+- +-
Say hello if visiting :) by Gecko
11 Jan 2023, 07:43:05 pm

Seti is down again by Mike
09 Aug 2017, 10:02:44 am

Some considerations regarding OpenCL MultiBeam app tuning from algorithm view by Raistmer
11 Dec 2016, 06:30:56 am

Loading APU to the limit: performance considerations by Mike
05 Nov 2016, 06:49:26 am

Better sleep on Windows - new round by Raistmer
26 Aug 2016, 02:02:31 pm

Author Topic: New approach to benchmarking  (Read 12750 times)

Offline Raistmer

  • Working Code Wizard
  • Volunteer Developer
  • Knight who says 'Ni!'
  • *****
  • Posts: 14349
New approach to benchmarking
« on: 05 Dec 2008, 02:39:02 pm »
Now, in the age of multicore systems, stopping BOINC totaly for testing seems unneeded. Moreover, it biases benchmark results cause applications under investigation operate in different environment (they own whole L2 cache that usually shared between cores). It's better to free only single core for benchmarking leaving BOINC on others.
So, my proposal is to create 2 copies of  cc_config.xml file. One with <ncpus>0</ncpus> and second with <ncpus>N-1</ncpus> where N  is number of cores on host.
At benchmark begin, after "net stop boinc" copy should be added to replace current cc_config.xml with edited one.
Then "net start boinc" and benchmarking.
At the end of benchmark another "net stop boinc", copy (to restore original cc_config.xml ) and "net start boinc" should replace single "net start boinc" in current benchmark.

copy destination will depend on BOINC path on your host so I don't post exact command for copy.

Sample cmd file attached.


[attachment deleted by admin]
« Last Edit: 10 Dec 2008, 03:17:16 pm by Raistmer »

 

Welcome, Guest.
Please login or register.
 
 
 
Forgot your password?
Members
Total Members: 97
Latest: ToeBee
New This Month: 0
New This Week: 0
New Today: 0
Stats
Total Posts: 59559
Total Topics: 1672
Most Online Today: 17
Most Online Ever: 983
(20 Jan 2020, 03:17:55 pm)
Users Online
Members: 0
Guests: 21
Total: 21
Powered by EzPortal