Posts by Mr.Pernod

1) Message boards : Number crunching : Credit discussion II (Message 619070)
Posted 14 Aug 2007 by Profile Mr.Pernod
Post:
however interesting this digression is, it does not seem to bring us any closer to facilitating Bill&Patsy's desire to reward computational effort based on scientific "worth"

;-)
2) Message boards : Number crunching : Credit discussion II (Message 619045)
Posted 14 Aug 2007 by Profile Mr.Pernod
Post:
<SNIP>
please define the scientific "worth" of a SETI-result that does NOT contain any likely or confirmed signal(s)

It narrows done the areas that future studies should look at for ETI, by elimination.

would this elimination have a lesser, equal or higher scientific "worth" than a result containing a likely or confirmed signal?

(not taking into account a possible higher scan resolution in a future science application that would necessitate a rescan of all areas)

That depends, if the receivers at Aricebo were on 24/7 and other scientists did not direct the receiver platform then we would have scanned the whole area several times. And because we got zero responses this time for that pinprick of sky, what did we get previously and what might we find in the future. The whole survey depends on persistency. see Near Time Persistency Checker and link from there.

What I read in this is that the scientific "worth" of a result might increase or decrease over time, depending on other results covering the same area/(near)frequency, yet observed at a different point in time.
If my understanding of this is correct, a constant reevaluation of the "reward" based on scientific "worth" for work done should take place.
3) Message boards : Number crunching : Credit discussion II (Message 619037)
Posted 14 Aug 2007 by Profile Mr.Pernod
Post:
<SNIP>
please define the scientific "worth" of a SETI-result that does NOT contain any likely or confirmed signal(s)

It narrows done the areas that future studies should look at for ETI, by elimination.

would this elimination have a lesser, equal or higher scientific "worth" than a result containing a likely or confirmed signal?

(not taking into account a possible higher scan resolution in a future science application that would necessitate a rescan of all areas)
4) Message boards : Number crunching : Credit discussion II (Message 619031)
Posted 14 Aug 2007 by Profile Mr.Pernod
Post:
<SNIP>
Consider: If a project has, say, a quasi-infinite loop that does almost nothing, apparently you approve as long as the machine cycles are rewarded. But, per this example, the naked definition of Cobblestones doesn't prove that they represent any science value at all. Why do you continue to try to force the Science into the Cobblestones, instead of looking at the actual science that is produced by those Cobblestones? The scientific "worth" of one Cobblestone cannot sensibly be compared in vacuo to another.

So, I continue to advocate that the more science that is performed for a given amount of work (Cobblestones, if you will), the more recognition that should be accorded. Conversely, those that waste Cobblestones should not be rewarded.

please define the scientific "worth" of a SETI-result that does NOT contain any likely or confirmed signal(s)
5) Message boards : Number crunching : Because of the 'validate errors' (Message 618268)
Posted 12 Aug 2007 by Profile Mr.Pernod
Post:
<SNIP>

I OC the Intel Core2 Extreme QX6700 from 2.66 to 3.17 GHz, so it's not so much..
You must ask msattler because of his OC! ;-)

<SNIP>


overclocking your FSB can have a dramatic effect on other devices on the motherboard if your motherboard does NOT have AGP/PCI/PCI-e locks or manually adjustable bus-speeds or dividers.

I recently tried overclocking a board without those options and the onboard NIC (on the PCI-bus) failed completely at an overclock of 54MHz on the FSB (from 200 to 254) which put the PCI-bus on a frequency of 42MHz (from 33MHz), which is comparable to your overclock from 266 to 317. At a lower overclock the NIC seems to work ok.

Check your BIOS for an AGP/PCI/PCI-e divider or manual speedsettings and put those values so that the busspeeds are close to stock.
6) Message boards : Cafe SETI : quotes . . . (Message 348292)
Posted 25 Jun 2006 by Profile Mr.Pernod
Post:
Be cautious with your judgement, for you never truly know - Damian Wilson
7) Message boards : Cafe SETI : quotes . . . (Message 336367)
Posted 14 Jun 2006 by Profile Mr.Pernod
Post:
Half of this world is only joking, the others are purely provoking - Damian Wilson
8) Message boards : Number crunching : Combining CPU's to process Work Units (Message 327887)
Posted 5 Jun 2006 by Profile Mr.Pernod
Post:
A multi-threaded science application could be nice, but I'm not sure the work that has to be done allows for it.

In my opinion, on hyperthreaded systems, it could potentially reduce cache-contention between two seperately running results, because both virtual cpu's would be working on the same dataset, however, on multi-cpu-systems it would require synchronizing the cache of both physical cpu's on a regular basis, which would introduce some delay in the processing.
It would also involve a serious rethink of which dependencies exist in the way the calculations on the data are ordered.
In my opinion, it would be a nice exercise for the programmers of the science application, but I have some reservation as to the possible gains in total processing throughput on most systems.
9) Message boards : Number crunching : Do I have to leave SETI after 6.5 years of membership because I'm always late? (Message 322342)
Posted 1 Jun 2006 by Profile Mr.Pernod
Post:
just a thought....
if BOINC is set to "Do work while computer is in use? yes", it might give the user a feeling of a slight lag when it is releasing the cpu to userprocesses.

the computer is listed as having 2 cpu's.

why not set BOINC to "Do work while computer is in use? yes" and "On multiprocessors, use at most 1 processors" so the possible feeling of lag is removed, but BOINC will still be getting the full benefit of 1 cpu for most of the day.
10) Message boards : Number crunching : There are credits and then there are credits! (Message 317981)
Posted 26 May 2006 by Profile Mr.Pernod
Post:
I wonder how many hosts he had to merge to get that number....

