GPU app - beta version for linux nvidia

Message boards : News : GPU app - beta version for linux nvidia
Message board moderation

To post messages, you must log in.

Previous · 1 · 2 · 3 · 4 · 5 · 6 · Next

AuthorMessage
Profile Eric Driver
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 8 Jul 11
Posts: 1163
Credit: 192,059,671
RAC: 179,006
Message 2271 - Posted: 25 Mar 2019, 20:20:02 UTC - in response to Message 2269.  

Yesterday was triple the output of the Formula BOINC Sprint's best day.


Unfortunately, I don't think those numbers are accurate, just based on the numbers of results coming back. It's correlated to the obscenely high GPU credits being awarded. Maybe it's time I change to "Credit New", from the current "Credit based on runtime"?


I do believe you need to change the credit system. On my GPUs it takes running about 3 WUs to fully load the GPU. If I run 6 WUs per GPU I get twice the credit, but the amount of work done is probably about the same. I only ran 6 WUs per GPU on 1 GPU for a couple of hours. I started to run other ones for a few minutes, but then backed off to 3 since it doesn't really seem fair.

Thanks for the great job. Things do run really well.


OK. Some will not be happy about this, but after enough people suggested it, I restarted the validator using the default credit system. So I believe this will be the infamous "CreditNew".
ID: 2271 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
tito

Send message
Joined: 18 Aug 13
Posts: 2
Credit: 1,003,792
RAC: 0
Message 2272 - Posted: 25 Mar 2019, 20:25:04 UTC - in response to Message 2270.  

LOL
I think none. I'm windows user and there wasn't need to install additional CUDA.
Unfortunately I have limited internet access on linux box so will wait for Windows version. Then try 1080Ti.
Wating...
ID: 2272 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
mmonnin

Send message
Joined: 1 Feb 17
Posts: 23
Credit: 39,568,985
RAC: 679
Message 2274 - Posted: 25 Mar 2019, 21:21:42 UTC - in response to Message 2268.  

Only errors here.
GTX645; newest drivers 418
Anyone can help?

PG PPS sieve works fine.


2nd person for a 6xx card to not work with 418. The card might just be too old.
ID: 2274 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
[H]auntjemima

Send message
Joined: 30 Apr 18
Posts: 11
Credit: 2,344,276
RAC: 0
Message 2275 - Posted: 26 Mar 2019, 0:07:08 UTC

lol, at least you guys are getting tasks. I just keep getting "got 0 new tasks".

I disabled cpu tasks, just to test the GPU versions. I am running driver 418.56 AND I have a compute 3.0 card. Are there any tasks or am I configured incorrectly?
ID: 2275 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
mmonnin

Send message
Joined: 1 Feb 17
Posts: 23
Credit: 39,568,985
RAC: 679
Message 2277 - Posted: 26 Mar 2019, 0:42:02 UTC - in response to Message 2275.  

lol, at least you guys are getting tasks. I just keep getting "got 0 new tasks".

I disabled cpu tasks, just to test the GPU versions. I am running driver 418.56 AND I have a compute 3.0 card. Are there any tasks or am I configured incorrectly?


It seems like the 760 should be able to receive tasks. The app is Linux only so the 210 won't work. And it's to old.
ID: 2277 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
[H]auntjemima

Send message
Joined: 30 Apr 18
Posts: 11
Credit: 2,344,276
RAC: 0
Message 2278 - Posted: 26 Mar 2019, 0:44:33 UTC - in response to Message 2277.  

lol, at least you guys are getting tasks. I just keep getting "got 0 new tasks".

I disabled cpu tasks, just to test the GPU versions. I am running driver 418.56 AND I have a compute 3.0 card. Are there any tasks or am I configured incorrectly?


It seems like the 760 should be able to receive tasks. The app is Linux only so the 210 won't work. And it's to old.


Yeah, the 760 is the only one I'm using for it. The 210 does nothing. And in fact has never worked for BOINC ever lol
ID: 2278 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Eric Driver
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 8 Jul 11
Posts: 1163
Credit: 192,059,671
RAC: 179,006
Message 2279 - Posted: 26 Mar 2019, 1:10:12 UTC - in response to Message 2275.  

lol, at least you guys are getting tasks. I just keep getting "got 0 new tasks".

I disabled cpu tasks, just to test the GPU versions. I am running driver 418.56 AND I have a compute 3.0 card. Are there any tasks or am I configured incorrectly?


There should be plenty of tasks available. And I see no reason why the scheduler would not give you work.

Is the 3.01 executable in your project directory? I had the same problem last night, where I had the old 3.00 executable and it wouldn't download the new exe or send new tasks...but it eventually started working.
ID: 2279 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
[H]auntjemima

Send message
Joined: 30 Apr 18
Posts: 11
Credit: 2,344,276
RAC: 0
Message 2280 - Posted: 26 Mar 2019, 1:18:02 UTC - in response to Message 2279.  
Last modified: 26 Mar 2019, 1:20:38 UTC

lol, at least you guys are getting tasks. I just keep getting "got 0 new tasks".

