Error when downloading!! Again (/&%¤#¤%&/(

Message boards : Number crunching : Error when downloading!! Again (/&%¤#¤%&/(
Message board moderation

To post messages, you must log in.

AuthorMessage
DanHansen@Denmark
Volunteer tester
Avatar

Send message
Joined: 14 Nov 12
Posts: 194
Credit: 5,881,465
RAC: 0
Denmark
Message 1661051 - Posted: 3 Apr 2015, 12:20:38 UTC
Last modified: 3 Apr 2015, 12:37:00 UTC

Hi,


Anybody who can help me with these issues??? Why is it these GPU tasks fails!?!? I know it's Asteroids jobs, but SETI AstroPulse v. 7.04 (Linux/GPU Jobs) doesn't work either...







No SETI AstroPulse v. 7.04 either ;(



Astropulse is generating work again
The databases are still being repaired, but we are now in a state where we can create new workunits and collect results. 13 Mar 2015, 22:44:48 UTC



Hoping to hear from you ;)

Happy holidays ;)

Kind Regards,
Dan
Project Headless CLI Linux Multiple GPU Boinc Servers
Ubuntu Server 14.04.1 64bit
Kernel 3.13.0-32-generic
CPU's i5-4690K
GPU's GT640/GTX750TI
Nvidia v.340.29
BOINC v.7.2.42

ID: 1661051 · Report as offensive
rob smith Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer moderator
Volunteer tester

Send message
Joined: 7 Mar 03
Posts: 22158
Credit: 416,307,556
RAC: 380
United Kingdom
Message 1661068 - Posted: 3 Apr 2015, 13:41:27 UTC

Since most (all?) of the highlighted download failures appear to be connected to Asteroids you would be far better asking the question over there.


The one group highlighted that are S@H's are not an uncommon sight. Its all to do with the way S@H prepares and distributes work. A batch of 100 work units are prepared for distribution, and are sent out to the first valid requests. Once that batch of 100 is sent out another batch is prepared. If your request arrives in the time it takes to prepare a batch there will be no work for you - its a bit of a pain, you have to wait for your next valid request slot to try again (abut 5 minutes?). Even in normal times you may have to wait a couple of cycles to get some work if you are unlucky to make a request when the output cache is empty.
Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?
ID: 1661068 · Report as offensive
Profile HAL9000
Volunteer tester
Avatar

Send message
Joined: 11 Sep 99
Posts: 6534
Credit: 196,805,888
RAC: 57
United States
Message 1661093 - Posted: 3 Apr 2015, 15:15:32 UTC
Last modified: 3 Apr 2015, 15:27:58 UTC

The Astroids download issue is an ongoing issue the admins has been working on. See this thread for details. I get about 2500 errors for ever 20-30 good tasks.

For most projects you can check the Server Status to see the amount of work "Results ready to send". In this case you will notice that Astropulse is 0. Astropulse work is generally always in high demand/short supply.

Also there is not currently a stock MB Nvidia linux app. Which you can check by accessing the Applications list for the project.
SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[
ID: 1661093 · Report as offensive
DanHansen@Denmark
Volunteer tester
Avatar

Send message
Joined: 14 Nov 12
Posts: 194
Credit: 5,881,465
RAC: 0
Denmark
Message 1661158 - Posted: 3 Apr 2015, 19:01:28 UTC
Last modified: 3 Apr 2015, 19:35:56 UTC

Hi Rob & Hal9000 ;)


Thanks for your reply.

Hal9000, OK! I see... This is very useful. Now I can just look there before searching for errors. Thanks!
Maybe this is why there was only 1 job done on one of my systems in a pretty long time:
    21 Mar 2015, 19:49:40 UTC Completed, waiting for validation 2.03 0.33 pending AstroPulse v7 v7.04 (opencl_nvidia_100)


I'll just find a project which uses the GPU's more. Much easier. I just hoped SETI would have done some more of these, but then again, I'm sure everybody needs something ;)

Rob, ".....If your request arrives in the time it takes to prepare a batch there will be no work for you - its a bit of a pain, you have to wait for your next valid request slot to try again (abut 5 minutes?). Even in normal times you may have to wait a couple of cycles to get some work if you are unlucky to make a request when the output cache is empty..."
OK, we'll just have to wait and see what happens. It's just that last year when the I attached to a project with a new system it went so much faster. Not with SETI, because back the "AstroPulse v7 v7.04 (opencl_nvidia_100)" wasn't ready. We'll just have to hope the jobs will be created ;)


