| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Support
Forum: BOINC Agent Support
Thread: odd difference between vista and ubuntu! |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 21
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
A power management issue remains the most likely issue. This is the only cause that will affect reported runtime by a large amount over all projects, when benchmarks are normal.
You may need to disable Cool'n'Quiet in the BIOS. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hello czahn,
Didactylos is right. The new AMD chips have much more advanced power features. Better describe your system. Belarc Advisor at http://www.belarc.com/free_download.html is a convenient utility. Lawrence |
||
|
|
czahn
Cruncher Joined: May 3, 2007 Post Count: 16 Status: Offline Project Badges: 
|
cool and quiet is disabled and has been ,there are no other power management features enabled that I know of unless there are some others in vista that are not known to me.
The only thing that has changed is the install of vista64 and 4 addl gigs of ram. the bios is the same as it was before vista and I have gone thru it to look for anything strange. the setup is a 9850 be amd chip, asus m3a78t motheboard ,6 gigs of ram,fresh install of vista 64,no other programs running but boinc and system programs ,160gig hdd ,cool and quiet disabled,acc disabled. Let me know if there any other settings in vista to try to toggle. also ran belarc and everything looks normal. thanks to all that are helping me figure this out if you need more info let me know thanks! |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
One last question, then: are you reporting CPU time or wallclock time?
Are you reporting final CPU times or estimated times? Okay, two questions.... |
||
|
|
czahn
Cruncher Joined: May 3, 2007 Post Count: 16 Status: Offline Project Badges:
|
the times are actual completion times at 100% ready to report, not the estimated time ,which can be very different and way off.
|
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Perhaps you are simply seeing some longer work units.
Can you calculate granted credit/hour for both cases? |
||
|
|
czahn
Cruncher Joined: May 3, 2007 Post Count: 16 Status: Offline Project Badges:
|
here is what I am going to try ,I am going to set it up to dual boot both vista and ubuntu and see what difference in the times are .
I am going to do this to make sure it is nothing in the bios and is a dedicated vista problem. It could be longer times ,but it just seems odd that in a difference of about 2 hours of install time for vista I start getting all longer work units ,while the intel vista box Is still running them out at about 4 hours each . by the way whats the best way to calculate the granted points per hour? I have compared 2 units from both boxes and for the same time the box is coming up short by about 25+ points. and it is also way over estimating the claimed points . |
||
|
|
BobCat13
Senior Cruncher Joined: Oct 29, 2005 Post Count: 295 Status: Offline Project Badges:
|
here is what I am going to try ,I am going to set it up to dual boot both vista and ubuntu and see what difference in the times are . I am going to do this to make sure it is nothing in the bios and is a dedicated vista problem. It could be longer times ,but it just seems odd that in a difference of about 2 hours of install time for vista I start getting all longer work units ,while the intel vista box Is still running them out at about 4 hours each . Here are my results of a very small test on the HFCC application in Linux vs. Windows. This was run on a single core AMD Athlon XP 2800+ with 1GB PC3200 on Win XP SP3 and Puppy Linux 4.1. Crunched the same task on each OS and the resulting times were: Linux 7:13:27 Win XP 9:44:37 I will now run each OS long enough to get at least 10 single replication valid results and calculate their granted credits per hour. by the way whats the best way to calculate the granted points per hour? I have compared 2 units from both boxes and for the same time the box is coming up short by about 25+ points. and it is also way over estimating the claimed points . Use the Results Status page: https://secure.worldcommunitygrid.org/ms/viewBoincResults.do Find the Valid HFCC tasks under each OS you use on the Phenom box and divide the Granted credits by the CPU Time (Hours). You might want to click on each task's name to be sure they are single replication tasks. Using tasks with more than one replication will not give you accurate numbers to check the applications' performance as your granted credit can be skewed by a wingman's high or low claim. You can safely ignore the over claims as the 64 bit boinc client benchmarks higher on the integer than the 32 bit version. Since HFCC uses floating point mostly, the increased integer speed of the 64 bit OS doesn't help speed up the crunching thus resulting in an over claim on the credit. |
||
|
|
czahn
Cruncher Joined: May 3, 2007 Post Count: 16 Status: Offline Project Badges:
|
@ bobcat13 Thank you for assuring me that I am not crazy!!
I am going to run the same type of test tonight to see the difference. |
||
|
|
BobCat13
Senior Cruncher Joined: Oct 29, 2005 Post Count: 295 Status: Offline Project Badges:
|
I will now run each OS long enough to get at least 10 single replication valid results and calculate their granted credits per hour. Here are the results from 10 tasks each under Puppy Linux 4.1 and Win XP SP3 on the Athlon XP 2800+: Linux: min. 12.0978 ---- max. 12.9599 ---- avg. 12.4925 Win XP: min. 9.8869 ---- max. 10.3178 ---- avg. 10.0454 Next, I will run the same test on a Q6600 on Vista x64 and Ubuntu x64, but will use 100 single replication results. |
||
|
|
|