Posts by DaBrat and DaBear

1) Message boards : Number crunching : Help with Optimized AP (Message 801418)
Posted 23 Aug 2008 by Profile DaBrat and DaBear
Post:
LOL really need to pay attention to hopw that reads... Thanks to Joker this is my finished file and the 6.03 is running better.... well for now

<app_info>
<app>
<name>astropulse</name>
</app>
<file_info>
<name>astropulse_4.35_windows_intelx86.exe</name>
<executable/>
</file_info>
<file_info>
<name>libfftw3f-3-1-1a_upx.dll</name>
<executable/>
</file_info>
<file_info>
<name>ap_graphics_4.35_windows_intelx86.exe</name>
<executable/>
</file_info>
<app_version>
<app_name>astropulse</app_name>
<version_num>435</version_num>
<file_ref>
<file_name>astropulse_4.35_windows_intelx86.exe</file_name>
<main_program/>
</file_ref>
<file_ref>
<file_name>libfftw3f-3-1-1a_upx.dll</file_name>
</file_ref>
<file_ref>
<file_name>ap_graphics_4.35_windows_intelx86.exe</file_name>
<open_name>graphics_app</open_name>
</file_ref>
</app_version>
<app>
<name>setiathome_enhanced</name>
</app>
<file_info>
<name>AK_v8_win_x64_SSE3x.exe</name>
<executable/>
</file_info>
<app_version>
<app_name>setiathome_enhanced</app_name>
<version_num>528</version_num>
<file_ref>
<file_name>AK_v8_win_x64_SSE3x.exe</file_name>
<main_program/>
</file_ref>
</app_version>
<app>
<name>setiathome_enhanced</name>
</app>
<file_info>
<name>AK_v8_win_x64_SSE3x.exe</name>
<executable/>
</file_info>
<app_version>
<app_name>setiathome_enhanced</app_name>
<version_num>603</version_num>
<file_ref>
<file_name>AK_v8_win_x64_SSE3x.exe</file_name>
<main_program/>
</file_ref>
</app_version>
</app_info>


Really hate it when I have to put on my glases for close work....lol. LEts see what happens at the end of the file. Had to rearrange the AP info to the top of my file to follow it.
2) Message boards : Number crunching : Help with Optimized AP (Message 801399)
Posted 23 Aug 2008 by Profile DaBrat and DaBear
Post:
Actually this is showing as the new tasks in my Boinc since since downloading

Seti @home Enhanced 6.03

Before I added the AP code it was all 5.28


will it matter?

Are you running AMD or intel... I am running SSE3 on an AMD right now Vista
3) Message boards : Number crunching : Help with Optimized AP (Message 801388)
Posted 23 Aug 2008 by Profile DaBrat and DaBear
Post:
Added and restarted and still the same... now running SETI Enhanced 6.03 instead of 5.28, don't think this falls under my optimized app specs. Running 30 mins for every five estimated completion. Can I just change the version number under my optimized app specs?
4) Message boards : Number crunching : Help with Optimized AP (Message 801387)
Posted 23 Aug 2008 by Profile DaBrat and DaBear
Post:
Thanks Joker.. I am a mainframe assembler type girl... not a Client Server XML type...lol. Appreciate the help. Crashed the first time but the second is running just a bit slower than before... differnt type WUs I believe but I see that graphics are enabled so something must be amiss
5) Message boards : Number crunching : Help with Optimized AP (Message 801381)
Posted 23 Aug 2008 by Profile DaBrat and DaBear
Post:
Eyes are not what they used to be and was wondering if someone who has already started AP running optimized could look at my Ap_info. Not sure if this is correct but it seems my regular WUs are now running a bit slower

- <app_info>
- <app>
<name>setiathome_enhanced</name>
</app>
- <file_info>
<name>AK_v8_win_SSE3.exe</name>
<executable />
</file_info>
- <app_version>
<app_name>setiathome_enhanced</app_name>
<version_num>517</version_num>
- <file_ref>
<file_name>AK_v8_win_SSE3.exe</file_name>
<main_program/>
</file_ref>
</app_version>
- <app_version>
<app_name>setiathome_enhanced</app_name>
<version_num>527</version_num>
- <file_ref>
<file_name>AK_v8_win_SSE3.exe</file_name>
<main_program/>
</file_ref>
</app_version>
- <app_version>
<app_name>setiathome_enhanced</app_name>
<version_num>528</version_num>
- <file_ref>
<file_name>AK_v8_win_SSE3.exe</file_name>
<main_program/>
</file_ref>
</app_version>
<app>
<name>astropulse</name>
</app>
<file_info>
<name>astropulse_4.35_windows_intelx86.exe</name>
<executable/>
</file_info>
<file_info>
<name>libfftw3f-3-1-1a_upx.dll</name>
<executable/>
</file_info>
<file_info>
<name>ap_graphics_4.35_windows_intelx86.exe</name>
<executable/>
</file_info>
<app_version>
<app_name>astropulse</app_name>
<version_num>435</version_num>
<file_ref>
<file_name>astropulse_4.35_windows_intelx86.exe</file_name>
<main_program/>
</file_ref>
<file_ref>
<file_name>libfftw3f-3-1-1a_upx.dll</file_name>
</file_ref>
<file_ref>
<file_name>ap_graphics_4.35_windows_intelx86.exe</file_name>
<open_name>graphics_app</open_name>
</file_ref>
</app_version>
</app_info>


