Welcome to MilkyWay@home

app v0.9

Message boards : Number crunching : app v0.9
Message board moderation

To post messages, you must log in.

1 · 2 · Next

AuthorMessage
Profile Travis
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 30 Aug 07
Posts: 2046
Credit: 26,480
RAC: 0
Message 8518 - Posted: 17 Jan 2009, 15:42:54 UTC

Please post any problems with app v0.9 here.
ID: 8518 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile speedimic
Avatar

Send message
Joined: 22 Feb 08
Posts: 260
Credit: 57,387,048
RAC: 0
Message 8522 - Posted: 17 Jan 2009, 17:56:11 UTC - in response to Message 8518.  

...no source code for this one?
mic.


ID: 8522 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Travis
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 30 Aug 07
Posts: 2046
Credit: 26,480
RAC: 0
Message 8523 - Posted: 17 Jan 2009, 17:59:01 UTC - in response to Message 8522.  

...no source code for this one?


Its the same as 0.8.
ID: 8523 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile speedimic
Avatar

Send message
Joined: 22 Feb 08
Posts: 260
Credit: 57,387,048
RAC: 0
Message 8525 - Posted: 17 Jan 2009, 18:37:12 UTC - in response to Message 8523.  

ok, then then I got it runnin' since yesterday without any problem...
but now I can compare with stock:

stock09: fitness: -3.046272445362276
my 08/09: fitness: -3.046272445362276

This host: 43423

mic.


ID: 8525 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile [AF>Occitania>Lengadocia...

Send message
Joined: 30 Mar 08
Posts: 25
Credit: 75,915,107
RAC: 0
Message 8535 - Posted: 17 Jan 2009, 20:00:35 UTC

Hello,

No problem on ubuntu 64 bits.
Except time to compute : 1:30 instead of 42 minutes for 0.07.

Best regards
ID: 8535 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Travis
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 30 Aug 07
Posts: 2046
Credit: 26,480
RAC: 0
Message 8537 - Posted: 17 Jan 2009, 20:15:07 UTC - in response to Message 8535.  

Hello,

No problem on ubuntu 64 bits.
Except time to compute : 1:30 instead of 42 minutes for 0.07.

Best regards


really? it's twice as slow?

ID: 8537 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile speedimic
Avatar

Send message
Joined: 22 Feb 08
Posts: 260
Credit: 57,387,048
RAC: 0
Message 8543 - Posted: 17 Jan 2009, 21:52:23 UTC - in response to Message 8537.  

For me the new v8/9 is faster. 13min against 17-18. (Intel compiler SSE41)
mic.


ID: 8543 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Wang Solutions
Avatar

Send message
Joined: 22 Dec 07
Posts: 13
Credit: 46,606,530
RAC: 0
Message 8545 - Posted: 17 Jan 2009, 22:02:19 UTC - in response to Message 8543.  
Last modified: 17 Jan 2009, 22:05:53 UTC

I am getting 81 minutes to completion now on 0.09 versus 38 minutes with 0.07.

Q6600 on Kubuntu 7.10 64-bit.
Proud member of BOINC@AUSTRALIA
ID: 8545 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile [AF>Occitania>Lengadocia...

Send message
Joined: 30 Mar 08
Posts: 25
Credit: 75,915,107
RAC: 0
Message 8546 - Posted: 17 Jan 2009, 22:11:25 UTC
Last modified: 17 Jan 2009, 22:11:55 UTC

