Advanced search

Message boards : Number crunching : New applications

Author Message
Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1947
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 8465 - Posted: 15 Apr 2009 | 19:05:25 UTC
Last modified: 15 Apr 2009 | 19:09:54 UTC

We are testing the new applications for Windows and Linux. They should be out tomorrow.

The are very many added features from the science point of view. From a practical point of view the multiplier will pass from 1.5x to 2.0x for everybody (SETI 2.4x). We will give further 25% for results returned within two days.
Speed should be the same.

gdf

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1947
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 8495 - Posted: 16 Apr 2009 | 12:48:09 UTC - in response to Message 8465.

The applications are ready and tested.
We are waiting tomorrow, as today we are submitting several updated WUs.
If we change two things, we would not know what's wrong in case of problems.

Best, gdf

Fuzzy Duck
Send message
Joined: 28 Mar 09
Posts: 6
Credit: 6,972,294
RAC: 0
Level
Ser
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 8515 - Posted: 17 Apr 2009 | 8:19:34 UTC - in response to Message 8495.

I assume that the new apps will d/l automatically?

Profile Paul D. Buck
Send message
Joined: 9 Jun 08
Posts: 1050
Credit: 37,321,185
RAC: 0
Level
Val
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 8529 - Posted: 17 Apr 2009 | 15:50:11 UTC - in response to Message 8515.

I assume that the new apps will d/l automatically?

Yes, they will ... I already have two 6.63 tasks on my queue ... I should be starting them in about 4-6 hours ... I hope they don't error out

jrobbio
Send message
Joined: 13 Mar 09
Posts: 59
Credit: 324,366
RAC: 0
Level

Scientific publications
watwatwatwat
Message 8531 - Posted: 17 Apr 2009 | 23:24:02 UTC - in response to Message 8529.

I assume that the new apps will d/l automatically?

Yes, they will ... I already have two 6.63 tasks on my queue ... I should be starting them in about 4-6 hours ... I hope they don't error out


I managed to complete my first 6.63 task here. The timestep is quite high though. I might try the new Boinc version and see how it copes.

Rob

Profile Paul D. Buck
Send message
Joined: 9 Jun 08
Posts: 1050
Credit: 37,321,185
RAC: 0
Level
Val
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 8535 - Posted: 18 Apr 2009 | 4:01:32 UTC

I completed my first and the T/S was about the same ... the run time slightly higher.

The 4 in work should start pumping out in the next couple hours, the next in line looks to be 8:12 which is about 2 hours longer than before. So, I will have to look to see what is up with that if it is an outlier or just a longer running task ...

More to come I am sure ...

Profile [FVG] bax
Avatar
Send message
Joined: 18 Jun 08
Posts: 29
Credit: 17,772,874
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwat
Message 8537 - Posted: 18 Apr 2009 | 6:20:47 UTC - in response to Message 8465.

We are testing the new applications for Windows and Linux. They should be out tomorrow....


is this related with NO WORK available for PS3 ?


good job guys ;)

Profile mike047
Send message
Joined: 21 Dec 08
Posts: 47
Credit: 7,330,049
RAC: 0
Level
Ser
Scientific publications
watwatwatwatwatwatwat
Message 8545 - Posted: 18 Apr 2009 | 8:24:47 UTC - in response to Message 8537.

We are testing the new applications for Windows and Linux. They should be out tomorrow....


is this related with NO WORK available for PS3 ?


good job guys ;)


ATM, I see plenty of PS3 work and No GPU work..........
____________
mike

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1947
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 8736 - Posted: 22 Apr 2009 | 19:56:39 UTC - in response to Message 8545.

We have performed a further application update to fix one bug which did not allow us to run the new type of workunits.

gdf

Profile Paul D. Buck
Send message
Joined: 9 Jun 08
Posts: 1050
Credit: 37,321,185
RAC: 0
Level
Val
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 8738 - Posted: 22 Apr 2009 | 20:11:31 UTC - in response to Message 8736.

We have performed a further application update to fix one bug which did not allow us to run the new type of workunits.

gdf

So we should start to see 6.64 tasks soon?

jrobbio
Send message
Joined: 13 Mar 09
Posts: 59
Credit: 324,366
RAC: 0
Level

Scientific publications
watwatwatwat
Message 8744 - Posted: 22 Apr 2009 | 20:45:39 UTC - in response to Message 8738.
Last modified: 22 Apr 2009 | 20:47:46 UTC

We have performed a further application update to fix one bug which did not allow us to run the new type of workunits.

gdf

So we should start to see 6.64 tasks soon?


