| View previous topic :: View next topic |
| Author |
Message |
joshvee
Black Belt 4th Degree
Joined: 12 Jul 2003
Posts: 4411
Location: Australia
|
 Posted: Sun Sep 14, 2003 9:26 pm Post subject: |
 |
|
OK next question.....If you are only running 1 instance of F@H on a hyperthreading machine, it uses BOTH virtual CPUs so doesnt it run twice as fast?
Now if you run 2 instances, they use 1 virtual CPU each....well doesn't that mean that each instance is only running half as quickly as 1 instance running both virtual CPUs? and if so, then whats the point of running 2 instances.
_________________
 |
|
| Back to top |
|
 |
joshvee
Black Belt 4th Degree
Joined: 12 Jul 2003
Posts: 4411
Location: Australia
|
| Posted: Sun Sep 14, 2003 9:30 pm Post subject: |
|
|
OK nevermind that last question....I found the answer...
_________________
|
|
| Back to top |
|
|
Tulatin
Enlightened Master
Joined: 11 Dec 2002
Posts: 14664
Location: Canada
|
| Posted: Sun Sep 14, 2003 9:46 pm Post subject: |
|
|
| yep josh, it only takes advantage of 1 proc of a time. You could always run one command line and one gui model though... 4000 posts. |
|
| Back to top |
|
|
joshvee
Black Belt 4th Degree
Joined: 12 Jul 2003
Posts: 4411
Location: Australia
|
| Posted: Sun Sep 14, 2003 11:33 pm Post subject: |
|
|
ok I now have the 2 console clients running (tks Aussie for that link...smurf you should add it to the web page)
next question.....I never ran the console clients before (always ran graphical clients) can I close the consle windows? will the clients still be running? and if so, how do I open the consoles again?
_________________
|
|
| Back to top |
|
|
Aussie
Mobo-fu Master
Joined: 22 Nov 1999
Posts: 10244
Location: Mackay, Queensland, Australia
|
| Posted: Mon Sep 15, 2003 2:16 am Post subject: |
|
|
| Quote: | | I never ran the console clients before (always ran graphical clients) can I close the consle windows? |
Yes.
| Quote: | | will the clients still be running? |
No.
If you want to run the console version but "hide it" this is what you want:-
http://home.comcast.net/~wxdude1/emsite/hideitx.html
I've never used this program so you are on your own. 
Aussie
_________________
Our lives begin to end the day we become silent about things that matter - Martin Luther King |
|
| Back to top |
|
|
joshvee
Black Belt 4th Degree
Joined: 12 Jul 2003
Posts: 4411
Location: Australia
|
| Posted: Mon Sep 15, 2003 4:08 am Post subject: |
|
|
hmm I think I'll just leave it showing.  
Now I'm running 2 clients I have noticed that 1 runs MUCH faster than the other. 1 does a frame every 3.5 mins and the other does a frame every 13 mins. Anyone know why so much difference?
When running only 1 client it did a frame every 6 mins regardless of the WU.
Seems odd 
_________________
|
|
| Back to top |
|
|
Aussie
Mobo-fu Master
Joined: 22 Nov 1999
Posts: 10244
Location: Mackay, Queensland, Australia
|
| Posted: Mon Sep 15, 2003 4:30 am Post subject: |
|
|
The time taken to complete a frame will vary from one work unit to the next, you are running two instances of the console and each is working on a different work unit so it is only natural for the "frame time" to differ.
I also believe that the P4's "virtual CPU's" when hyperthreading are not identical in their capabilities, that is one is stronger than the other.
The variation in work units and the points credited is quite dramatic I have discovered from a little experimentation. Using one machine running 24/7, recording the time taken and points credited for each work unit I have seen the time taken to generate one point vary from 12.25 minutes to 26.16 minutes between work units. That is quite a variation considering that everything other than the work unit was constant.
I would expect the "points generation rate" of a given machine to vary over a much narrower range between work units if the points scoring system was accurate.
Aussie
_________________
Our lives begin to end the day we become silent about things that matter - Martin Luther King |
|
| Back to top |
|
|
joshvee
Black Belt 4th Degree
Joined: 12 Jul 2003
Posts: 4411
Location: Australia
|
| Posted: Mon Sep 15, 2003 5:01 am Post subject: |
|
|
| Aussie wrote: | The time taken to complete a frame will vary from one work unit to the next, you are running two instances of the console and each is working on a different work unit so it is only natural for the "frame time" to differ.
I also believe that the P4's "virtual CPU's" when hyperthreading are not identical in their capabilities, that is one is stronger than the other.
The variation in work units and the points credited is quite dramatic I have discovered from a little experimentation. Using one machine running 24/7, recording the time taken and points credited for each work unit I have seen the time taken to generate one point vary from 12.25 minutes to 26.16 minutes between work units. That is quite a variation considering that everything other than the work unit was constant.
I would expect the "points generation rate" of a given machine to vary over a much narrower range between work units if the points scoring system was accurate.
Aussie |
Hmmm yes makes sence....just checked out the two WUs and the fast one (3min per frame) is only worth 15 points and is running on processor 0 (which would be the real processor.)
The slow one is worth a lot more points (70) and is running on processor 1 (the virtual processor)
_________________
|
|
| Back to top |
|
|
Tulatin
Enlightened Master
Joined: 11 Dec 2002
Posts: 14664
Location: Canada
|
| Posted: Mon Sep 15, 2003 3:28 pm Post subject: |
|
|
| Word of the wise josh - swap the processor that they run on - give the big boy the main (51. real) proc. |
|
| Back to top |
|
|
joshvee
Black Belt 4th Degree
Joined: 12 Jul 2003
Posts: 4411
Location: Australia
|
| Posted: Mon Sep 15, 2003 10:19 pm Post subject: |
|
|
| Tulatin wrote: | | Word of the wise josh - swap the processor that they run on - give the big boy the main (51. real) proc. |
hmm tried that and it didn't make any difference now thats odd.
_________________
|
|
| Back to top |
|
|
|
