Index  | Recent Threads  | Unanswered Threads  | Who's Active  | Guidelines  | Search
 

Quick Go »
No member browsing this thread
Thread Status: Active
Total posts in this thread: 254
Posts: 254   Pages: 26   [ Previous Page | 9 10 11 12 13 14 15 16 17 18 | Next Page ]
[ Jump to Last Post ]
Post new Thread
Author
Previous Thread This topic has been viewed 523869 times and has 253 replies Next Thread
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: Computing for Sustainable Water Problems Thread

I wonder how fast Linux will run them comparatively to XP & 7 ?
Once they get Linux sorted out that is... wink
I have the choice of XP x64 or Linux x64 so I hope one of them will do well...
[Apr 24, 2012 2:57:51 PM]   Link   Report threatening or abusive post: please login first  Go to top 
BSD
Senior Cruncher
Joined: Apr 27, 2011
Post Count: 224
Status: Offline
Reply to this Post  Reply with Quote 
Re: Computing for Sustainable Water Problems Thread

On this Win7 X64 laptop that's running TThrottle with CPU set to 50%, normally the WU completed CPU time is half that of the Elapsed time. But, these WU for this project the CPU time is about 25% that of Elapsed. Very strange indeed. Going to change TTrottle to 100% to see what happens. This of course will make the CPU run hotter.

CPU: 04:13:36 Hrs
Elapsed: 19:11:52

Not understanding the combination of CPU set at 50% (what? CPU time or Cores?) and TThrottle. That could be working against each other, so would recommend 100% CPU time [if that's what you're restricting] and let TThrottle limit the CPU time through the chosen temp ceiling for maximum throughput at any part of the day [hot noons, cool nights]

--//--

For laptops or desktops that I want reduced heat and fan noise.

BOINC Manager preferences: Use at most 100% CPU time
TThrottle Run percentage: 50 Max CPU %

I use the Max CPU % vice Set core temp setting. This has worked well for me on other projects. Might be the temp setting is set to a lower threshold and is slowing down the CPU lower than 50% on this laptop, I check that out.
[Apr 24, 2012 3:04:33 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Mathilde2006
Senior Cruncher
Germany
Joined: Sep 30, 2006
Post Count: 269
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: Computing for Sustainable Water Problems Thread

I wonder how fast Linux will run them comparatively to XP & 7 ?
Once they get Linux sorted out that is... wink
I have the choice of XP x64 or Linux x64 so I hope one of them will do well...



With my I7 -920:

XP-64 ~ 7 hours (6.9- 7.1)

Ubuntu 11.10 x64 5.25 hours - only one WU tested.

Just installed Vista X64 (currently running update installation = no SP)
25 % reached after 1:22 h
----------------------------------------

[Apr 24, 2012 4:18:26 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: Computing for Sustainable Water Problems Thread

We are looking into the runtime differences for CFSW. It is not a 32/64 bit issue but XP is running much slower than Win7.
Between 20Apr12 06:37:40 UTC and 22Apr12 04:57:39, the claimed points for CFSW units on my Athlon 64 3500+ running XP Pro SP3 dropped from about 18 per hour to about 7.7 (and are still dropping, now about 7.0), despite all the units still taking very close to 10.0 hours. By comparison, SN2S units on the same machine are still claiming about 14.4 per hour. What might be the explanation for that drop?
Examples:
---------------
cfsw_ 0020_ 00020074_ 1-- 605 Valid 18/04/12 03:16:04 21/04/12 19:48:15 5.07 111.5 / 145.9
cfsw_ 0020_ 00020074_ 0-- 605 Valid 18/04/12 01:47:03 20/04/12 06:37:40 9.97 180.4 / 145.9 << me
claimed 18.09, granted 14.63 per hour
---------------
cfsw_ 0058_ 00058031_ 1-- 605 Valid 18/04/12 19:37:45 23/04/12 23:40:04 12.67 181.8 / 129.4
cfsw_ 0058_ 00058031_ 0-- 605 Valid 18/04/12 19:37:15 22/04/12 04:57:39 9.96 77.1 / 129.4 << me
claimed 7.74, granted 12.99 per hour
---------------
SN2S_ AH003972_ 0000032_ 0450_ 0-- 613 Valid 21/04/12 07:32:32 22/04/12 01:06:39 5.53 112.2 / 117.6
SN2S_ AH003972_ 0000032_ 0450_ 1-- 613 Valid 21/04/12 07:32:06 23/04/12 06:18:17 8.59 123.1 / 117.6 << me
claimed 14.33, granted 13.69 per hour
[Apr 24, 2012 4:38:28 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: Computing for Sustainable Water Problems Thread

