Welcome to MilkyWay@home

App v0.10

Message boards : Number crunching : App v0.10
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 8573 - Posted: 18 Jan 2009, 14:44:30 UTC

Let me know if the recompiles of 0.9 are running as fast as they used to. Thanks!
ID: 8573 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile banditwolf
Avatar

Send message
Joined: 12 Nov 07
Posts: 2425
Credit: 524,164
RAC: 0
Message 8575 - Posted: 18 Jan 2009, 14:55:35 UTC

Are any of these apps being tested first or being put out and let the public test them? Shouldn't these new apps be put out as "test apps" so that adds another level of testing after your tests are done to check that the app works before they are made the current official app?

I think too many apps are being thrown out too quickly in the last week, and it is only causing confusing for some people. Along with the many errors that have occured only cause more havoc.
Doesn't expecting the unexpected make the unexpected the expected?
If it makes sense, DON'T do it.
ID: 8575 · 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 8576 - Posted: 18 Jan 2009, 15:01:45 UTC - in response to Message 8575.  

Are any of these apps being tested first or being put out and let the public test them? Shouldn't these new apps be put out as "test apps" so that adds another level of testing after your tests are done to check that the app works before they are made the current official app?

I think too many apps are being thrown out too quickly in the last week, and it is only causing confusing for some people. Along with the many errors that have occured only cause more havoc.


We could do that, but the only problem is when we were running two seperate apps (one regular and one test), the server really suffered from bad performance. I think from here on out most changes to the application will be in the form of different compiler flags and small changes in the code for optimization so I don't see this as a huge issue.

When we move to beta (remember this project is STILL alpha), I think then we'll get another server that's either fast enough to handle running a test app and regular app or have two seperate machines, one running test apps and the other doing the regular MW stuff.
ID: 8576 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile banditwolf
Avatar

Send message
Joined: 12 Nov 07
Posts: 2425
Credit: 524,164
RAC: 0
Message 8577 - Posted: 18 Jan 2009, 15:06:27 UTC - in response to Message 8576.  

If you limit the # of test wu's to one or two hundred then it shouldn't add much strain at all but still give results to show if the test app worked or not.
Doesn't expecting the unexpected make the unexpected the expected?
If it makes sense, DON'T do it.
ID: 8577 · 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 8583 - Posted: 18 Jan 2009, 15:23:21 UTC - in response to Message 8577.  
Last modified: 18 Jan 2009, 15:23:42 UTC

If you limit the # of test wu's to one or two hundred then it shouldn't add much strain at all but still give results to show if the test app worked or not.


We can give it a shot for the next code update and see how the server handles it. If the server can handle it it's probably a good practice to get into the habit of anyways.
ID: 8583 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
cwhyl

Send message
Joined: 11 Nov 07
Posts: 41
Credit: 1,000,181
RAC: 0
Message 8609 - Posted: 18 Jan 2009, 21:07:46 UTC
Last modified: 18 Jan 2009, 21:38:34 UTC

64-bit Linux validates ok but takes double time to complete compared with 32-bit
ID: 8609 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Phil
Avatar

Send message
Joined: 13 Feb 08
Posts: 1124
Credit: 46,740
RAC: 0
Message 8615 - Posted: 18 Jan 2009, 21:40:15 UTC - in response to Message 8573.  

Let me know if the recompiles of 0.9 are running as fast as they used to. Thanks!


Working the same as v0.07 did.
The DCF is 0.12
ID: 8615 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile rottenkiwi

Send message
Joined: 15 Mar 08
Posts: 3
Credit: 17,692,264
RAC: 0
Message 8628 - Posted: 19 Jan 2009, 0:24:05 UTC

0.10 4 Linux x86_64 takes 4200 s, 0.10 4 Lin. i686 takes 1800 s on same PC.
ID: 8628 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile KWSN Checklist
Avatar

Send message
Joined: 12 Aug 08
Posts: 253
Credit: 275,593,872
RAC: 0
Message 8630 - Posted: 19 Jan 2009, 1:07:42 UTC
Last modified: 19 Jan 2009, 1:12:18 UTC

0.07 8400 sec., 0.10 8200 sec. on Linux x86_64 Intel D940.
ID: 8630 · 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 8634 - Posted: 19 Jan 2009, 2:01:30 UTC

OS X still takes about twice as long on Intel Mac as it did before.
ID: 8634 · 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 8643 - Posted: 19 Jan 2009, 7:15:49 UTC
Last modified: 19 Jan 2009, 7:16:35 UTC

I found that v0.09 took more than twice as long as 0.07. v0.10 is possibly faster than 0.09 but still a lot slower than 0.07.
Proud member of BOINC@AUSTRALIA
ID: 8643 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Tin Man

Send message
Joined: 1 Jan 08
Posts: 3
Credit: 31,972,566
RAC: 0
Message 8644 - Posted: 19 Jan 2009, 7:41:26 UTC

Was crunching work units in just over an hour, but for the last 2 days they have been taking twice as long to complete.
ID: 8644 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
6dj72cn8

Send message
Joined: 26 Dec 07
Posts: 41
Credit: 2,582,082
RAC: 0
Message 8645 - Posted: 19 Jan 2009, 7:50:02 UTC - in response to Message 8634.  