Just making sure all is in the correct order since it doesn't seem to cut and paste truly from my app Info into the forum.
6) Message boards : Number crunching : Question on Astropulse versus regular WU credit (Message 801047)
Posted 22 Aug 2008 by Profile DaBrat and DaBear
Post:
[ Which explains why the whale had an interesting and active, though brief existence, and the bowl of petunias simply said "oh, no, not again."



Dare I ask?


Have you seen the movie Magnolia ?

Opening scene... frogs...

Some days... Weird Just Happens ! or something like that ;-)

Cheers,



LOL!! I actually looked it up it sounded so odd. Since H2G2 is not my usual fair for entertainemnt I was absolutely stiumped. Not that knowing about the petunias and the whale helps....lol! Maybe... stuff happens and then you die?...lol
7) Message boards : Number crunching : Question on Astropulse versus regular WU credit (Message 801037)
Posted 22 Aug 2008 by Profile DaBrat and DaBear
Post:
[ Which explains why the whale had an interesting and active, though brief existence, and the bowl of petunias simply said "oh, no, not again."



Dare I ask?
8) Message boards : Number crunching : Question on Astropulse versus regular WU credit (Message 801015)
Posted 22 Aug 2008 by Profile DaBrat and DaBear
Post:
Wow you learn so much on these SETI boards. Can someone return my kid's AP?
9) Message boards : Number crunching : Question on Astropulse versus regular WU credit (Message 800998)
Posted 22 Aug 2008 by Profile DaBrat and DaBear
Post:
So the credit system is a fluid system and this is just not an overall decrease in points awarded. I can handle that. Kinda hard to stomach the idea of returning more completed units than someone else because my comps are newer or faster and being penalized for that. I am just hoping the 'Holy Grail' for this project, based on teiring and points will not be put out of reach.

Sigh.

The credit multiplier is based on the median computer.

As the median rises (new, faster computers join, old, slower computers drop out or die), the number of FLOPs increases -- and so does the benchmark.

Let's say that based on FLOPs, the hypothetical median computer claimed 50 credits for a work unit. Based on benchmark * time, that same computer would have claimed 45 credits. The scaling factor (to bring FLOPs into line with the definition of a cobblestone) would be 0.9.

Jump ahead a couple of years. The hypothetical median computer is 50% faster.

A 50 credit work unit would be 50 credits based on the same number of FLOPs. Had the credit been done by benchmark * time (the benchmark is higher, and the time is shorter), the credit should have been 45 credits. The scaling factor is still 0.9.

Does that help any?



Sigh.....


Just said that without the explanation. It wasn't a question rather a statement of understanding. The question arose due to the fact that overall RAC seems to be decreasing. Throw the mysterious CPU hour eating AP into the mix and it would appear to someone not inside your comps, that some sort of adjustment if being made, instead of short term may be long term.


It would appear to anyone crunching the project for any length of time that RAC is being adjusted downward. Thats why the question was posted here.
10) Message boards : Number crunching : Question on Astropulse versus regular WU credit (Message 800994)
Posted 22 Aug 2008 by Profile DaBrat and DaBear
Post:
[quoteGiven this as a rule, if you crunch twice as much work with an optimized application as someone with an identical computer running standard Multibeam, how could you not get twice as much credit?[/quote]

This is not how I read your original statement. It said to me that due to optimization causing some computers to get twice the credit, the adjustment was in place to correct this in order not to have a 'credit penalty'.
11) Message boards : Number crunching : Question on Astropulse versus regular WU credit (Message 800976)
Posted 22 Aug 2008 by Profile DaBrat and DaBear
Post:
So the credit system is a fluid system and this is just not an overall decrease in points awarded. I can handle that. Kinda hard to stomach the idea of returning more completed units than someone else because my comps are newer or faster and being penalized for that. I am just hoping the 'Holy Grail' for this project, based on teiring and points will not be put out of reach.
12) Message boards : Number crunching : Question on Astropulse versus regular WU credit (Message 800855)
Posted 22 Aug 2008 by Profile DaBrat and DaBear
Post:
Ok I understood your explanation but the following is the only thing I take exception to:

>>The reward for going to the trouble of optimizing is that we get more credit per hour than we would otherwise be entitled to receive -- just because we did the same work twice as fast is no reason to impose a credit-penalty.<<

In a production environment, which in my opinion this is, producing more work twice as fast SHOULD be rewarded with twice the credit.

JMHO.....

The optimization apps are out there for all who care to go to the trouble. By this I read that even though I may return twice the work as someone which has chosen NOT to optimize, I would not be fairly rwared in points.