It says the application (note: hardest page to find!) was applied at 6pm so any prepared after then I guess.

Should I cancel the 6.63 in my queue? Edit: Just did and received a 6.64.

Rob

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1947
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 8749 - Posted: 22 Apr 2009 | 21:41:00 UTC - in response to Message 8744.

you don't need to cancel old WUs, these are just fine.
We need the new application for the WUs which we will submit tomorrow.

gdf

Rabinovitch
Avatar
Send message
Joined: 25 Aug 08
Posts: 143
Credit: 64,937,578
RAC: 0
Level
Thr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwat
Message 8754 - Posted: 23 Apr 2009 | 1:19:38 UTC - in response to Message 8736.

We have performed a further application update to fix one bug which did not allow us to run the new type of workunits.

gdf


So it could be very interesting to know, what is this "new type of WUs"... Could you please inform us? :-)

____________
From Siberia with love!

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1947
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 8763 - Posted: 23 Apr 2009 | 7:15:44 UTC - in response to Message 8754.

It is a rather technical but important change, from now on we will be able to use the Amber force fields, which is more accurate for certain applications.

gdf

Profile Paul D. Buck
Send message
Joined: 9 Jun 08
Posts: 1050
Credit: 37,321,185
RAC: 0
Level
Val
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 8804 - Posted: 23 Apr 2009 | 21:21:09 UTC - in response to Message 8763.

It is a rather technical but important change, from now on we will be able to use the Amber force fields, which is more accurate for certain applications.

And all this time I was hoping that we would start using the puce force fields ...

For those mathematically inclined: AMBER force Fields (an acronym for Assisted Model Building and Energy Refinement)...

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1947
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 8807 - Posted: 23 Apr 2009 | 21:54:17 UTC - in response to Message 8804.

Details of the new WUs will be posted soon.

gdf

Profile Hydropower
Avatar
Send message
Joined: 3 Apr 09
Posts: 70
Credit: 6,003,024
RAC: 0
Level
Ser
Scientific publications
watwatwatwatwatwat
Message 8840 - Posted: 24 Apr 2009 | 14:47:43 UTC - in response to Message 8807.
Last modified: 24 Apr 2009 | 15:25:24 UTC

I observe that the 6.64 application runs 4x as long on a seemingly same type of workunit, is that intentional ? It also uses 0.25% CPU time instead of 0.15% with version 6.63. The GPU is much cooler. My impression is that this is not as efficient as before...

Profile Michael Goetz
Avatar
Send message
Joined: 2 Mar 09
Posts: 124
Credit: 7,573,744
RAC: 0
Level
Ser
Scientific publications
watwatwatwatwatwatwatwat
Message 8843 - Posted: 24 Apr 2009 | 15:46:29 UTC - in response to Message 8840.
Last modified: 24 Apr 2009 | 15:54:19 UTC

I observe that the 6.64 application runs 4x as long on a seemingly same type of workunit, is that intentional ? It also uses 0.25% CPU time instead of 0.15% with version 6.63. The GPU is much cooler. My impression is that this is not as efficient as before...


My first 6.64 started running a little while ago. I am not noticing any difference in elapsed time or GPU temp so far, but I am seeing the CPU utilization going up to 25% from 15%, same as you.

My hardware is a Q6600/GTX280(FOC). Your CPU is faster and our GPUs are similar in speed.

I'm not sure what conclusions to draw from this. I find it very interesting and somewhat perplexing that BOTH of us saw 15% CPU utilization under 6.63 and 25% utilization under 6.64, despite having GPUs that run around the same speed (and hence should require the same amount of servicing by the CPU), yet your CPU is faster.

EDIT:

I just noticed that I have queued up a work unit that looks very different than prior work units. I guess this work unit is one of the new ones just released today. Of note is that the estimated run time is double that of the older tasks.

I suspect what you're seeing is also one of the new ones, and perhaps they're just bigger.

Profile Hydropower
Avatar
Send message
Joined: 3 Apr 09
Posts: 70
Credit: 6,003,024
RAC: 0
Level
Ser
Scientific publications
watwatwatwatwatwat
Message 8844 - Posted: 24 Apr 2009 | 15:51:03 UTC - in response to Message 8843.