Really Travis

0.07 2400 to 2500 s
0.09 5400 s

Q9300 on ubuntu 8.04 hardy heron 64 bit see the change here

62198099 62264858 17 Jan 2009 17:32:28 UTC 17 Jan 2009 19:55:32 UTC Over Success Done 5,408.09 42.56 39.84
62182048 61575414 17 Jan 2009 16:17:04 UTC 17 Jan 2009 19:55:32 UTC Over Success Done 5,404.16 42.52 39.84
62182047 61575327 17 Jan 2009 16:17:04 UTC 17 Jan 2009 19:25:53 UTC Over Success Done 5,535.09 43.56 41.09
62178946 62245955 17 Jan 2009 15:51:41 UTC 17 Jan 2009 18:34:30 UTC Over Success Done 5,405.97 42.54 39.84
62178945 62245954 17 Jan 2009 15:51:41 UTC 17 Jan 2009 18:09:50 UTC Over Success Done 5,400.31 42.49 39.84
62169653 62236797 17 Jan 2009 15:06:42 UTC 17 Jan 2009 17:32:28 UTC Over Success Done 2,541.26 20.00 41.09
62169652 62236796 17 Jan 2009 15:06:42 UTC 17 Jan 2009 17:32:28 UTC Over Success Done 2,594.65 20.42 41.09
62165248 62232484 17 Jan 2009 14:45:41 UTC 17 Jan 2009 16:38:01 UTC Over Success Done 2,483.09 19.54 39.84
62165247 62232483 17 Jan 2009 14:45:41 UTC 17 Jan 2009 16:31:23 UTC Over Success Done 2,491.66 19.61 39.84
62160186 62227517 17 Jan 2009 14:21:07 UTC 17 Jan 2009 16:17:03 UTC Over Success Done 2,486.10 19.56 39.84
62160185 62227516 17 Jan 2009 14:21:07 UTC 17 Jan 2009 15:42:47 UTC Over Success Done 2,500.76 19.68 39.84
62157831 62225196 17 Jan 2009 14:09:29 UTC 17 Jan 2009 15:40:00 UTC Over Success Done 2,467.98 19.42 41.09
62151734 62219171 17 Jan 2009 13:38:09 UTC 17 Jan 2009 15:23:51 UTC Over Success Done 2,536.18 19.96 41.09
62151719 62219156 17 Jan 2009 13:38:09 UTC 17 Jan 2009 15:06:42 UTC Over Success Done 2,547.07 20.04 41.09
ID: 8546 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
MB Atlanos

Send message
Joined: 2 Sep 07
Posts: 18
Credit: 180,611
RAC: 0
Message 8555 - Posted: 18 Jan 2009, 0:53:34 UTC

no difference in runtime between v0.7 and v0.9 for PPC on a G4 Mac mini:
http://milkyway.cs.rpi.edu/milkyway/results.php?hostid=1905
ID: 8555 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
paris
Avatar

Send message
Joined: 26 Apr 08
Posts: 87
Credit: 64,801,496
RAC: 0
Message 8556 - Posted: 18 Jan 2009, 1:22:00 UTC

Several of my machines (running OS X) are taking about twice as long to complete a unit as they did before. I don't know if it is due to the new types of units or the new application. Is there a new adjustment to the credit award in the works?
ID: 8556 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
paris
Avatar

Send message
Joined: 26 Apr 08
Posts: 87
Credit: 64,801,496
RAC: 0
Message 8557 - Posted: 18 Jan 2009, 2:11:02 UTC

(addendum to previous message):
The problem appears to be on the Intel powered Macs; the PPC Macs are running about the same as before.
ID: 8557 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Travis
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 30 Aug 07
Posts: 2046
Credit: 26,480
RAC: 0
Message 8558 - Posted: 18 Jan 2009, 2:57:08 UTC - in response to Message 8557.  

(addendum to previous message):
The problem appears to be on the Intel powered Macs; the PPC Macs are running about the same as before.


I changed some of the compiler flags which supposedly would make it run faster. Seems like they haven't so I'm going to update the app again which should fix the problem.
ID: 8558 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
jedirock
Avatar

Send message
Joined: 8 Nov 08
Posts: 178
Credit: 6,140,854
RAC: 0
Message 8560 - Posted: 18 Jan 2009, 3:21:49 UTC - in response to Message 8558.  

I changed some of the compiler flags which supposedly would make it run faster. Seems like they haven't so I'm going to update the app again which should fix the problem.