[EDIT]
need to type faster

keep an eye on his machines, he sure has a lot of similar hosts, which I expect to be gone soon.
[/EDIT]
11) Message boards : Number crunching : Response to concerns regarding the new credit system. (Message 310492)
Posted 19 May 2006 by Profile Mr.Pernod
Post:
funny how this "I have some spare cpu-cycles, let me help science" turned into "gimme my credit or else...."
12) Message boards : Number crunching : Error 403 (Message 307401)
Posted 16 May 2006 by Profile Mr.Pernod
Post:
a similar problem with validation occured over at Einstein last friday.
they had to reissue the work with the validation-errors and run a script to grant credit once the reissued results were returned and validated.


So worst case scenerio here is that I've lost 50+ hours of processing time? Now that would be discouraging... :(

Over at Einstein they kept the invalidated results and issued new results to other machines and then when the newly issued results were returned and validated they ran a script to award the same credit to the invalidated results.
One can only hope the SETI-crew does the same thing in this case.
13) Message boards : Number crunching : Error 403 (Message 307369)
Posted 16 May 2006 by Profile Mr.Pernod
Post:
So can I assume that all will be well and we'll get credit for the Vaidate error work once "kryten" is fully back online? I currently have over 50 hours worth of work showing Validate Error :(

a similar problem with validation occured over at Einstein last friday.
they had to reissue the work with the validation-errors and run a script to grant credit once the reissued results were returned and validated.
14) Message boards : Number crunching : SETI Enhanced - Protest Thread!!! Post to Vote: Yes! (to problems) (Message 307050)
Posted 16 May 2006 by Profile Mr.Pernod
Post:
Hi Mr. Pernod!

Is the Pentium-M Version working better than the SSE2 version - on a Pentium-M that is :o)

I've got a bunch of those...

Regards Hans

Hans,
the Pentium-M package is a combination of Enhanced 5.12 SSE2 for P4 and the Pentium M 4.11
15) Message boards : Number crunching : SETI Enhanced - Protest Thread!!! Post to Vote: Yes! (to problems) (Message 306750)
Posted 15 May 2006 by Profile Mr.Pernod
Post:
[off-topic]
The only WU's I have trashed, were because I aborted the 4.18's and because I fiddled about with Crunch3r's app. Haven't got it to work (yet)...

Fuzzy, try these

[/off-topic]
16) Message boards : Number crunching : Sticky post for current version? (Message 306724)
Posted 15 May 2006 by Profile Mr.Pernod
Post:
I agree, I don't think we'll see a "final" version this week.
what versionnumber it will be? dunno, your guess is as good as mine.
which brings us back to the original idea of the thread.
I don't think a link on the homepage or a sticky with a link to the Applications Page is a bad idea.
it will give the people who are using the optimized applications a signal to check their queues and the optimizers homepages.
17) Message boards : Number crunching : Sticky post for current version? (Message 306554)
Posted 15 May 2006 by Profile Mr.Pernod
Post:
That would be reasonable except that my P4 2.8G has the exact same app_info.xml as my P4 1.4G and it is only downloading 4.18 and 5.12 - no sign of a 5.13.

interesting.
your P4 2.8 is still crunching on results that were released a couple of days before 5.13 was released for the Mac OS, while your P4 1.4 is crunching results from the day 5.13 was released or newer.
18) Message boards : Number crunching : Sticky post for current version? (Message 306475)
Posted 15 May 2006 by Profile Mr.Pernod
Post:
ok, here's how I see it (someone correct me if I'm wrong, please)
due to the current version-discrepancies between different systems/operating systems, the splitter is giving out work that has both version numbers in the header-information.
your client with the adapted app_info.xml sees 5.13 as the newest version and ignores the 5.12 version-information in the result and crunches the work with the "spoofed" 5.13 (crunch3rs 5.12) on your machine and reports this back as being crunched by 5.13
19) Message boards : Cafe SETI : Words of UnWisdom (Message 306456)
Posted 15 May 2006 by Profile Mr.Pernod
Post:
a cause worth dying for isn't worth living for
20) Message boards : Number crunching : Sticky post for current version? (Message 306365)
Posted 15 May 2006 by Profile Mr.Pernod
Post:
well, it IS the official listing, so I suspect a sticky on the forums wouldn't be much more up-to-date.

btw, any chance you are running crunch3rs optimized application with an app_info.xml that includes a few lines for 5.13 on your P4 1.6?
WU 77831013 is showing your host to run 5.13 through Crunch3rs 5.12 while the other hosts are using the official 5.12....


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.