Another thing is, that I've been looking for a third project using GPU's so that I had a better chance to gets jobs for my GPU's. And because I don't want all these GPU's to be idle. This wasn't my plan when spending all that money ;)
So, I chose to try Einstein@, but of course I couldn't create an account. They have closed for new accounts. Just attach and use email they write. Well, this is ok if you are using e.g. the BoincManager, but when using the command line tool "BOINCCMD" I'm not sure if this is at all possible!?!?
I usually use a BOINCCMD project attach command and my ID to connect to a project!. Please let me know if you guys have an idea. Sorry for the newbie question :o)

I'm also trying to compare jobs from a Asus GT640, Asus GTX470 & the newest LowProfile card GTX750ti. But, I just can't figure it out. When looking at properties for a workunit from e.g. Asteroid. The only thing I've been able to find is "Estimated task size, in this case 1.380.023 GFLOPs". This job is running! The example in the bottom is a finished job.

    Project Asteroids@home
    Name period_search_6709_1426501029.728701_316143_0

    Application Period Search Application 101.12 (cuda55)
    Workunit name period_search_6709_1426501029.728701_316143
    State Running
    Received 03-04-2015 21:04:42
    Report deadline 14-04-2015 09:04:41
    Estimated app speed 147,83 GFLOPs/sec
    Estimated task size 1.380.023 GFLOPs
    Resources 0,01 CPUs + 1 NVIDIA GPU
    CPU time at last checkpoint 00:00:03
    CPU time 00:00:03
    Elapsed time 00:17:25
    Estimated time remaining 02:15:48
    Fraction done 19,090%
    Virtual memory size 24.774,20 MB
    Working set size 88,14 MB
    Directory slots/4
    Process ID 1459



But when viewing jobs finished, done by another GPU I can't find the same "type" or "name" to measuere the GPU's against each other. Would like to see how much faster a GTX750ti is compared to GT640. Data from a job done by GT640 looks like this:

    Navn period_search_1010_1426497677.081390_420611_1
    Workunit 28683823
    Opret 16 Mar 2015, 21:20:01 UTC
    Afsender 28 Mar 2015, 22:36:09 UTC
    Report deadline 8 Apr 2015, 10:36:09 UTC
    Received 29 Mar 2015, 11:06:10 UTC
    Server state Over
    Outcome Success
    Client state Ingen
    Exit status 0 (0x0)
    Computere 82538
    Run time 2 hours 33 min 37 sec
    CPU time 23 sec
    Validate state Valid
    Opret 480.00
    Device peak FLOPS 176.69 GFLOPS
    Applikationer Period Search Application v101.12 (cuda55)





KR
Dan


Project Headless CLI Linux Multiple GPU Boinc Servers
Ubuntu Server 14.04.1 64bit
Kernel 3.13.0-32-generic
CPU's i5-4690K
GPU's GT640/GTX750TI
Nvidia v.340.29
BOINC v.7.2.42

ID: 1661158 · Report as offensive
Profile HAL9000
Volunteer tester
Avatar

Send message
Joined: 11 Sep 99
Posts: 6534
Credit: 196,805,888
RAC: 57
United States
Message 1661169 - Posted: 3 Apr 2015, 19:59:09 UTC - in response to Message 1661158.  

I was able to create an account with Einstein@Home & 2 other projects using the "lost password" link with my 2nd account. It has only ever been connected to SETI@home, but my username & other account info was carried over. I imagine this is part of the BOINC project syncing process. As the account creation date listed on the other projects displays March 4th. I know for a fact I was not attaching to any projects or any other messing around that day. Given I was having my HVAC replaced.
SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[
ID: 1661169 · Report as offensive
Juha
Volunteer tester

Send message
Joined: 7 Mar 04
Posts: 388
Credit: 1,857,738
RAC: 0
Finland
Message 1661408 - Posted: 4 Apr 2015, 11:26:33 UTC - in response to Message 1661158.  

So, I chose to try Einstein@, but of course I couldn't create an account. They have closed for new accounts. Just attach and use email they write. Well, this is ok if you are using e.g. the BoincManager, but when using the command line tool "BOINCCMD" I'm not sure if this is at all possible!?!?
I usually use a BOINCCMD project attach command and my ID to connect to a project!. Please let me know if you guys have an idea. Sorry for the newbie question :o)


You use "--create_account URL email password name" followed by "--lookup_account URL email password" is necessary.
ID: 1661408 · Report as offensive
DanHansen@Denmark
Volunteer tester
Avatar

Send message
Joined: 14 Nov 12
Posts: 194
Credit: 5,881,465
RAC: 0
Denmark
Message 1661592 - Posted: 4 Apr 2015, 22:48:02 UTC
Last modified: 4 Apr 2015, 23:28:15 UTC

Hi HAL9000 & Juha ;)


Thanks. I'll try both things! Because I've just looked at all my jobs done for SETI and only 1 !! Only 1 was done by a GPU/Linux and that was on 23 Mar 2014, 23:03:40 UTC !! I'm so g.. d... disappointed. I really wanted to support SETI with all these GPU systems. I just found a new Low Profile card which does more than twice the amount of work than the GT640. I just bought the GTX750TI and I'm testing right now. 3 or 4 month ago when I noticed that none Linux/GPU jobs where being done I shut them down.. All of them but one. I don't want an electrical bill that high when I'm doing nothing to help.