What flags did you use?
ID: 8560 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Travis
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 30 Aug 07
Posts: 2046
Credit: 26,480
RAC: 0
Message 8574 - Posted: 18 Jan 2009, 14:49:26 UTC - in response to Message 8560.  
Last modified: 18 Jan 2009, 14:49:49 UTC

I changed some of the compiler flags which supposedly would make it run faster. Seems like they haven't so I'm going to update the app again which should fix the problem.

What flags did you use?


I was trying some of the ones suggested in the code forum. I just went back to the old flags, so v0.10 is using:

osx, x86_64:
-O3 -mfpmath=sse -mtune=nocona -msse3 -ftree-vectorize -funroll-loops

osx, i686:
-O2 -msse2 -mfpmath=sse -mtune=prescott -ftree-vectorize -funroll-loops

osx, ppc:
-O2 -maltivec -mabi=altivec -mcpu=7400 -funroll-loops

linux, x86_64:
-O2 -msse2 -ftree-vectorize -pedantic -funroll-loops

linux, i686:
-m32 -O2 -ftree-vectorize -pedantic -funroll-loops


I'm not sure what compiler flags Dave is using for the windows versions.
ID: 8574 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile speedimic
Avatar

Send message
Joined: 22 Feb 08
Posts: 260
Credit: 57,387,048
RAC: 0
Message 8578 - Posted: 18 Jan 2009, 15:09:39 UTC - in response to Message 8574.  

I think -pedantic will slow thing down a bit...
mic.


ID: 8578 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Travis
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 30 Aug 07
Posts: 2046
Credit: 26,480
RAC: 0
Message 8582 - Posted: 18 Jan 2009, 15:20:53 UTC - in response to Message 8578.  

I think -pedantic will slow thing down a bit...


I thought that was just for warnings?
ID: 8582 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile speedimic
Avatar

Send message
Joined: 22 Feb 08
Posts: 260
Credit: 57,387,048
RAC: 0
Message 8585 - Posted: 18 Jan 2009, 15:48:12 UTC - in response to Message 8582.  

I think -pedantic will slow thing down a bit...


I thought that was just for warnings?


You're right. I mixed that up with some fp-whatever-option...
mic.


ID: 8585 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
jedirock
Avatar

Send message
Joined: 8 Nov 08
Posts: 178
Credit: 6,140,854
RAC: 0
Message 8586 - Posted: 18 Jan 2009, 17:27:40 UTC - in response to Message 8574.  
Last modified: 18 Jan 2009, 17:27:57 UTC

I was trying some of the ones suggested in the code forum. I just went back to the old flags, so v0.10 is using:

osx, x86_64:
-O3 -mfpmath=sse -mtune=nocona -msse3 -ftree-vectorize -funroll-loops

osx, i686:
-O2 -msse2 -mfpmath=sse -mtune=prescott -ftree-vectorize -funroll-loops

osx, ppc:
-O2 -maltivec -mabi=altivec -mcpu=7400 -funroll-loops

...

I'm not sure what compiler flags Dave is using for the windows versions.

Yeah, the OS X flags look fine to me (where I specialize ;-) ).
ID: 8586 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
mbs

Send message
Joined: 29 Nov 08
Posts: 1
Credit: 251,947
RAC: 0
Message 8621 - Posted: 18 Jan 2009, 22:24:59 UTC - in response to Message 8586.  

osx, x86_64:
-O3 -mfpmath=sse -mtune=nocona -msse3 -ftree-vectorize -funroll-loops
...


Yeah, the OS X flags look fine to me (where I specialize ;-) ).


Please use -march=nocona instead of -mtune=nocona.
-mtune will tune the code for specific processor, but -march will actually use the processor specific instructions (and implicitly set -mtune for the target architecture)
ID: 8621 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
1 · 2 · Next

Message boards : Number crunching : app v0.9

©2024 Astroinformatics Group