I disabled cpu tasks, just to test the GPU versions. I am running driver 418.56 AND I have a compute 3.0 card. Are there any tasks or am I configured incorrectly?


There should be plenty of tasks available. And I see no reason why the scheduler would not give you work.

Is the 3.01 executable in your project directory? I had the same problem last night, where I had the old 3.00 executable and it wouldn't download the new exe or send new tasks...but it eventually started working.


It isnt downloading any master file at all. I reset the project and even tried removing and then adding it back. Folder remains empty and tells me the same issue as above. 0 tasks sent.

edit: Let me clarify... initially there was a 2.12 file in there, and I removed it and then reset the project. Since then, no new file has appeared.

edit2: waited the 3 minute requirement, clicked update and got 100 tasks. Thanks!
ID: 2280 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
[H]auntjemima

Send message
Joined: 30 Apr 18
Posts: 11
Credit: 2,344,276
RAC: 0
Message 2281 - Posted: 26 Mar 2019, 1:23:43 UTC

100 tasks all computation error, besides the ones I was able to abort before they did. All compute error immediately.. output is below.

<core_client_version>7.6.31</core_client_version>
<![CDATA[
<message>
process exited with code 1 (0x1, -255)
</message>
<stderr_txt>
Setting GPU device number 0.
CHECKPOINT_FILE = wu_sf5_DS-15x11_Grp91994of800000_checkpoint.
Checkpoint Flag = 0.
Reading file ../../projects/numberfields.asu.edu_NumberFields/sf5_DS-15x11_Grp91994of800000.dat
K = x^2 + 5
S = [2, 5]
Disc Bound = 320000000000000
Skip = (P^4)*(Q^5)
Num Congruences = 5
SCALE = 1.000000
|dK| = 20
Signature = [0,1]
Opening output file ../../projects/numberfields.asu.edu_NumberFields/wu_sf5_DS-15x11_Grp91994of800000_0_r891750085_0
Now starting the targeted Martinet search:
Error code 700: an illegal memory access was encountered file polDiscTest_gpuCuda.cu line 415.
polDisc Test had an error. Aborting.

</stderr_txt>
]]>
ID: 2281 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Eric Driver
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 8 Jul 11
Posts: 1163
Credit: 192,059,671
RAC: 179,006
Message 2282 - Posted: 26 Mar 2019, 3:44:08 UTC - in response to Message 2281.  

So it would appear 6xx cards have a problem with the gpu app. Most everyone with issues has that in common.
ID: 2282 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Eric Driver
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 8 Jul 11
Posts: 1163
Credit: 192,059,671
RAC: 179,006
Message 2283 - Posted: 26 Mar 2019, 4:04:04 UTC - in response to Message 2282.  

I added the min_cuda_version to the plan_class, to see if that's the root cause of the problem with the 6xx cards (if it is, I would expect them to stop receiving tasks)

On the flip side, let me know if properly functioning hosts stop receiving tasks, and I can lower the min version.
ID: 2283 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
gemini8

Send message
Joined: 6 May 16
Posts: 6
Credit: 1,887,309
RAC: 1,057
Message 2284 - Posted: 26 Mar 2019, 5:59:10 UTC

If enough people have older drivers I might just have to build from an older SDK. I'll wait to see what the concensus is.

Please build for older drivers also. I'm sure there are many, like myself, with older cards that would love to help but the 418.xx drivers aren't supported for our cards.

And there are others like myself who don't want to unnecessarily fiddle with their systems as they're running just fine.
I downgraded the Nvidia drivers on my Win box after the initial performance breakdown of the 4xx.xx drivers, and the older drivers are working nicely with my 970 and 1050Ti cards. Meanwhile both 1050Ti are sitting in my Debian system, running on pre 4xx.xx drivers as well.

Btw.: Which hardware requirements does the GPU app have?
Greetings, Jens
ID: 2284 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Chooka

Send message
Joined: 3 May 18
Posts: 15
Credit: 17,304,326
RAC: 11
Message 2286 - Posted: 26 Mar 2019, 8:45:19 UTC - in response to Message 2284.  
Last modified: 26 Mar 2019, 8:50:21 UTC

Soooo is there a time frame before I can throw my Radeon cards at this?
It's a bit of a shame to see my hours and hours of CPU crunching now being smashed by NVIDIA GPU work :(

Oh...and you're going to need to rework your badge system with all these credits. 100Mil is nothing now.

ID: 2286 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
[H]auntjemima

Send message
Joined: 30 Apr 18
Posts: 11
Credit: 2,344,276
RAC: 0
Message 2288 - Posted: 26 Mar 2019, 11:07:54 UTC - in response to Message 2282.  

So it would appear 6xx cards have a problem with the gpu app. Most everyone with issues has that in common.


I am running a 7xx card.
ID: 2288 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Alez

Send message
Joined: 31 Oct 12
Posts: 14
Credit: 30,251,746
RAC: 6,405
Message 2289 - Posted: 26 Mar 2019, 13:00:39 UTC - in response to Message 2271.  



