Message boards : Number crunching : sixtrack 4.67
Message board moderation

To post messages, you must log in.

1 · 2 · Next

AuthorMessage
_heinz

Send message
Joined: 27 Nov 05
Posts: 1
Credit: 49,615
RAC: 0
Message 13337 - Posted: 12 Apr 2006, 21:26:41 UTC

I use a Pentium4, 2,6GHz and I need 3h 32min per WU
under BOINC 5.2.13
Is such a long time per WU OK, or is my machine too slow??
ID: 13337 · Report as offensive     Reply Quote
Gaspode the UnDressed

Send message
Joined: 1 Sep 04
Posts: 506
Credit: 118,619
RAC: 0
Message 13338 - Posted: 12 Apr 2006, 22:50:04 UTC

3h30 seems about right. Some will go much faster. Some may take a while longer. If you're concerned that these WUs take so long try CPDN - a single Wu there can take weeks to run...

Gaspode the UnDressed
http://www.littlevale.co.uk
ID: 13338 · Report as offensive     Reply Quote
River~~

Send message
Joined: 13 Jul 05
Posts: 456
Credit: 75,142
RAC: 0
Message 13343 - Posted: 13 Apr 2006, 5:54:41 UTC - in response to Message 13337.  

I use a Pentium4, 2,6GHz and I need 3h 32min per WU
under BOINC 5.2.13
Is such a long time per WU OK, or is my machine too slow??


Try using a 500MHz box - crunch times of 18 hours (& sometimes over a day)

;-)

Seriously, if your work comes in before the deadline, then your box is fast enough. The project team set the deadline to what they need: trust them.

Especially on this project where the team are more proactive on deadlines than most teams - here deadlines vary from 3 days up to 14, whereas most projects set a single length of deadline for all work.

And as Mike says, your times seem about right for the speed of box.

River~~
ID: 13343 · Report as offensive     Reply Quote
River~~

Send message
Joined: 13 Jul 05
Posts: 456
Credit: 75,142
RAC: 0
Message 13344 - Posted: 13 Apr 2006, 5:59:29 UTC - in response to Message 13337.  

...


by the way, I hope you don't mind me emntioning this:

It is better to put your statistics into your sig (see forum preferences) than to include the bbcode in the post itself. That way the sig does not get quoted into replies.

Also it saves you a whole load of cut & paste as when making a post you just tick or untick a box in order to choose whether or not to add your standard sig.

R~~
ID: 13344 · Report as offensive     Reply Quote
River~~

Send message
Joined: 13 Jul 05
Posts: 456
Credit: 75,142
RAC: 0
Message 13345 - Posted: 13 Apr 2006, 8:24:41 UTC

PS:

also please don't panic if you get a few WU that run for under half an hour - that also is normal on this project. Of the short-runners, there are some that complete successfully in under a minute and some that run for a few dozen minutes.

R~~
ID: 13345 · Report as offensive     Reply Quote
stormdog

Send message
Joined: 30 Sep 04
Posts: 4
Credit: 73,494
RAC: 0
Message 13346 - Posted: 13 Apr 2006, 10:00:41 UTC

I'm jumping in with the similar question. Recently I've upgraded to Athlon 64 3800+ Dual core. Normal WU now takes 5 hour 20 minutes to complete on each core. This looks extremely slow for me, especially comparing with my old Athlon 1300. Is it normal? I'm using BOINC 5.2.13 and Win2K SP4.

Best Regards,
Andrew
ID: 13346 · Report as offensive     Reply Quote
J Langley

Send message
Joined: 31 Dec 05
Posts: 68
Credit: 8,691
RAC: 0
Message 13348 - Posted: 13 Apr 2006, 11:04:36 UTC - in response to Message 13346.  

I'm jumping in with the similar question. Recently I've upgraded to Athlon 64 3800+ Dual core. Normal WU now takes 5 hour 20 minutes to complete on each core. This looks extremely slow for me, especially comparing with my old Athlon 1300. Is it normal? I'm using BOINC 5.2.13 and Win2K SP4.

Best Regards,
Andrew


I think that is about right. My Athlon XP 2800+ would take around 7 hours to complete 2 WUs. I believe that for an X2, the 3800 refers to the total performance of the CPU, not the performance of each core. (But I could be wrong.)
ID: 13348 · Report as offensive     Reply Quote
achim.huber

Send message
Joined: 13 Jul 05
Posts: 5
Credit: 451,738
RAC: 0
Message 13351 - Posted: 13 Apr 2006, 14:11:24 UTC - in response to Message 13346.  