What would you do? Wait and see what happens? Or implement those apps from e.g. lunatics and do that way around? http://lunatics.kwsn.net/index.php?module=Downloads Thought about a year ago, but I'm totally green here.

***********************

Found this:

If you're running a command-line or pre-5.0 version of BOINC, create an account first.

Sorry, but web acccount creation is disabled
Just run BOINC, select Attach Project, and enter an email address and password.


***********************

Tried it!!! It worked perfectly! Thanks a lot! For others, here how to do it:

To create a account for Einstein using CLI
Command: # boinccmd --create_account http://einstein.phys.uwm.edu/ myemail@gmail.com mypassword theusernameiwant

Reply from CLI Einsten:
# account key: 722453269c2226a71e7f0e76465764212

Use the ID/account key to attach to the project:
Command: boinccmd --project_attach http://einstein.phys.uwm.edu/ 722453269c2226a71e7f0e76465764212

Remember! If you are using password and port other than the default, add "--host 192.168.x.xx:xxxxx --passwd mypassword" after "boinccmd" but before everything else ;)


Thanks for your help! 10 seconds after I attached Einsten BRP5-cuda32 jobs started to pour in ;)




KR
Dan
Project Headless CLI Linux Multiple GPU Boinc Servers
Ubuntu Server 14.04.1 64bit
Kernel 3.13.0-32-generic
CPU's i5-4690K
GPU's GT640/GTX750TI
Nvidia v.340.29
BOINC v.7.2.42

ID: 1661592 · Report as offensive
Profile BilBg
Volunteer tester
Avatar

Send message
Joined: 27 May 07
Posts: 3720
Credit: 9,385,827
RAC: 0
Bulgaria
Message 1661736 - Posted: 5 Apr 2015, 13:53:25 UTC - in response to Message 1661592.  
Last modified: 5 Apr 2015, 14:01:36 UTC

Because I've just looked at all my jobs done for SETI and only 1 !! Only 1 was done by a GPU/Linux and that was on 23 Mar 2014, 23:03:40 UTC !! I'm so g.. d... disappointed.

You look at the wrong place - task lists are purged (info about validated tasks 'vanish' from the web page ~day after validation)

Application details for host 7247578
http://setiathome.berkeley.edu/host_app_versions.php?hostid=7247578

AstroPulse v7 7.04 x86_64-pc-linux-gnu (opencl_nvidia_100):
"Number of tasks completed 90"

This means at least 90 AstroPulse v7 tasks are done by NVIDIA GPU on Linux
(not all "Completed and validated" tasks are counted on "Number of tasks completed" value)


... Or implement those apps from e.g. lunatics ...

This is where you can read how Linux users successfully run NVIDIA CUDA SETI@home app on Linux (Download link - at the (jason_gee) post linked):
http://setiathome.berkeley.edu/forum_thread.php?id=77018&postid=1660631#1660631

Ask there if you need info on how to 'install' preserving CPU tasks
 


- ALF - "Find out what you don't do well ..... then don't do it!" :)
 
ID: 1661736 · Report as offensive
DanHansen@Denmark
Volunteer tester
Avatar

Send message
Joined: 14 Nov 12
Posts: 194
Credit: 5,881,465
RAC: 0
Denmark
Message 1662198 - Posted: 7 Apr 2015, 5:18:14 UTC
Last modified: 7 Apr 2015, 5:20:28 UTC

Hi BilBg,


Thanks for the heads up ;) Thanks for the links. As you remember I'm a clown when it comes to building XML files ;)

Have a nice easter ;)

KR
Dan
Project Headless CLI Linux Multiple GPU Boinc Servers
Ubuntu Server 14.04.1 64bit
Kernel 3.13.0-32-generic
CPU's i5-4690K
GPU's GT640/GTX750TI
Nvidia v.340.29
BOINC v.7.2.42

ID: 1662198 · Report as offensive

Message boards : Number crunching : Error when downloading!! Again (/&%¤#¤%&/(


 
©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.