Computation resets for AstroPulse (opencl_nvidia_100)

Message boards : Number crunching : Computation resets for AstroPulse (opencl_nvidia_100)
Message board moderation

To post messages, you must log in.

AuthorMessage
Anders

Send message
Joined: 5 Aug 11
Posts: 11
Credit: 831,583
RAC: 0
Netherlands
Message 1393214 - Posted: 21 Jul 2013, 9:54:25 UTC
Last modified: 21 Jul 2013, 9:56:22 UTC

I have set GPU to be suspended while computer in use. When the computer idles and SETI@home start to crunch WU of type AstroPulse version 6.01 opencl_nvidia_100 it resets the elapsed computation time to zero and sets back remaining time to the original value, i.e. it start all over again from the start and continues to do this indefinitely.


At start up it says in the log entry:

21/07/2013 11:19:43 | SETI@home | Restarting task ap_14no08ae_B1_P1_00065_20130715_28579.wu_0 using astropulse_v6 version 601 in slot 1

But when resuming it says:

21/07/2013 11:20:05 | | Suspending GPU computation - computer is in use
21/07/2013 11:20:08 | | Resuming GPU computation
21/07/2013 11:20:08 | SETI@home | Restarting task ap_29au08ab_B1_P1_00355_20130713_04765.wu_0 using astropulse_v6 version 604 (opencl_nvidia_100) in slot 7

Which is a little bit strange as the WU is said to be version 6.01 in the Tasks view of the BOINC manager.

Anyway, AFAIK I do not have the same problem with other GPU WU and for now the best I can do is just abort this kind of WU.

Any comments on this is welcomed....
ID: 1393214 · Report as offensive
Profile Raistmer
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 16 Jun 01
Posts: 6325
Credit: 106,370,077
RAC: 121
Russia
Message 1393224 - Posted: 21 Jul 2013, 10:34:52 UTC - in response to Message 1393214.  

Log quotes you listed are for different astropulse tasks.
Look at task names.

Do you have any errored tasks already to force manual aborts?

SETI apps news
We're not gonna fight them. We're gonna transcend them.
ID: 1393224 · Report as offensive
Profile Mike Special Project $75 donor
Volunteer tester
Avatar

Send message
Joined: 17 Feb 01
Posts: 34253
Credit: 79,922,639
RAC: 80
Germany
Message 1393237 - Posted: 21 Jul 2013, 11:41:52 UTC

You only have finnished CPU tasks atm which are 6.01.
GPU ones are 6.04.

Check preferences in Boinc manager and check use GPU always to see if this helps.



With each crime and every kindness we birth our future.
ID: 1393237 · Report as offensive
Anders

Send message
Joined: 5 Aug 11
Posts: 11
Credit: 831,583
RAC: 0
Netherlands
Message 1393492 - Posted: 22 Jul 2013, 6:14:44 UTC - in response to Message 1393224.  

Log quotes you listed are for different astropulse tasks.
Look at task names.


Indeed you are right, didn't looked to careful there.

Do you have any errored tasks already to force manual aborts?


Not sure what you ask, but I aborted two and have one suspended atm. No point in processing them as they will not complete.
ID: 1393492 · Report as offensive
Anders

Send message
Joined: 5 Aug 11
Posts: 11
Credit: 831,583
RAC: 0
Netherlands
Message 1393496 - Posted: 22 Jul 2013, 6:20:47 UTC - in response to Message 1393237.  

You only have finnished CPU tasks atm which are 6.01.
GPU ones are 6.04.


Correct, my bad.

Check preferences in Boinc manager and check use GPU always to see if this helps.



Not sure why I would like to do that... GPU crunching works fine for other WU (cuda 22, 23, 32 etc) so it just the AstroPulse that seams to be "broken".
ID: 1393496 · Report as offensive
Anders

Send message
Joined: 5 Aug 11
Posts: 11
Credit: 831,583
RAC: 0
Netherlands
Message 1395109 - Posted: 26 Jul 2013, 10:54:09 UTC
Last modified: 26 Jul 2013, 10:55:35 UTC

More problems with GPU crunching:

While, for instance, a WU like 17fe09aa.30537.5798.13.12.155_1 (SETI@home cuda42) works just fine I now start to see the same reset problem with other SETI@home GPU WU as well. As a notice known affected WU are:

17fe09aa.30537.5798.13.12.88_1 (cuda42)
18ap08ah.12735.22976.9.12.35_1 (cuda32)

Any idea/suggestions why this happens?
ID: 1395109 · Report as offensive
Profile Vicki
Avatar

Send message
Joined: 30 Nov 01
Posts: 65
Credit: 1,640,576
RAC: 46
New Zealand
Message 1395759 - Posted: 28 Jul 2013, 2:52:45 UTC - in response to Message 1395109.  

More problems with GPU crunching:

While, for instance, a WU like 17fe09aa.30537.5798.13.12.155_1 (SETI@home cuda42) works just fine I now start to see the same reset problem with other SETI@home GPU WU as well. As a notice known affected WU are:

17fe09aa.30537.5798.13.12.88_1 (cuda42)
18ap08ah.12735.22976.9.12.35_1 (cuda32)

Any idea/suggestions why this happens?