I bet you are also using BOINC version 6.6.20 ? I just read another thread and decided to test it. I got rid of BOINC 6.6.20 and reinstalled 6.4.7 and bingo, I am back to my old high speed crunching !

Profile Michael Goetz
Avatar
Send message
Joined: 2 Mar 09
Posts: 124
Credit: 7,573,744
RAC: 0
Level
Ser
Scientific publications
watwatwatwatwatwatwatwat
Message 8846 - Posted: 24 Apr 2009 | 15:56:19 UTC - in response to Message 8844.

Yes, I've been using 6.6.20 for a while. It seems to work fine when you only have one GPU, and solves some problems I had with 6.4.7, most notably the failure to keep all CPU cores busy.

P.S. Check the edit I made to my previous post.

Profile Hydropower
Avatar
Send message
Joined: 3 Apr 09
Posts: 70
Credit: 6,003,024
RAC: 0
Level
Ser
Scientific publications
watwatwatwatwatwat
Message 8848 - Posted: 24 Apr 2009 | 16:07:28 UTC - in response to Message 8846.
Last modified: 24 Apr 2009 | 16:09:02 UTC

Hi, I read your edit. I suspected a change in workunit structure at first too, but I can conclude that BOINC is the real cause of the issue. The same workunit I was working on before and which I paused, crunches 4x as fast with BOINC 6.4.7. on the GPU.
(edit)
I can also notice by the GPU temperature, it shot up by 10 degrees after going back to BOINC 6.4.7. It is much more efficient.

Profile Paul D. Buck
Send message
Joined: 9 Jun 08
Posts: 1050
Credit: 37,321,185
RAC: 0
Level
Val
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 8867 - Posted: 24 Apr 2009 | 19:57:23 UTC

The 6.4.x series needs manual tuning to get it to properly feed all GPUs, 6.5.0 does not need this manual tuning.

See the other threads for my comments on 6.6.20, .23 and .24

Profile Michael Goetz
Avatar
Send message
Joined: 2 Mar 09
Posts: 124
Credit: 7,573,744
RAC: 0
Level
Ser
Scientific publications
watwatwatwatwatwatwatwat
Message 8911 - Posted: 25 Apr 2009 | 13:38:13 UTC - in response to Message 8843.

I just noticed that I have queued up a work unit that looks very different than prior work units. I guess this work unit is one of the new ones just released today. Of note is that the estimated run time is double that of the older tasks.

I suspect what you're seeing is also one of the new ones, and perhaps they're just bigger.


As it turns out, that new work unit is now running, and despite having an initial estimate that was much higher, it seems to be running in the same time as the older work units.

GPU temperature is normal, and, oddly enough, CPU usage for this task is back down to 15%.

Mike

TomaszPawel
Send message
Joined: 18 Aug 08
Posts: 121
Credit: 59,836,411
RAC: 0
Level
Thr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 9374 - Posted: 6 May 2009 | 12:13:30 UTC - in response to Message 8911.
Last modified: 6 May 2009 | 12:15:17 UTC

Hi!

I use 32 bit OS and crunch gpugrid :)

Do you have also 64 bit version of your aplication?

I mean real 64 bit, no 32bit runing on 64 bit os.

If you have 64 bit is it faster then 32?

I don't know if it is time to upgrade to 64 bit os.

My comp have 4 gb ram, on XP 32 bit I have 3400Mb witch is ok ...
____________
POLISH NATIONAL TEAM - Join! Crunch! Win!

Profile Paul D. Buck
Send message
Joined: 9 Jun 08
Posts: 1050
Credit: 37,321,185
RAC: 0
Level
Val
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 9399 - Posted: 6 May 2009 | 20:56:15 UTC - in response to Message 9374.

Hi!

I use 32 bit OS and crunch gpugrid :)

Do you have also 64 bit version of your aplication?

I mean real 64 bit, no 32bit runing on 64 bit os.

If you have 64 bit is it faster then 32?

I don't know if it is time to upgrade to 64 bit os.

My comp have 4 gb ram, on XP 32 bit I have 3400Mb witch is ok ...

Since there is little time spent on the CPU, changing the application to 64-bit won't make much difference.

Projects where that can help include Prime Grid, ABC and at least one other and my mind whet blank. Oh MW with the optimized applications.

Post to thread

Message boards : Number crunching : New applications

//