| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Community
Forum: Chat Room
Thread: How well our your computers performing? |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 13
|
|
| Author |
|
|
pcwr
Ace Cruncher England Joined: Sep 17, 2005 Post Count: 10903 Status: Offline Project Badges: 
|
Okay, here's one for you performance stats followers.
----------------------------------------Take your number of PV WUs times the number of total days of cache, divide by 10 then divide by the number of processors. Do not add up the hours of your outstanding PVs, just the number of WUs you have in PV state. eg 36*5 / 10 / 7 = 1.02 I currently have 36 PV WUs, 5 days worth of total cache, 1 quad, 1 dual , 1 single My current figure is above 2.5 If you have 50 PV, 3 cache, 7 processors you get 50*3/10/7 = 2.1 Higher the figure you get, the better your machines are working. What is yours? Patrick  [Edit 1 times, last edit by pcwr at Mar 16, 2011 11:39:09 PM] |
||
|
|
Powhatan
Advanced Cruncher Joined: Oct 20, 2009 Post Count: 58 Status: Offline Project Badges: 
|
What about large numbers of PV due to wingmen slow to crunch or no replies or aborts and resends?
|
||
|
|
pcwr
Ace Cruncher England Joined: Sep 17, 2005 Post Count: 10903 Status: Offline Project Badges:
|
What about large numbers of PV due to wingmen slow to crunch or no replies or aborts and resends? The simple formula that I came up with will show that you will have a higher figure, therefore the better your installations are working. If your wingmen were more efficient at returning WUs, your PVs would be lower, therefore leading to a lower figure. By having a shorter cache, you may have more PVs, but the formula would basically compensate. eg 25*5/10/5 = 2.5 35*3/10/5 = 2.1 I didn't think that one would want to calculate how many PVs you have for each project and then multiply those figures by weighted fractions depending on which project it was.  Patrick [Edit 1 times, last edit by pcwr at Mar 17, 2011 8:08:51 AM] |
||
|
|
Ingleside
Veteran Cruncher Norway Joined: Nov 19, 2005 Post Count: 974 Status: Offline Project Badges:
|
Higher the figure you get, the better your machines are working. Hmm... by my calculations, my figure is... 0.0000625.   "I make so many mistakes. But then just think of all the mistakes I don't make, although I might." |
||
|
|
petehardy
Senior Cruncher USA Joined: May 4, 2007 Post Count: 318 Status: Offline Project Badges: 
|
Here's mine:
----------------------------------------470*1/10/44=1.068181818181818   "Patience is a virtue", I can't wait to learn it! |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Start your computer or lower your cache ;-)
... but the value is rather meaningless as when you switch from HCC1 at 1 day cache to FAAH, you see the value plummeting, simply because the run times per task are totally different and one is quorum 2 and runs short and the other is long but zero redundant. I'm going by hours in PV per mean core day efficiency, thus get 1.78 or at a mean contribution efficiency of 86/100th per core day (86%), have a factor of 1.78 in PV, or nearly 37 hours per core in PV... That was the day before when posting elsewhere about this and then a higher force kicked in and made it drop 0.5... suddenly the next 15 HCMD2 validated instantly though not in cache for longer than 18 hours and a bunch of the slaggers found a wingman. From 43 results to 29 results in inmateship. --//-- |
||
|
|
pcwr
Ace Cruncher England Joined: Sep 17, 2005 Post Count: 10903 Status: Offline Project Badges:
|
Here's mine: 470*1/10/44=1.068181818181818 So your cache is only 30 mins / processor? No wonder you have a large PV. Patrick [Edit 1 times, last edit by pcwr at Mar 17, 2011 11:52:43 AM] |
||
|
|
pcwr
Ace Cruncher England Joined: Sep 17, 2005 Post Count: 10903 Status: Offline Project Badges:
|
Start your computer or lower your cache ;-) ... but the value is rather meaningless as when you switch from HCC1 at 1 day cache to FAAH, you see the value plummeting, simply because the run times per task are totally different and one is quorum 2 and runs short and the other is long but zero redundant. I'm going by hours in PV per mean core day efficiency, thus get 1.78 or at a mean contribution efficiency of 86/100th per core day (86%), have a factor of 1.78 in PV, or nearly 37 hours per core in PV... That was the day before when posting elsewhere about this and then a higher force kicked in and made it drop 0.5... suddenly the next 15 HCMD2 validated instantly though not in cache for longer than 18 hours and a bunch of the slaggers found a wingman. From 43 results to 29 results in inmateship. --//-- Yes I know that some projects have no wingmen whereas some have a large number. I was just trying to come up with a simple formula to see how one's computers were performing. Yes you can multiply by the length of the WUs and multiply by the number of wingmen required per WU etc, but which would you prefer to calculate, 4 figures or 25+? Patrick |
||
|
|
pcwr
Ace Cruncher England Joined: Sep 17, 2005 Post Count: 10903 Status: Offline Project Badges:
|
Higher the figure you get, the better your machines are working. Hmm... by my calculations, my figure is... 0.0000625. I take it most of your projects don't require wingmen and that your computer(s) are not running 24/7. Patrick |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Start your computer or lower your cache ;-) ... but the value is rather meaningless as when you switch from HCC1 at 1 day cache to FAAH, you see the value plummeting, simply because the run times per task are totally different and one is quorum 2 and runs short and the other is long but zero redundant. I'm going by hours in PV per mean core day efficiency, thus get 1.78 or at a mean contribution efficiency of 86/100th per core day (86%), have a factor of 1.78 in PV, or nearly 37 hours per core in PV... That was the day before when posting elsewhere about this and then a higher force kicked in and made it drop 0.5... suddenly the next 15 HCMD2 validated instantly though not in cache for longer than 18 hours and a bunch of the slaggers found a wingman. From 43 results to 29 results in inmateship. --//-- Yes I know that some projects have no wingmen whereas some have a large number. I was just trying to come up with a simple formula to see how one's computers were performing. Yes you can multiply by the length of the WUs and multiply by the number of wingmen required per WU etc, but which would you prefer to calculate, 4 figures or 25+? Patrick I don't, Excel does ;-) |
||
|
|
|