Welcome to MilkyWay@home

New faster application?


Advanced search

Message boards : Number crunching : New faster application?
Message board moderation

To post messages, you must log in.

Previous · 1 . . . 4 · 5 · 6 · 7 · 8 · 9 · 10 . . . 12 · Next

AuthorMessage
zombie67 [MM]
Avatar

Send message
Joined: 29 Aug 07
Posts: 115
Credit: 478,798,803
RAC: 491,511
300 million credit badge14 year member badge
Message 8434 - Posted: 16 Jan 2009, 0:55:04 UTC - in response to Message 8432.  
Last modified: 16 Jan 2009, 0:56:36 UTC

I just ran the SSE2 app on that same machine. The results matched the SSE3 results. So neither SSE2 or SSE3 match the stock app.



Are these the apps from the MW.rar download?

If not, would you please specify exactly which apps you tested?

Thanks

Yes, they were both from the rar download.

ID: 8434 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ProfileLabbie
Avatar

Send message
Joined: 29 Aug 07
Posts: 327
Credit: 116,463,193
RAC: 0
100 million credit badge14 year member badge
Message 8435 - Posted: 16 Jan 2009, 0:55:39 UTC - in response to Message 8434.  

Yes, they were both from the rar download.


Thanks


Calm Chaos Forum...Join Calm Chaos Now
ID: 8435 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ProfileGavin Shaw
Avatar

Send message
Joined: 16 Jan 08
Posts: 98
Credit: 1,371,299
RAC: 0
1 million credit badge14 year member badge
Message 8436 - Posted: 16 Jan 2009, 0:59:36 UTC

So where do we stand?

As Labbie said can people who have tested specify what download they used. Also can anyone get in contact with the guys/gals (or they themselves) who made the apps and find out if they did any testing before release. For your information I have been using the SSE3 version from the MW.rar file downloaded from Ice's site.

For the moment I have set the machine to NNW and I will let it crunch out the remaining units. Then I guess I'll take it elsewhere while this is all sorted out.

And can anyone confirm what Logan brought up?


@ Travis

Why linux, OSX, and Win stock apps gave different results with the same wu?


If this is the case does that not mean there could be an issue with the stock app too?

Never surrender and never give up. In the darkest hour there is always hope.

ID: 8436 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profilebanditwolf
Avatar

Send message
Joined: 12 Nov 07
Posts: 2425
Credit: 524,164
RAC: 0
500 thousand credit badge14 year member badge
Message 8437 - Posted: 16 Jan 2009, 1:03:05 UTC - in response to Message 8433.  

So I guess this SSE3-Version saves the party for us.

Result of the Test-WU:
fitness: -3.046272445362275


Yes, I also got
fitness: -3.046272445362275

But that still does not match the stock app output of
fitness: -3.046272445362277


Are they close enough it won't change the outcome (of the search) or should they be exact?
Doesn't expecting the unexpected make the unexpected the expected?
If it makes sense, DON'T do it.
ID: 8437 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ProfileTravis
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 30 Aug 07
Posts: 2046
Credit: 26,480
RAC: 0
10 thousand credit badge14 year member badge
Message 8438 - Posted: 16 Jan 2009, 1:06:11 UTC - in response to Message 8437.  

So I guess this SSE3-Version saves the party for us.

Result of the Test-WU:
fitness: -3.046272445362275


Yes, I also got
fitness: -3.046272445362275

But that still does not match the stock app output of
fitness: -3.046272445362277


Are they close enough it won't change the outcome (of the search) or should they be exact?


Yeah that's fine. I think the cutoff where it would start to matter would be differences greater than 10^-10ish
ID: 8438 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Cluster Physik

Send message
Joined: 26 Jul 08
Posts: 627
Credit: 94,940,203
RAC: 0
50 million credit badge13 year member badgeextraordinary contributions badge
Message 8439 - Posted: 16 Jan 2009, 1:06:57 UTC - in response to Message 8436.  
Last modified: 16 Jan 2009, 1:11:10 UTC

And can anyone confirm what Logan brought up?


@ Travis

Why linux, OSX, and Win stock apps gave different results with the same wu?


If this is the case does that not mean there could be an issue with the stock app too?

There are some issues with the stock app, but this accuracy thing is nothing you can really avoid between different architectures when using standard math libraries. And Travis answer (actually a question) to that problem

At what level of accuracy?

implies he is aware of that.