We are looking into the runtime differences for CFSW. It is not a 32/64 bit issue but XP is running much slower than Win7. At this point the cause is unknown but we are investigating.


I look forward to reading your results! Thanks, Jon.
[Apr 24, 2012 4:43:39 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: Computing for Sustainable Water Problems Thread

On this Win7 X64 laptop that's running TThrottle with CPU set to 50%, normally the WU completed CPU time is half that of the Elapsed time. But, these WU for this project the CPU time is about 25% that of Elapsed. Very strange indeed. Going to change TTrottle to 100% to see what happens. This of course will make the CPU run hotter.

CPU: 04:13:36 Hrs
Elapsed: 19:11:52

Not understanding the combination of CPU set at 50% (what? CPU time or Cores?) and TThrottle. That could be working against each other, so would recommend 100% CPU time [if that's what you're restricting] and let TThrottle limit the CPU time through the chosen temp ceiling for maximum throughput at any part of the day [hot noons, cool nights]

--//--

For laptops or desktops that I want reduced heat and fan noise.

BOINC Manager preferences: Use at most 100% CPU time
TThrottle Run percentage: 50 Max CPU %

I use the Max CPU % vice Set core temp setting. This has worked well for me on other projects. Might be the temp setting is set to a lower threshold and is slowing down the CPU lower than 50% on this laptop, I check that out.

Never noticed that TThrottle had a CPU % setting feature, so loaded it up and indeed you can enter a percent there. I'd never use that, just the temp ceiling, and tweak that if it's requirement to have a percent of CPU juice for ad hoc user disposal [the lackless response you're wanting]. Had CPU % at 100% [default I suppose]. The more tools used, the more risk of conflicts.

I've noticed that TThrottle was not working to satisfaction on my new 8 core toy, so unloaded it after seeing a big drop in throughput. It seemed to have been working against the system's own CoolSense, that now works automatic to run in performance optimized mode when stationary [on desk] and in coolest mode when on lap (with a notepal 3 fan cooler pad below it to keep certain 'hardware' at 34C). Since, running 100% and never an issue with CFSW.

--//--
[Apr 24, 2012 5:00:17 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: Computing for Sustainable Water Problems Thread

On credit, visit the charts on offer http://bit.ly/WCGCPH. CFSW is now the lowest per-hour credit rewarding science at WCG. HPF2 used to be the incumbent for that spot. Soon as a badge level aspired has been reached it's back to med science [yes I'm cherry picking too at WCG].

--//--
----------------------------------------
[Edit 1 times, last edit by Former Member at Apr 24, 2012 5:13:12 PM]
[Apr 24, 2012 5:03:28 PM]   Link   Report threatening or abusive post: please login first  Go to top 
BSD
Senior Cruncher
Joined: Apr 27, 2011
Post Count: 224
Status: Offline
Reply to this Post  Reply with Quote 
Re: Computing for Sustainable Water Problems Thread

On this Win7 X64 laptop that's running TThrottle with CPU set to 50%, normally the WU completed CPU time is half that of the Elapsed time. But, these WU for this project the CPU time is about 25% that of Elapsed. Very strange indeed. Going to change TTrottle to 100% to see what happens. This of course will make the CPU run hotter.