OS X still takes about twice as long on Intel Mac as it did before.

Then it must depend which Intel Mac you have. 0.7, 0.8, 0.9 and 0.10 have all been pretty much the same speed for me.

ID: 8645 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
cwhyl

Send message
Joined: 11 Nov 07
Posts: 41
Credit: 1,000,181
RAC: 0
Message 8647 - Posted: 19 Jan 2009, 8:32:26 UTC

32-bit Linux is one or two minutes faster than before.
64-bit Linux takes double time.
ID: 8647 · 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 8655 - Posted: 19 Jan 2009, 11:43:27 UTC - in response to Message 8647.  

Same for me
0.10
Linux 64 bits still around 1h30 instead of 42 minutes 0.07

62698794 62752634 19 Jan 2009 8:41:38 UTC 19 Jan 2009 11:11:42 UTC Over Success Done 5,307.89 41.77 39.84
62682174 62736354 19 Jan 2009 7:29:18 UTC 19 Jan 2009 10:19:08 UTC Over Success Done 5,533.33 43.54 41.09
62675156 62729576 19 Jan 2009 7:00:09 UTC 19 Jan 2009 8:41:38 UTC Over Success Done 5,390.89 42.42 39.84
ID: 8655 · 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 8658 - Posted: 19 Jan 2009, 13:14:04 UTC - in response to Message 8645.  

OS X still takes about twice as long on Intel Mac as it did before.

Then it must depend which Intel Mac you have. 0.7, 0.8, 0.9 and 0.10 have all been pretty much the same speed for me.



I have two core duo mini's runing Tiger and one core 2 duo MacBook running Leopard. What are you running?
ID: 8658 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile DaveSun
Avatar

Send message
Joined: 10 Nov 07
Posts: 28
Credit: 2,549,231
RAC: 0
Message 8659 - Posted: 19 Jan 2009, 13:16:34 UTC

Just checking on v .10 on my windows machines looks like completion time has increased 50% over v .07. Completion times have gone from 2000 secs to 3000 secs.
ID: 8659 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Jim Wilkins

Send message
Joined: 1 Jan 08
Posts: 40
Credit: 1,676,165
RAC: 0
Message 8661 - Posted: 19 Jan 2009, 14:33:28 UTC - in response to Message 8659.  

I have seen very little increase or decrease from v. 0.7 on either my Mac or PC.

Jim
ID: 8661 · 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 8666 - Posted: 19 Jan 2009, 16:13:16 UTC

No change between version 0.1 and 0.07 on windows 32 bits
core duo T7250 2GHz

around 1 hour and 10 minutes.
ID: 8666 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
C

Send message
Joined: 22 Nov 07
Posts: 36
Credit: 1,224,316
RAC: 0
Message 8668 - Posted: 19 Jan 2009, 18:08:16 UTC

There is definitely something wrong with the OS X version for Intel.

I have four computers doing MW:

333MHz iMac PPC G3 OS X 10.3.9 ~30500 cpusec/WU
1.33GHz iBook PPC G4 OS X 10.5.6 ~ 9100 cpusec/WU 4 times faster clock, 30% of the imac time, altivec helps a bit
1.67GHz MacBookPro Core Duo OS X 10.4.11 ~ 8100 cpusec/wu 5 times faster clock, 25% of the imac time; I would expect around 6100 cpusec, based on clock speed alone. If I remember correctly, it used to actually take in the low 4000's, because of SSE2 and others(?)
2.4GHz MacBookPro Core 2 Duo OS X 10.4.11 ~ 7200 cpusec/wu 7.2 times faster clock, 23% of the imac time; I would expect around 4200 cpusec based on clock speed alone. I think it used to take around 3600.

My conclusion is there is definitely something lacking in the Intel Mac compilation.

C
Team MacNN
ID: 8668 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
1 · 2 · Next

Message boards : Number crunching : App v0.10

©2024 Astroinformatics Group