Actually, this sort of differences in the last bits is what one often calls numerical noise. Both versions are likely to calculate results within the same error bars when compared to a (hypothetical) implementation with infinite (or just somewhat extended) precision.
ID: 8439 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ProfileGavin Shaw
Avatar

Send message
Joined: 16 Jan 08
Posts: 98
Credit: 1,371,299
RAC: 0
1 million credit badge14 year member badge
Message 8440 - Posted: 16 Jan 2009, 1:22:31 UTC - in response to Message 8439.  

And can anyone confirm what Logan brought up?


@ Travis

Why linux, OSX, and Win stock apps gave different results with the same wu?


If this is the case does that not mean there could be an issue with the stock app too?

There are some issues with the stock app, but this accuracy thing is nothing you can really avoid between different architectures when using standard math libraries. And Travis answer (actually a question) to that problem

At what level of accuracy?

implies he is aware of that.

Actually, this sort of differences in the last bits is what one often calls numerical noise. Both versions are likely to calculate results within the same error bars when compared to a (hypothetical) implementation with infinite (or just somewhat extended) precision.


Then perhaps the project can specify what the accepted 'error margin' is. Or is that the no difference greater than 10^-10ish.

Never surrender and never give up. In the darkest hour there is always hope.

ID: 8440 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Misfit
Avatar

Send message
Joined: 27 Aug 07
Posts: 915
Credit: 1,503,319
RAC: 0
1 million credit badge14 year member badge
Message 8441 - Posted: 16 Jan 2009, 2:11:31 UTC

It's like Einstein and akosf all over again. Good times.
me@rescam.org
ID: 8441 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profilebanditwolf
Avatar

Send message
Joined: 12 Nov 07
Posts: 2425
Credit: 524,164
RAC: 0
500 thousand credit badge14 year member badge
Message 8442 - Posted: 16 Jan 2009, 2:17:34 UTC - in response to Message 8438.  

Yeah that's fine. I think the cutoff where it would start to matter would be differences greater than 10^-10ish


So then the "new" optimized apps are fine to use as long as they are correct to the 10th place? So far only the ss3 has been confirmed to work out fine within the limit right? What about the sse2.rar app?
Doesn't expecting the unexpected make the unexpected the expected?
If it makes sense, DON'T do it.
ID: 8442 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ProfileTravis
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 30 Aug 07
Posts: 2046
Credit: 26,480
RAC: 0
10 thousand credit badge14 year member badge
Message 8443 - Posted: 16 Jan 2009, 2:22:37 UTC - in response to Message 8442.  

Yeah that's fine. I think the cutoff where it would start to matter would be differences greater than 10^-10ish


So then the "new" optimized apps are fine to use as long as they are correct to the 10th place? So far only the ss3 has been confirmed to work out fine within the limit right? What about the sse2.rar app?


Yeah, so long as it's just numerical noise.
ID: 8443 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
James Nunley

Send message
Joined: 29 Nov 07
Posts: 39
Credit: 74,300,629
RAC: 0
50 million credit badge14 year member badge
Message 8446 - Posted: 16 Jan 2009, 4:21:32 UTC

Could someone please post a link to which one of these apps are supposedly ok. It's kind of confusing as there is more than one file download linked in the thread.

thanks.
ID: 8446 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profilebanditwolf
Avatar

Send message
Joined: 12 Nov 07
Posts: 2425
Credit: 524,164
RAC: 0
500 thousand credit badge14 year member badge
Message 8447 - Posted: 16 Jan 2009, 4:47:34 UTC - in response to Message 8446.  

Could someone please post a link to which one of these apps are supposedly ok. It's kind of confusing as there is more than one file download linked in the thread.

thanks.


If I have read it all I believe the sse3 app is the only one tested so far. No one else has said about the others.
Doesn't expecting the unexpected make the unexpected the expected?
If it makes sense, DON'T do it.
ID: 8447 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
James Nunley

Send message
Joined: 29 Nov 07
Posts: 39
Credit: 74,300,629
RAC: 0
50 million credit badge14 year member badge
Message 8448 - Posted: 16 Jan 2009, 4:54:32 UTC

However there is more than one sse3 app linked in this thread

one is

astronomy_sse3_app and is 889kb


and the other is


astronomy_SSE3_s and is 920kb
ID: 8448 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ProfileLabbie
Avatar

Send message
Joined: 29 Aug 07
Posts: 327
Credit: 116,463,193
RAC: 0
100 million credit badge14 year member badge
Message 8449 - Posted: 16 Jan 2009, 5:11:02 UTC - in response to Message 8438.  