13) Message boards : Number crunching : Question on Astropulse versus regular WU credit (Message 800748)
Posted 22 Aug 2008 by Profile DaBrat and DaBear
Post:
As I said in my previous post... I hope that credit will be dialed in. That answers my question but didn't want you to miss the fact that is was not an optimized app. So back to my initial question, will the credit or will it not be comparable. This will get into the discussion of equal pay for equal work no matter what the time frame the work was done in, but not trying to go there. Whether by CPU hour or 'flips and FLOPS'. In general will RAC continue to decline in a system with a ranking tier? Or will comps, even with optimized apps have to work twice as hard and as long to get the same value for their contribution?

The reason the third comp was not initially optimized was in anticipation of the Astropulse WUs.... Lets just say that DaBear (the comp owner) is into the science and happens to be my kid. And originally pointed out that the value seemed rather low, which of course was a disappointment. Took two months but he finally let me optimized his comp after that.


14) Message boards : Number crunching : Question on Astropulse versus regular WU credit (Message 800591)
Posted 21 Aug 2008 by Profile DaBrat and DaBear
Post:
Thanks JD what ya got for AMD SSE3?

Looks like you're already running it ;-)

Cheers,



Thanks again!!! Buit works a WHOLE lot better on my intels....lol!!
15) Message boards : Number crunching : Question on Astropulse versus regular WU credit (Message 800588)
Posted 21 Aug 2008 by Profile DaBrat and DaBear
Post:
Thanks JD what ya got for AMD SSE3?
16) Message boards : Number crunching : Question on Astropulse versus regular WU credit (Message 800582)
Posted 21 Aug 2008 by Profile DaBrat and DaBear
Post:
Ned I don't mean to be repetitive... but the WU that was used for comparison was NOT done with an optimized version of the app. I think I stated that in every post. Hopefully credit wwill be dialed in. Take a closer look at WUs just after the astropulse 'quick before they disappear'. The machine was not optimized until AFTER the math was done. The point in optimizing the machine was to reject further AP WUs.

I believe I put the link to the WU in the previous post and you can see it is not optimized.

People keep going on about the optimization that was adone AFTERWARDS but the fact is an optimized machine, which once again it was not, will not accept AP WUs. Nor was it optimized with the compared result.

Guess I'll just paste it here:


<core_client_version>5.10.45</core_client_version>
<![CDATA[
<stderr_txt>
setiathome_enhanced 5.27 DevC++/MinGW

Work Unit Info:
...............
WU true angle range is : 0.432046
Optimal function choices:
-----------------------------------------------------
name
-----------------------------------------------------
v_BaseLineSmooth (no other)
v_vGetPowerSpectrum 0.00129 0.00000
sse1_ChirpData_ak 0.02012 0.00000
v_vpfTranspose8x4ntw 0.00974 0.00000
BH SSE folding 0.00224 0.00000

Flopcounter: 15833217136241.709000

Spike count: 0
Pulse count: 1
Triplet count: 2
Gaussian count: 0

</stderr_txt>
]]>

Validate state Initial
Claimed credit 44.9296575849168


17061 secs CPU time for 44.xx credit NOT OPTIMIZED
564,497.40 secs for 719 credits AP.... could have crunched 33 of the above in the same time frame for twice the credit....

Does this make my question clearer? Before the question is asked... yes same machine
17) Message boards : Number crunching : Question on Astropulse versus regular WU credit (Message 800288)
Posted 21 Aug 2008 by Profile DaBrat and DaBear
Post:
The machine was not optimized until AFTER the Astropulse was uploaded.. You are looking at recent WUs within the past three dcays... before that... the machine was not optimized as a matter of fact there really is no difference in performance. That is NOT why the difference in credits.

http://setiathome.berkeley.edu/result.php?resultid=950774976

This was not run with an optimized version and is the unit I used for comparison. I actually optimized after doing the math.
18) Message boards : Number crunching : Question on Astropulse versus regular WU credit (Message 800267)
Posted 21 Aug 2008 by Profile DaBrat and DaBear
Post:
Sorry I am so lost... on this machine I basically got what they gave me and optimized after the Astropulse. What determines if you get enhanced or multibeam?

So basically, this is not an issue of equal pay for equal work? I a little confused on the credit system... If you looked at that machine and saw the amount of time it took to crunch it seems that the credit system would be a little more balanced in that respect... not half as much.
19) Message boards : Number crunching : Question on Astropulse versus regular WU credit (Message 800194)
Posted 20 Aug 2008 by Profile DaBrat and DaBear
Post:
Doing a little bit of math, I find that the time it took for one of my comps to crunch and astropulse when compared to a regular WU resulted in about HALF the credit per CPU hr.

Will this be the norm? In the time it took to crunch the one AP, I would have crunched 33 WU equivalent to 44.xx credits each, for a value of twice as many credits.
20) Questions and Answers : Windows : Problems with 6.2.26 (Message 798441)
Posted 15 Aug 2008 by Profile DaBrat and DaBear
Post:
Thanks so much for helping me locate that file.. thought I would have to revert to old client in order to use my opt app.


Next 20


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.