How many gpu tasks are you running at once? If more than 1, drop it back to 1 as your gpu only has 512k of ram to work with. Also, since your dektop has 4 cpus, allow seti / Bonic to only use 3 of them.

Other thoughts, Try closing bonic, exit all tasks & run the Luntatics installer v0.41. you can download that from here
http://mikesworldnet.de/produkte.html Be sure to grab the 64 bit installer for your system as it has 64 bit windows on it.

After restarting Bonic, your tasks should continue where they left off from. Note, future work units will show on the seti site as anoynmous platform.
I suggest, when installing that you select either cuda ver 3.2 or 4.2.
A city destroyed by an earthquake is an opportunity to Rebuild, redeign & make it a better place to be. Better, stronger, faster like the 6 Million Dollar Man
ID: 1395759 · Report as offensive
rcthardcore

Send message
Joined: 23 Nov 08
Posts: 48
Credit: 1,306,006
RAC: 0
United States
Message 1396058 - Posted: 29 Jul 2013, 4:53:31 UTC

I would love to get some AP GPU workunits to crunch. My GTX 580 is ready!
ID: 1396058 · Report as offensive
Profile petri33
Volunteer tester

Send message
Joined: 6 Jun 02
Posts: 1668
Credit: 623,086,772
RAC: 156
Finland
Message 1397848 - Posted: 2 Aug 2013, 7:54:08 UTC - in response to Message 1393214.  

A long time ago I had similar problems. I had to check 'keep suspended tasks in memory'.
To overcome Heisenbergs:
"You can't always get what you want / but if you try sometimes you just might find / you get what you need." -- Rolling Stones
ID: 1397848 · Report as offensive
Anders

Send message
Joined: 5 Aug 11
Posts: 11
Credit: 831,583
RAC: 0
Netherlands
Message 1398615 - Posted: 4 Aug 2013, 11:50:08 UTC - in response to Message 1397848.  

A long time ago I had similar problems. I had to check 'keep suspended tasks in memory'.


In my case it cannot be related to that because I have already ticked that option even before the problem start to occur. However I have not noticed any problem with GPU WU since after the last report, in particular I can see this in my statistics because the average credits per day has increased by approximately 35%.
ID: 1398615 · Report as offensive
Anders

Send message
Joined: 5 Aug 11
Posts: 11
Credit: 831,583
RAC: 0
Netherlands
Message 1398617 - Posted: 4 Aug 2013, 11:58:24 UTC - in response to Message 1395759.  
Last modified: 4 Aug 2013, 11:59:35 UTC

How many gpu tasks are you running at once?


Only one at the time, but the scheduler swaps between several GPU tasks.

Also, since your dektop has 4 cpus, allow seti / Bonic to only use 3 of them.


It is not clear to me why I would like, or need, to do this?

Other thoughts, Try closing bonic, exit all tasks & run the Luntatics installer v0.41. you can download that from ...


All I do is just crunching WU with idle time so I do not have any more special needs than what the BONIC Manager already provides me with so I think I will stay with it. But thanks anyway for the advice.
ID: 1398617 · Report as offensive
Profile Mike Special Project $75 donor
Volunteer tester
Avatar

Send message
Joined: 17 Feb 01
Posts: 34253
Credit: 79,922,639
RAC: 80
Germany
Message 1398619 - Posted: 4 Aug 2013, 12:10:48 UTC - in response to Message 1398617.  
Last modified: 4 Aug 2013, 12:12:24 UTC

How many gpu tasks are you running at once?


Only one at the time, but the scheduler swaps between several GPU tasks.

Also, since your dektop has 4 cpus, allow seti / Bonic to only use 3 of them.


It is not clear to me why I would like, or need, to do this?

Other thoughts, Try closing bonic, exit all tasks & run the Luntatics installer v0.41. you can download that from ...


All I do is just crunching WU with idle time so I do not have any more special needs than what the BONIC Manager already provides me with so I think I will stay with it. But thanks anyway for the advice.


The GPU needs to get feed so a free CPU core really helps to avoid trouble.
Plus your GPU will run smoother and faster.

As i can see your CPU needs approx 16000 to 20000 seconds for one unit.
Running the Lunatics installer will give you an optimzed CPU application which reduces your processing time significantly.
So your PC would run more efficient.


With each crime and every kindness we birth our future.
ID: 1398619 · Report as offensive
Claggy
Volunteer tester

Send message
Joined: 5 Jul 99
Posts: 4654
Credit: 47,537,079
RAC: 4
United Kingdom
Message 1398641 - Posted: 4 Aug 2013, 14:53:01 UTC - in response to Message 1398615.  

A long time ago I had similar problems. I had to check 'keep suspended tasks in memory'.


In my case it cannot be related to that because I have already ticked that option even before the problem start to occur.

Another reason why that will have no effect is because GPU tasks are never kept in GPU memory when suspended (the only exception is during CPU benchmarks), So that Selection only effects CPU tasks.

Claggy
ID: 1398641 · Report as offensive

Message boards : Number crunching : Computation resets for AstroPulse (opencl_nvidia_100)


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.