Welcome to MilkyWay@home

de_14_3s_free_5... shows no progress on 0.20 CPU client

Message boards : Number crunching : de_14_3s_free_5... shows no progress on 0.20 CPU client
Message board moderation

To post messages, you must log in.

1 · 2 · 3 · Next

AuthorMessage
[boinc.at] Nowi

Send message
Joined: 22 Mar 09
Posts: 99
Credit: 503,422,495
RAC: 0
Message 37262 - Posted: 12 Mar 2010, 20:58:05 UTC
Last modified: 12 Mar 2010, 20:59:23 UTC

Hello,

I have downloaded this WU (http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=78693225) and it shows no progress after more than 30 minutes (0.000 %).

What happened?
What shall I do?

Boinc 6.10.36
Win7 64Bit
0.20 sse3 optimized app

thanks for your replies
ID: 37262 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile arkayn
Avatar

Send message
Joined: 14 Feb 09
Posts: 999
Credit: 74,932,619
RAC: 0
Message 37263 - Posted: 12 Mar 2010, 21:07:57 UTC - in response to Message 37262.  
Last modified: 12 Mar 2010, 21:09:31 UTC

I ended up aborting the 20 tasks as they all did the same thing, right now I am crunching Collatz as they go fine.

This is either another bad search, or something the .20b app cannot handle.

[edit]
Seeing that you are using a CPU app and I am running a GPU app, I am going to say that it is a bad search.
ID: 37263 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Donnie
Avatar

Send message
Joined: 19 Jul 08
Posts: 67
Credit: 272,086,462
RAC: 0
Message 37264 - Posted: 12 Mar 2010, 21:14:00 UTC - in response to Message 37263.  
Last modified: 12 Mar 2010, 21:14:56 UTC

I ended up aborting the 20 tasks as they all did the same thing, right now I am crunching Collatz as they go fine.

This is either another bad search, or something the .20b app cannot handle.

[edit]
Seeing that you are using a CPU app and I am running a GPU app, I am going to say that it is a bad search.


I was aborting a few of them until the progress bar finally moved (above 100% and counted backwards). When they finally reported, they took almost twice as long as the "normal" WUs, and paid a liitle better. I just let 'em run!

BTW, GPUs only with no app_info file. I just let the stock app run!!
ID: 37264 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile arkayn
Avatar

Send message
Joined: 14 Feb 09
Posts: 999
Credit: 74,932,619
RAC: 0
Message 37265 - Posted: 12 Mar 2010, 21:19:39 UTC

I had one go into 4 minutes and was still at 0.000%, it was still using 99% of my GPU to do nothing.

I will not crunch them.
ID: 37265 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile [KWSN]John Galt 007
Avatar

Send message
Joined: 12 Dec 08
Posts: 56
Credit: 269,889,439
RAC: 0
Message 37266 - Posted: 12 Mar 2010, 21:22:02 UTC

ID: 37266 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Crunch3r
Volunteer developer
Avatar

Send message
Joined: 17 Feb 08
Posts: 363
Credit: 258,227,990
RAC: 0
Message 37267 - Posted: 12 Mar 2010, 21:34:41 UTC - in response to Message 37266.  
Last modified: 12 Mar 2010, 21:35:06 UTC

http://milkyway.cs.rpi.edu/milkyway/workunit.php?wuid=73633688

They pay well...417 credits...


Not really. I had one on a 4870 that took 830 sec/417 credits... compare that to 177sec/213 credits for the normal ones (de_s222).

Something is wrong here.

Join Support science! Joinc Team BOINC United now!
ID: 37267 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Len LE/GE

Send message
Joined: 8 Feb 08
Posts: 261
Credit: 104,050,322
RAC: 0
Message 37268 - Posted: 12 Mar 2010, 21:39:06 UTC

The old searches did run like 250s here and gave 213.76 credits

I see the new searches run from 461.05s to 663.27s for 417.48 credits

While the 461s for 417 credits is in range, 663s for the same credits is far off and should give around 565.

The progress bar first stayed at 0.0% and was above 130% when finished.
ID: 37268 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile [KWSN]John Galt 007
Avatar

Send message
Joined: 12 Dec 08
Posts: 56
Credit: 269,889,439
RAC: 0
Message 37269 - Posted: 12 Mar 2010, 21:44:22 UTC - in response to Message 37267.  

Hmmm...mine was a de_13_3s_free_5_22297_1268428455 and it ran on a stock 4870, i7-920, 3 GPU, Win x64 system...I did have 1 that ran 380sec/417cr, but that was a de_13_3s_const_1_22237_1268428453...
Click to help Seti City.




ID: 37269 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
John Clark

Send message
Joined: 4 Oct 08
Posts: 1734
Credit: 64,228,409
RAC: 0
Message 37281 - Posted: 12 Mar 2010, 23:24:21 UTC

Similar discussion going on here
Go away, I was asleep


ID: 37281 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile skildude
Avatar

Send message
Joined: 13 Jan 08
Posts: 52
Credit: 59,823,507
RAC: 0
Message 37288 - Posted: 13 Mar 2010, 0:55:26 UTC - in response to Message 37281.  

I only noticed the WU's hanging when I suspended the GPU work and it came back with >100% while it was still running. I restarted BOINC. everything was fine until I snoozed the GPU/resumed and I had to abort the WU I was on. these new WU's make my screen get a lot more wonky than when I was running the other WU's
Blah blah blah you know the rest
ID: 37288 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Joses
Avatar

Send message
Joined: 8 Jul 09
Posts: 19
Credit: 1,667,175
RAC: 0
Message 37295 - Posted: 13 Mar 2010, 6:11:40 UTC

Looks bizarre!
I occasionally crunch a cosmology unit but those things are massive with few save-points, so I got to stop everything else so that those things crunch in one pass.