Unfortunately, I don't think those numbers are accurate, just based on the numbers of results coming back. It's correlated to the obscenely high GPU credits being awarded. Maybe it's time I change to "Credit New", from the current "Credit based on runtime"?



OK. Some will not be happy about this, but after enough people suggested it, I restarted the validator using the default credit system. So I believe this will be the infamous "CreditNew".


And now the credit awarded is so low it's not worth putting my GPU's on the project.
ID: 2289 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Wailing Angus Beef

Send message
Joined: 15 Jan 15
Posts: 4
Credit: 26,754,561
RAC: 15
Message 2290 - Posted: 26 Mar 2019, 13:25:42 UTC

It was paying around 20 per run sec on my GTX 1080 which is better than many GPU apps but not crazy. Now it's only paying 0.08 per run sec, 1/250 of what it was which is crazy low. It's worse paying than Asteroids on a GPU. Einstein pays 5-7 per run sec.
ID: 2290 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
pututu

Send message
Joined: 8 Jul 17
Posts: 10
Credit: 30,887,005
RAC: 0
Message 2291 - Posted: 26 Mar 2019, 13:33:36 UTC - in response to Message 2290.  
Last modified: 26 Mar 2019, 14:04:49 UTC

It was paying around 20 per run sec on my GTX 1080 which is better than many GPU apps but not crazy. Now it's only paying 0.08 per run sec, 1/250 of what it was which is crazy low. It's worse paying than Asteroids on a GPU. Einstein pays 5-7 per run sec.


Maybe I'll try CPU and see if the points are better than GPU payout. :(

edit: If there is no significant point (not number of WUs completion) gain over CPU app, it certainly defeats the purpose of creating GPU app in the first place. I know some of us crunchers are crazy about points rather than the number of WUs completion. Like Asteroid, maybe pay a fix credits for each WU completion which is what is important to NumberField project, rather than the point. In general, the points are to attract crunchers. My one cent.
ID: 2291 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Eric Driver
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 8 Jul 11
Posts: 1163
Credit: 192,059,671
RAC: 179,006
Message 2293 - Posted: 26 Mar 2019, 15:32:19 UTC - in response to Message 2290.  

It was paying around 20 per run sec on my GTX 1080 which is better than many GPU apps but not crazy. Now it's only paying 0.08 per run sec, 1/250 of what it was which is crazy low. It's worse paying than Asteroids on a GPU. Einstein pays 5-7 per run sec.


OK. I changed credit to a fixed credit based on average runtime. Hopefully this will be a good compromise. I tried to target what Einstein pays (actually a tad more). I can adjust later as necessary. Now we wait and see...
ID: 2293 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
mmonnin

Send message
Joined: 1 Feb 17
Posts: 23
Credit: 39,568,985
RAC: 679
Message 2294 - Posted: 26 Mar 2019, 15:34:22 UTC - in response to Message 2291.  

It was paying around 20 per run sec on my GTX 1080 which is better than many GPU apps but not crazy. Now it's only paying 0.08 per run sec, 1/250 of what it was which is crazy low. It's worse paying than Asteroids on a GPU. Einstein pays 5-7 per run sec.


Maybe I'll try CPU and see if the points are better than GPU payout. :(

edit: If there is no significant point (not number of WUs completion) gain over CPU app, it certainly defeats the purpose of creating GPU app in the first place. I know some of us crunchers are crazy about points rather than the number of WUs completion. Like Asteroid, maybe pay a fix credits for each WU completion which is what is important to NumberField project, rather than the point. In general, the points are to attract crunchers. My one cent.


Tasks vary in length so fixed credit is unfair as well. The only other way is if the FLOPs calculated during the task was known for each task. The prior method was open to abuse since it was a function of the GPUs FLOPs and time.
ID: 2294 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Eric Driver
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 8 Jul 11
Posts: 1163
Credit: 192,059,671
RAC: 179,006
Message 2295 - Posted: 26 Mar 2019, 16:46:15 UTC - in response to Message 2294.  


Tasks vary in length so fixed credit is unfair as well. The only other way is if the FLOPs calculated during the task was known for each task. The prior method was open to abuse since it was a function of the GPUs FLOPs and time.


It may be unfair, but I would argue it's more fair than the previous methods. Credit from runtime is open to abuse and CreditNew pays next to nothing. I have computed the average FLOPS over 200 WUs for the current dataset and based the credit on that. In the long run this should be fair and accurate based on the law of large numbers (isn't this the principle behind CredtNew anyways?). The only abuse could be cherry picking, but with GPU tasks only taking a few minutes, someone would have to be insane to sit there and try to distinguish the fast from the slow.
ID: 2295 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Previous · 1 · 2 · 3 · 4 · 5 · 6 · Next

Message boards : News : GPU app - beta version for linux nvidia


Main page · Your account · Message boards


Copyright © 2021 Arizona State University