| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
| World Community Grid Forums
|
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 16
|
|
| Author |
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
snoekie,
----------------------------------------Think it's time to ask for a copy/paste post of the content in the global_prefs.xml and if existing the global_prefs_override.xml sitting in the BOINC program directory. Also the client_state.xml section which looks like this: <time_stats> These will tell what BOINC uses as preferences and how it translates to effective crunching.
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 1 times, last edit by Sekerob at Jun 24, 2007 6:04:34 PM] |
||
|
|
JmBoullier
Former Community Advisor Normandy - France Joined: Jan 26, 2007 Post Count: 3716 Status: Offline Project Badges:
|
Hi snoekie!
----------------------------------------I think the confusion might come from your understanding of processors with hyperthreading. Before starting grid computing I was thinking like you that a HT processor was some kind of dual processor. But after reading explanations by experts on various fora I have realized my mistake: a P4 HT (my case) is a true single core processor with two threads to feed instructions in the processor. That makes it more efficient when switching between tasks, but there is still only one core to do the job. Therefore, in an "idle" machine, if you run two concurrent WUs on a HT processor each one receives half the computing power and lasts twice longer than if it were alone. In fact, in such a machine, the most efficient is to set the number of processors in BOINC to 1, although it would also run fine with 2 as you could notice, if you forget about estimating when a task will finish. Edit: I forgot to say that the confusion is largely helped by Windows' taskmanager that incorrectly shows 50% when you have only one job running, while the HT processor is actually crunching at 100% (assuming BOINC is throttled at 100%). Anybody, please don't hesitate to correct me if I have misunderstood something. Cheers. Jean. ---------------------------------------- [Edit 1 times, last edit by JmBoullier at Jun 23, 2007 10:38:35 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hello JmBoullier,
Anybody, please don't hesitate to correct me if I have misunderstood something. Your understanding is . . . good enough. Hyperthreading increases the silicon area of the Pentium 4 core by about 5%, but increases total throughput by about 15% normally when running 2 threads. Each individual thread runs slower. Some specialized video-editing programs that understand hyperthreading increase throughput by about 45%, which is probably why Intel is so interested in this technique.However, the WCG programs are all running through the single FPU, so it does not make much sense to run 2 WCG threads on a Pentium 4, particularly if any other programs will run concurrently. Running one FPU thread and one standard program, such as a web browser or word processing program lets hyperthreading show off its capabilities. Running Dragon Natural Speaking through the FPU while trying to run a BOINC program through the FPU will probably slow things down dreadfully. Lawrence |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Sekerob, I am that ignorant that I don't know where to look for that item.
Found, and it is this 1 0 0 0 0 10.000000 1.000000 50.000000 Couldn't find the wording you listed |
||
|
|
JmBoullier
Former Community Advisor Normandy - France Joined: Jan 26, 2007 Post Count: 3716 Status: Offline Project Badges:
|
Lawrence, thank you much for this additional info. Jean.
---------------------------------------- |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
After thinking about it, I think I am beginning to understand.
It would explain some of the delay, I have the processor useage set at 80%, but am puzzled as to why this would take up nearly 3 times the actual time. Thanks for the pointers |
||
|
|
|