Turned Milkyway back on again now that I got past another large Cosmology unit, and wow... 1 minute each and results as "Completed, marked as invalid".

Seems each project has it's different requirements.
Seti has long-range requirement that can wait until mid april - therefore no rush.
Einstien appears also friendly to smaller/slower desktops too.
Milkyway needs results rather fast, so keeps bumping the slower projects that can wait longer.
Cosmology is massive with few save-points, so got to run them in one shot.

I notice there is something mentioned about scripts refusing WUs and I hope that is about some other subject, and not simply me telling BOINC to please not pick-up the next WU since I'd like to crunch a share of other projects....

Seems like whatever that was, was just a 15 minute blip and not a normal thing (now that I see a somewhat normal unit again --- Dohhh!!! it's another high priority WU again).
http://www.joescat.com/boinc/
ID: 37295 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
BarryAZ

Send message
Joined: 1 Sep 08
Posts: 520
Credit: 302,524,931
RAC: 15
Message 37298 - Posted: 13 Mar 2010, 7:11:25 UTC - in response to Message 37288.  

I have seen this as well on multiple workstations -- the de_s222's are good, the de_12, de_13 and de_14 are bad.
ID: 37298 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
_heinz

Send message
Joined: 23 Feb 09
Posts: 28
Credit: 10,775,220
RAC: 0
Message 37301 - Posted: 13 Mar 2010, 8:27:46 UTC

The same for me... resetted project...
nothing helped...
ID: 37301 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
John Clark

Send message
Joined: 4 Oct 08
Posts: 1734
Credit: 64,228,409
RAC: 0
Message 37303 - Posted: 13 Mar 2010, 8:51:29 UTC

Nothing to do with you, it's what the servers are giving us.

Wait until they go away, but crunch normally.
Go away, I was asleep


ID: 37303 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Joe

Send message
Joined: 1 Sep 08
Posts: 1
Credit: 39,265,086
RAC: 0
Message 37306 - Posted: 13 Mar 2010, 9:36:37 UTC

I have a 5970 GPU and no problems with the _13 WUs. After 150sec 0,00% it jumps to 143% and go down to 100%. Afer 195 sec I earn about 417 credits...

ID: 37306 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
[boinc.at] Nowi

Send message
Joined: 22 Mar 09
Posts: 99
Credit: 503,422,495
RAC: 0
Message 37307 - Posted: 13 Mar 2010, 10:13:45 UTC

So, reading all that, I allowed one WU to complete and I observed the things mentioned above:

Long time progress at 0.00 %, then jumping over 100 % and counting backwards. WU is finished when progress descends to 100 %.

outcome: (optimized 0.20 sse3 64Bit on overclocked Q9550)
2.74 credits/minute (de_14_3s...) new one
1.86 credits/minute (de_s222...) old one

increase of about 47 %.

That´s fine. So I will let them run in the future and will not wonder about the strange progress bar... ;-)
ID: 37307 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
John Clark

Send message
Joined: 4 Oct 08
Posts: 1734
Credit: 64,228,409
RAC: 0
Message 37312 - Posted: 13 Mar 2010, 14:39:33 UTC

That is the conclusion I have come to as well. It is just the unexpected behaviour of the progress bar that makes us suspect the validity of the WUs in question.
Go away, I was asleep


ID: 37312 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Paul Forsdick

Send message
Joined: 19 Feb 09
Posts: 29
Credit: 5,452,255
RAC: 0
Message 37317 - Posted: 13 Mar 2010, 18:48:28 UTC

Hi
I have got Boinctasks and it is showing 5 hours 10 minutes 49.381% 5 hrs 19 to go

Boinc shows 5 hours 10 minutes 0% 5 hrs 19 to go

it looks like a boinc problem and the work is running properly according to Boinctasks

paul
ID: 37317 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Purple Rabbit
Avatar

Send message
Joined: 9 Nov 08
Posts: 44
Credit: 128,043,914
RAC: 0
Message 37318 - Posted: 13 Mar 2010, 18:59:31 UTC
Last modified: 13 Mar 2010, 19:37:36 UTC

I've had several de_14_*** tasks "successfully" complete (according to the BOINC Client) after 2 seconds. These are CPU tasks on multiple computers. The tasks were downloaded this morning and were first issue (_0 tasks). Needless to say they were invalid.

I've had many more of them over the past day or so. STD error shows nothing (as you might expect), just a really fast finish!
ID: 37318 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Cluster Physik

Send message
Joined: 26 Jul 08
Posts: 627
Credit: 94,940,203
RAC: 0
Message 37320 - Posted: 13 Mar 2010, 23:10:45 UTC - in response to Message 37263.  
Last modified: 13 Mar 2010, 23:27:05 UTC

This is either another bad search, or something the .20b app cannot handle.

I think I have an idea what is causing the strange progress bar behaviour. The project just increased the WU size another time with this new series. Now, one WU integrates over more than 2 billion points in space for the main integral. As this is used to calculate the progress (it's just the ratio between between already completed points and total points) and the code uses only a 32bit integer to store this number it is very likely an overflow of this 32bit values. I still have to check for some side effects (especially for the GPU version as it is also used to predict the sleep times of the CPU during the GPU computations; edit: no, that part uses already doubles), but it should be easy to change this to an 64bit integer or a double value to correct the progress bar behaviour. Furtunately, this small bug shouldn't affect the actual calculation at all ;)

I expect a 0.22 release tomorrow if I find the time to compile all versions.
ID: 37320 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
1 · 2 · 3 · Next

Message boards : Number crunching : de_14_3s_free_5... shows no progress on 0.20 CPU client

©2024 Astroinformatics Group