So I guess this SSE3-Version saves the party for us.

Result of the Test-WU:
fitness: -3.046272445362275


Yes, I also got
fitness: -3.046272445362275

But that still does not match the stock app output of
fitness: -3.046272445362277


Are they close enough it won't change the outcome (of the search) or should they be exact?


Yeah that's fine. I think the cutoff where it would start to matter would be differences greater than 10^-10ish


The app linked above is the only one I know of that is good. The last post above (the "Yeah that's fine" one) is from Travis.


Calm Chaos Forum...Join Calm Chaos Now
ID: 8449 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
James Nunley

Send message
Joined: 29 Nov 07
Posts: 39
Credit: 74,300,629
RAC: 0
50 million credit badge14 year member badge
Message 8451 - Posted: 16 Jan 2009, 5:16:07 UTC

Ok so that would make it the


astronomy_SSE3_s one


cool cause thats the one I'm using
ID: 8451 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
zombie67 [MM]
Avatar

Send message
Joined: 29 Aug 07
Posts: 115
Credit: 478,798,803
RAC: 491,511
300 million credit badge14 year member badge
Message 8453 - Posted: 16 Jan 2009, 5:33:16 UTC - in response to Message 8451.  

So, we have a new version of the SSE3 app that meets the 10 digit requirement. Does anyone have an updated SSE or SSE2 version?

ID: 8453 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ProfileLogan
Avatar

Send message
Joined: 15 Aug 08
Posts: 163
Credit: 3,876,869
RAC: 0
3 million credit badge13 year member badge
Message 8454 - Posted: 16 Jan 2009, 8:10:59 UTC - in response to Message 8426.  

@ Travis

Why linux, OSX, and Win stock apps gave different results with the same wu?

Best regards.


At what level of accuracy?


Didn't you tested it...? Bad boy...:)

Try it, and give us your results.

Best regards.
Logan.

BOINC FAQ Service (Ahora, también disponible en Español/Now available in Spanish)
ID: 8454 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
John Clark

Send message
Joined: 4 Oct 08
Posts: 1734
Credit: 64,228,409
RAC: 0
50 million credit badge13 year member badge
Message 8456 - Posted: 16 Jan 2009, 10:25:07 UTC

Travis/Dave

Part of the assimilator update allows us to track users that repeatedly return invalid results and stop awarding credit to them


How quickly do you guys expect this assimilator update to pick up users returning invalid results and stop awarding credit?

Is it very soon (<1 day) or more?
ID: 8456 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ProfileGalaxyIce
Avatar

Send message
Joined: 6 Apr 08
Posts: 2018
Credit: 100,142,856
RAC: 0
100 million credit badge14 year member badge
Message 8457 - Posted: 16 Jan 2009, 10:49:56 UTC - in response to Message 8456.  

Travis/Dave

Part of the assimilator update allows us to track users that repeatedly return invalid results and stop awarding credit to them


How quickly do you guys expect this assimilator update to pick up users returning invalid results and stop awarding credit?

Is it very soon (<1 day) or more?

And are you going to tell them or let them continue unaware?


ID: 8457 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
[BOINCstats] LostBoy

Send message
Joined: 8 Mar 08
Posts: 4
Credit: 774,815
RAC: 0
500 thousand credit badge14 year member badge
Message 8458 - Posted: 16 Jan 2009, 11:25:37 UTC - in response to Message 8453.  

So, we have a new version of the SSE3 app that meets the 10 digit requirement. Does anyone have an updated SSE or SSE2 version?


Here we go...

http://rapidshare.com/files/184217251/astronomy_sse2_app.rar.html

Original app:
fitness: -3.046272445362277

this one:
fitness: -3.046272445362278

have fun...

LB
ID: 8458 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Previous · 1 . . . 4 · 5 · 6 · 7 · 8 · 9 · 10 . . . 12 · Next

Message boards : Number crunching : New faster application?

©2022 Astroinformatics Group