| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: GO Fight Against Malaria
Thread: CPU time higher than elapsed time |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 16
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
It looks like more than one CPU was working on the WU.
CPU time should be substantially less than elapsed time (wall-clock-time) on single processor applications, e.g. 0m0.04s CPU time and 10m16.03s wall-clock-time. CPU time is just the actual time taken to complete a task. Wall-clock-time is CPU time plus I/O time plus any programmed delays plus resource delays so our WUs should always show smaller CPU time than wall-clock-time unless BOINC is using more than one processor for the WU. |
||
|
|
nanoprobe
Master Cruncher Classified Joined: Aug 29, 2008 Post Count: 2998 Status: Offline Project Badges: 
|
It looks like more than one CPU was working on the WU. CPU time should be substantially less than elapsed time (wall-clock-time) on single processor applications, e.g. 0m0.04s CPU time and 10m16.03s wall-clock-time. CPU time is just the actual time taken to complete a task. Wall-clock-time is CPU time plus I/O time plus any programmed delays plus resource delays so our WUs should always show smaller CPU time than wall-clock-time unless BOINC is using more than one processor for the WU. These are i7s with HT running 8 WUs. No WU was using more than 1 processor.
In 1969 I took an oath to defend and protect the U S Constitution against all enemies, both foreign and Domestic. There was no expiration date.
  |
||
|
|
Mysteron347
Senior Cruncher Australia Joined: Apr 28, 2007 Post Count: 179 Status: Offline Project Badges: 
|
I gain the impression that GFAM runs two applications serially.
In another forum, I was told that there's a similar observation about an application called Constellation, but that had been put down to some strange Java effect. I wonder whether the elapsed time shown is that of the second application whereas the CPU time is that of the WU as a whole? |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
How about the reverse logic? The controller app that does not take CPU time but a few seconds sends the elapsed time to BOINC as what you see in the manager tasks view. It's permanently loaded from task start to finish. The workers [many per task, a new one loaded with each job in a task], message the CPU time. These are concatenated and recorded in the Result log file. The final value in the log + the extra seconds at end of log are reported as the task CPU time. The same applies to DSFL (same engine) and CEP2 (different engine).
------------------------------------------//-- [Edit 1 times, last edit by Former Member at Nov 28, 2011 4:06:18 PM] |
||
|
|
nanoprobe
Master Cruncher Classified Joined: Aug 29, 2008 Post Count: 2998 Status: Offline Project Badges:
|
I guess it's just one of life's great mysteries.
---------------------------------------- FWIW I did notice that the WUs that showed this behavior would show it immediately after starting and continue until the WU finished. If I see any more I will watch if the time differential increases as the unit is crunched or stays static.
In 1969 I took an oath to defend and protect the U S Constitution against all enemies, both foreign and Domestic. There was no expiration date.
|
||
|
|
yoro42
Ace Cruncher United States Joined: Feb 19, 2011 Post Count: 8976 Status: Offline Project Badges: 
|
Ah, mysteries. Makes life interesting.
---------------------------------------- |
||
|
|
|