ap, using your specified installation method. It happens that neither had an in process result, so it was a fresh start.

The P4 EE HT machine example completed in a CPU time essentially identical to recent results downloaded in the same batch. Well within the noise, so I can't tell if it might be somewhat faster or (dare one imagine it) slower.

The Banias (Pentium M) machine took 2900 seconds to do a result in a batch that has been tightly clustered around 3500 seconds. This gives more hope that a real improvement was observed, though fluctuation in required CPU among result plus changes in the conditions of the use of the machine certainly are also suspects.

The Banias result quorum has three returned results, but the validation process has not yet delivered a verdict. You can see it at:
Banias result
The Gallatin result is in the same state:
Gallatin result

It may be relevant that my Banias has a 32k L1 data cache, as compared to only 8k for the Gallatin. Possibly it likes akosf's tuning better.

On the other hand result to result variation in SETI is even more a fact of life than Einstein, so I'd not put a lot of concern here.

I'd hate to deflect akosf from Einstein improvement to SETI improvement--I think Einstein has more need and better use for incremental computing capacity at this time.

(edited a few minutes after initial post to say the SETI message board is reporting recent significant validation problems and delay--this may be a tough moment to add evidence on this question)

RandyC
RandyC
Joined: 18 Jan 05
Posts: 6006
Credit: 111139797
RAC: 0

RE: Regarding akosf's SETI

Message 26387 in response to message 26386

Quote:
Regarding akosf's SETI ap as found at:
SETI K7-08
Quote:
Quote:
akosf, I'd love to try this.
So, installation process:
1. stop BOINC
2. copy seti...exe and app_info.xml to the project directory
3. start BOINC
OK, I've installed your SETI science ap on two machines previously running crunch3r's ap, using your specified installation method. It happens that neither had an in process result, so it was a fresh start.

Tried installing the above on my XP 1600+ system. The in-flight WU completed and seems to be good, but the following 6 WUs all errored out:

***UNHANDLED EXCEPTION****
Reason: Access Violation (0xc0000005) at address 0x00509404 read attempt to address 0x0210FFC01: 03/24/06 20:11:37

Backed it off immediately.

Seti Classic Final Total: 11446 WU.

The Uncle Bs
The Uncle Bs
Joined: 5 Mar 05
Posts: 6
Credit: 3239754
RAC: 0

RE: Well, I examined some

Message 26388 in response to message 26351

Quote:

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

Code of CPDN is worthless. (just my opinion)
I think a better compiler would be able to increase its speead about 30-50%. (careful estimate)
The handoptimizing is a bit hopeless because of very big code and the very slow feedback. (validation)

First off, akosf thank you for all your skilled talent(i think you have a bright future ahead of you). And all your dedicated time to tweaking this platform for Einstein, plus answering all the questions that have followed.

If you ever find the time to tweak CPDN, that project has the longest work unit times, thus less intrest from users. I`m sure that if the times became more in-line with Seti/Einstein times, more people would spread their cpu time between all project. Just a thought. Anyways akosf, thanks again.

archae86
archae86
Joined: 6 Dec 05
Posts: 3145
Credit: 7023854931
RAC: 1804616

RE: Regarding akosf's SETI

Message 26389 in response to message 26386

Quote:
Regarding akosf's SETI ap as found at:
SETI K7-08
The Banias (Pentium M) machine took 2900 seconds to do a result in a batch that has been tightly clustered around 3500 seconds. This gives more hope that a real improvement was observed, though fluctuation in required CPU among result plus changes in the conditions of the use of the machine certainly are also suspects.

Now my second Banias result has completed--it reported a CPU time of 3565 seconds, right in the tight band of the previous crunch3r results. So my hope that this might give .83 CPU ratio on the Banias machine is considerably attenuated.

On a more positive note, the initial Banias result has now validated. So has my initial Gallatin result.

So I've upped the SETI resource share on these two machines from 5 to 15%, and will report in a day or two when the execution time data look a bit clearer, or sooner if I see a problem.

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

RE: RE: RE: SETI has

Message 26390 in response to message 26370

Quote:
Quote:
Quote:
SETI has some very good optimised code.
I spent an afternoon to optimise it, and I could get just 10% improvement. (Crunch3r's Athlon XP SSE 2.09)

You mean, you had squeezed out another 10% out of a binary from Crunch3r? *wow*

Can I download that modified binary somewhere ;) ? Ten percent better is a good reason for me to try it :))

I upload to here. It was optimised to Duron (64+64kB cache), so don't rely on better time on other processors.
But, I think it will catch up the SSE3 code on A64 and P4.


Pentium4 SSE3 on Windows! :) Gimmie gimmie gimmie!

me-[at]-rescam.org

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

RE: Pentium4 SSE3 on

Message 26391 in response to message 26390

Quote:
Pentium4 SSE3 on Windows! :) Gimmie gimmie gimmie!

You can download the SSE3 optimized SETI application from here.

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

RE: The Banias (Pentium M)

Message 26392 in response to message 26386

Quote:

The Banias (Pentium M) machine took 2900 seconds to do a result in a batch that has been tightly clustered around 3500 seconds. This gives more hope that a real improvement was observed, though fluctuation in required CPU among result plus changes in the conditions of the use of the machine certainly are also suspects.

It may be relevant that my Banias has a 32k L1 data cache, as compared to only 8k for the Gallatin. Possibly it likes akosf's tuning better.

I suggest to you to run the SSE2 optimised code on your Banias, because K7-08 was optimised for Duron (K7 core with SSE) and Banias has better SSE execution unit. So, I think K7-08 won't be faster.

Comment viewing options

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