I'm jumping in with the similar question. Recently I've upgraded to Athlon 64 3800+ Dual core. Normal WU now takes 5 hour 20 minutes to complete on each core. This looks extremely slow for me, especially comparing with my old Athlon 1300. Is it normal? I'm using BOINC 5.2.13 and Win2K SP4.

Best Regards,
Andrew


It is extremly slow. I have also a Athlon 64 X2 3800+ using BOINC 5.2.13 and WinXP SP2. A normal WU should take 3 hours on each core.
ID: 13351 · Report as offensive     Reply Quote
Stefan

Send message
Joined: 13 Apr 06
Posts: 6
Credit: 142
RAC: 0
Message 13352 - Posted: 13 Apr 2006, 14:45:53 UTC

Is there an optimized client for LHC like there is for SETI or is it not optimizable?
ID: 13352 · Report as offensive     Reply Quote
achim.huber

Send message
Joined: 13 Jul 05
Posts: 5
Credit: 451,738
RAC: 0
Message 13353 - Posted: 13 Apr 2006, 15:31:32 UTC - in response to Message 13352.  
Last modified: 13 Apr 2006, 15:35:16 UTC

Is there an optimized client for LHC like there is for SETI or is it not optimizable?


No. They won't make an optimized client because thy need very accurate results.
ID: 13353 · Report as offensive     Reply Quote
Gaspode the UnDressed

Send message
Joined: 1 Sep 04
Posts: 506
Credit: 118,619
RAC: 0
Message 13354 - Posted: 13 Apr 2006, 16:59:37 UTC - in response to Message 13352.  

Is there an optimized client for LHC like there is for SETI or is it not optimizable?


Sixtrack is a benchmark application use by compiler writers to optimise their compilers. It is, by default, optimised.


Gaspode the UnDressed
http://www.littlevale.co.uk
ID: 13354 · Report as offensive     Reply Quote
Stefan

Send message
Joined: 13 Apr 06
Posts: 6
Credit: 142
RAC: 0
Message 13355 - Posted: 13 Apr 2006, 17:04:30 UTC - in response to Message 13354.  
Last modified: 13 Apr 2006, 17:29:08 UTC

Is there an optimized client for LHC like there is for SETI or is it not optimizable?


Sixtrack is a benchmark application use by compiler writers to optimise their compilers. It is, by default, optimised.



Ah ok I guess precision is better in this case than speed
ID: 13355 · Report as offensive     Reply Quote
Colin Porter

Send message
Joined: 14 Jul 05
Posts: 35
Credit: 71,636
RAC: 0
Message 13356 - Posted: 13 Apr 2006, 17:32:10 UTC - in response to Message 13355.  
Last modified: 13 Apr 2006, 17:32:37 UTC


I see...now is there a shortage of work or is something wrong with my computer?


Probably nothing wrong with your comp. It's just normal LHC.

We get a batch of WU's to crunch then nothing for a while (sometimes weeks) while results are analysed. Then it's back to work again.
ID: 13356 · Report as offensive     Reply Quote
stormdog

Send message
Joined: 30 Sep 04
Posts: 4
Credit: 73,494
RAC: 0
Message 13364 - Posted: 14 Apr 2006, 14:45:39 UTC - in response to Message 13351.  


It is extremly slow. I have also a Athlon 64 X2 3800+ using BOINC 5.2.13 and WinXP SP2. A normal WU should take 3 hours on each core.


It tooks some time to check my guess. I turned off Cool'n'Quiet driver. Now it crunches two times faster(not only for LHC@Home). I guess Cool'n'Quiet doesn't really work as expected. According to my understanding a box shouldn't be slower if Cool'n'Quiet is on AND the box is fully busy.

Well, the cruel reality proves the theory is far from practice. :-) Now the temprature of the processor is around 60C (20C higher) and the cooler is rotating 2 times faster (a bit more 3Kcycles). So far I'm happy, but should wait till the warm comes.

The strange thing is Benchmarking shows the same results whenever Cool'n'Quiet is off or on. But still crunches two times faster...

Best Regards,
Andrew
ID: 13364 · Report as offensive     Reply Quote
achim.huber

Send message
Joined: 13 Jul 05
Posts: 5
Credit: 451,738
RAC: 0
Message 13365 - Posted: 14 Apr 2006, 17:38:31 UTC - in response to Message 13364.  


[/quote]

The strange thing is Benchmarking shows the same results whenever Cool'n'Quiet is off or on. But still crunches two times faster...