CPU: 04:13:36 Hrs
Elapsed: 19:11:52

Not understanding the combination of CPU set at 50% (what? CPU time or Cores?) and TThrottle. That could be working against each other, so would recommend 100% CPU time [if that's what you're restricting] and let TThrottle limit the CPU time through the chosen temp ceiling for maximum throughput at any part of the day [hot noons, cool nights]

--//--

For laptops or desktops that I want reduced heat and fan noise.

BOINC Manager preferences: Use at most 100% CPU time
TThrottle Run percentage: 50 Max CPU %

I use the Max CPU % vice Set core temp setting. This has worked well for me on other projects. Might be the temp setting is set to a lower threshold and is slowing down the CPU lower than 50% on this laptop, I check that out.

Never noticed that TThrottle had a CPU % setting feature, so loaded it up and indeed you can enter a percent there. I'd never use that, just the temp ceiling, and tweak that if it's requirement to have a percent of CPU juice for ad hoc user disposal [the lackless response you're wanting]. Had CPU % at 100% [default I suppose]. The more tools used, the more risk of conflicts.

I've noticed that TThrottle was not working to satisfaction on my new 8 core toy, so unloaded it after seeing a big drop in throughput. It seemed to have been working against the system's own CoolSense, that now works automatic to run in performance optimized mode when stationary [on desk] and in coolest mode when on lap (with a notepal 3 fan cooler pad below it to keep certain 'hardware' at 34C). Since, running 100% and never an issue with CFSW.

--//--

Intresting, One of those "D'oh" moments. d oh whistling

This laptop has an Intel i7-2720QM CPU with maximum CPU operating temp 100C.

TThrottle had Temp Set Core set to 60C which I had never changed that setting before. I've always changed the Max CPU.

Tweaking the settings...

Set Max CPU 100%, Temp Set Core 60C = Task Mgr shows 50% CPU usage
Set Max CPU 100%, Temp Set Core 80C = Task Mgr shows 100% CPU usage

Set Max CPU 50%, Temp Set Core 60C = Task Mgr shows 20% CPU usage
Set Max CPU 50%, Temp Set Core 80C = Task Mgr shows 50% CPU usage

So, for this laptop I was causing my own extra slowness by setting the Temp Set Core too low.

Thanks Sek hugs
[Apr 24, 2012 6:47:48 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Dataman
Ace Cruncher
Joined: Nov 16, 2004
Post Count: 4865
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: Computing for Sustainable Water Problems Thread

We can identify all workunits that were impacted and we will need to re-run those workunits to be certain that only valid results are being returned to the researchers. This will require us re-running about ~15,000 workunits. We apologize for the inconvenience.


Are these resends going out as _1 & _2 wu's?
----------------------------------------


[Apr 24, 2012 6:51:10 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: Computing for Sustainable Water Problems Thread

We can identify all workunits that were impacted and we will need to re-run those workunits to be certain that only valid results are being returned to the researchers. This will require us re-running about ~15,000 workunits. We apologize for the inconvenience.
Are these resends going out as _1 & _2 wu's?
I've received these out of sequence today, as Quorum 1, which look like they may be re-runs of earlier ones, all _0:
------------------------
cfsw_ 0006_ 00006403_ 0-- - In Progress 24/04/12 03:50:27 04/05/12 03:50:27 0.00 0.0 / 0.0
cfsw_ 0041_ 00041037_ 0-- - In Progress 24/04/12 04:53:30 04/05/12 04:53:30 0.00 0.0 / 0.0
cfsw_ 0084_ 00084782_ 0-- - In Progress 24/04/12 06:06:07 04/05/12 06:06:07 0.00 0.0 / 0.0
cfsw_ 0098_ 00098209_ 0-- - In Progress 24/04/12 06:43:41 04/05/12 06:43:41 0.00 0.0 / 0.0
[Apr 24, 2012 9:07:22 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Posts: 254   Pages: 26   [ Previous Page | 9 10 11 12 13 14 15 16 17 18 | Next Page ]
[ Jump to Last Post ]
Post new Thread