[/quote]

I don't use Cool'n'Quiet because this feature only helps when the processor doesn't do anything and I always run BOINC when my Computer is on.
The reason that BOINC Benchmarks are the same is that BOINC runs with normal priority. So even with Cool'n'Quiet turned on the processor runs with full speed. The science apps run with lowest priority and with Cool'n'Quiet turned on your processor only runs with 1 GHz instead of 2 GHz.
ID: 13365 · Report as offensive     Reply Quote
River~~

Send message
Joined: 13 Jul 05
Posts: 456
Credit: 75,142
RAC: 0
Message 13372 - Posted: 15 Apr 2006, 10:49:01 UTC - in response to Message 13364.  

... I guess Cool'n'Quiet doesn't really work as expected. According to my understanding a box shouldn't be slower if Cool'n'Quiet is on AND the box is fully busy.


When they tell you that, they mean a box should not feel any slower to a "live" user. They do not mean the box runs background tasks as fast - that is where they make the savings. In all fairness if the background tasks take less than 50% of real time, then there would be no humnan-visible delays.

The fairest thing to say is that C'n'Q isn't designed with DC projects in mind. Even so there are two cases where it might be useful

1. to save power costs, if that matters to you - you will most likely find you do more crunching per pennyworth of electricity when C'n'Q is on - ie half the crunching but less than half the power drawn

2. to extend battery life on a laptop (tho personally I'd never run BOINC at all while on battery)

ID: 13372 · Report as offensive     Reply Quote
Travis DJ

Send message
Joined: 29 Sep 04
Posts: 196
Credit: 207,040
RAC: 0
Message 13393 - Posted: 17 Apr 2006, 3:38:01 UTC - in response to Message 13364.  
Last modified: 17 Apr 2006, 3:49:23 UTC

I turned off Cool'n'Quiet driver. Now it crunches two times faster(not only for LHC@Home). I guess Cool'n'Quiet doesn't really work as expected. According to my understanding a box shouldn't be slower if Cool'n'Quiet is on AND the box is fully busy.


This topic has been covered before. CnQ will idle the cpu when it isn't under USER initiated loads. When BOINC is installed and is allowed to run as a service, THAT is what makes the difference. So when the user is not producing a workload (i.e. the PC is idling) CnQ will reduce the CPU speed to the lowest speed to match user workload. BOINC will crunch, but at low speed - this is default behavior for all CnQ enabled systems. The only fixes are to either install BOINC to run only when a user is logged in, turn CnQ off in BIOS or easiest yet, change your power profile to "Always On" in Windows XP/2000.

ID: 13393 · Report as offensive     Reply Quote
stormdog

Send message
Joined: 30 Sep 04
Posts: 4
Credit: 73,494
RAC: 0
Message 13397 - Posted: 17 Apr 2006, 9:20:17 UTC - in response to Message 13393.  

This topic has been covered before. CnQ will idle the cpu when it isn't under USER initiated loads. When BOINC is installed and is allowed to run as a service, THAT is what makes the difference. So when the user is not producing a workload (i.e. the PC is idling) CnQ will reduce the CPU speed to the lowest speed to match user workload. BOINC will crunch, but at low speed - this is default behavior for all CnQ enabled systems. The only fixes are to either install BOINC to run only when a user is logged in, turn CnQ off in BIOS or easiest yet, change your power profile to "Always On" in Windows XP/2000.


This is not entirely true. My BOINC doesn't isntalled as a service. And the power profile of my box is "Always ON". CnQ is on leaded to greater cruch time anyway. It seems like achim.huber post is totally correct and CnQ just takes thread priority into account.
ID: 13397 · Report as offensive     Reply Quote
Travis DJ

Send message
Joined: 29 Sep 04
Posts: 196
Credit: 207,040
RAC: 0
Message 13407 - Posted: 19 Apr 2006, 5:02:44 UTC - in response to Message 13397.  
Last modified: 19 Apr 2006, 5:02:54 UTC

This is not entirely true. My BOINC doesn't isntalled as a service. And the power profile of my box is "Always ON". CnQ is on leaded to greater cruch time anyway. It seems like achim.huber post is totally correct and CnQ just takes thread priority into account.


Out of curiosity, which cpu driver is in use? The one MS provides in SP2 or the one from AMD's site? In the past (with AMD's driver) the fix was either 'always on' or CnQ in bios set to 'disabled'.
ID: 13407 · Report as offensive     Reply Quote
stormdog

Send message
Joined: 30 Sep 04
Posts: 4
Credit: 73,494
RAC: 0
Message 13410 - Posted: 19 Apr 2006, 12:59:54 UTC - in response to Message 13407.  


Out of curiosity, which cpu driver is in use?


I have AMD's driver, which comes with my motherboard, so it might be not the latest one.

Regards,
Andrew

ID: 13410 · Report as offensive     Reply Quote
1 · 2 · Next

Message boards : Number crunching : sixtrack 4.67


©2024 CERN