Joining Team Ukraine to show support?

Page 5 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

mmonnin03

Senior member
Nov 7, 2006
214
213
116
Several hours late with the Sprint announcement. 20UTC is 4PM EST. I'm wondering if its going away along with WUProp.
 

Icecold

Golden Member
Nov 15, 2004
1,090
1,008
146
If the Sprint happens I'll throw some machines on it for Team Ukraine. I suppose it remains to be seen if they switched to just announcing the project at the beginning of the sprint instead of 24 hours ahead of time, or if it's not happening/something is wrong, or like you said @mmonnin03 maybe it's going away entirely with WUProp.
 

Orange Kid

Elite Member
Oct 9, 1999
4,327
2,112
146
The Formula Boinc Sprint is SRBase and we are running it for the Ukraine

PrimeGrid does not start till the 21st and we will run that as Team AnandTech

I guess I am in on the FB Sprint as I was already there.
 

mmonnin03

Senior member
Nov 7, 2006
214
213
116
1st few in for Ukraine from me. More joining after GPUGrid tasks complete. At least one of the longer 540k credit tasks.
95s on 3080Ti, whew.
 

mmonnin03

Senior member
Nov 7, 2006
214
213
116
Both GPU and CPU. GPU app TF is worth more credit.
3080Ti 95s
3070Ti 150s
R Vii 290s
1070 Ti 584s
1070 718s

Also the project only works on GPU 0. If you have multiple GPUs, then both tasks will run on GPU 0. I guess a cc_config.xml could ignore the 1st GPU on a 2nd client although I've never tried that.
Edit: if you have Linux that supports CUDA 11 like 20.04 but a GPU w/o CUDA 11 support then you'll get both CUDA 10/11 tasks and the 11 tasks will fail. There is no server logic for that. So my 1070Ti aborts the CUDA111 tasks and will only run CODA100 in 20.04
The somewhat opposite is also true, a 3xxx NV card in 18.04 will abort CUDA111. I'd have to update OS or downgrade driver
 
Last edited:

SystemVipers

Member
May 18, 2013
162
171
116
Both GPU and CPU. GPU app TF is worth more credit.
3080Ti 95s
R Vii 290s
1070 Ti 584s

Also the project only works on GPU 0. If you have multiple GPUs, then both tasks will run on GPU 0. I guess a cc_config.xml could ignore the 1st GPU on a 2nd client although I've never tried that.
Edit: if you have Linux that supports CUDA 11 like 20.04 but a GPU w/o CUDA 11 support then you'll get both CUDA 10/11 tasks and the 11 tasks will fail. There is no server logic for that. So my 1070Ti aborts the CUDA111 tasks and will only run CODA100 in 20.04
The somewhat opposite is also true, a 3xxx NV card in 18.04 will abort CUDA111. I'd have to update OS or downgrade driver


yea I have that happening on a 2x 3060 rig, need to fix it, lol
 

Endgame124

Senior member
Feb 11, 2008
955
669
136
Hmm, I can't add the project through boinc (it fails) and I don't have an invitation code to do it through the website. What's the trick?
 

SystemVipers

Member
May 18, 2013
162
171
116
I can only get CUDA111 in 18.04 so I won't be running on a several cards.

Have you tried running 2 instances one for each card, I am trying that now

Hmm, I can't add the project through boinc (it fails) and I don't have an invitation code to do it through the website. What's the trick?

The invitation code for creating new accounts on SRBase is pillepalle, don't know if this one still works?
 

biodoc

Diamond Member
Dec 29, 2005
6,261
2,238
136
Hmm, I can't add the project through boinc (it fails) and I don't have an invitation code to do it through the website. What's the trick?
From the website:

The invitation code for creating new accounts is pillepalle
You can only create new accounts over the website.
 
  • Like
Reactions: SystemVipers

mmonnin03

Senior member
Nov 7, 2006
214
213
116
Have you tried running 2 instances one for each card, I am trying that now



The invitation code for creating new accounts on SRBase is pillepalle, don't know if this one still works?
I think the server code just looks at the driver version, gives the host CUDA111 tasks since the driver supports it. But the task requires glibc2.29 which isn't supported by 18.04. the CUDA100 app version doesn't require glibc2.29, something earlier but the server doesn't send it.
The dual GPU in one system has some CPDN thats that are a couple days from completion so I'm not going to reinstall 20.04 until those are done. Multiple clients wouldn't help in that case. If they were both pascal cards I could downgrade the driver with 2 clients.
 
  • Like
Reactions: SystemVipers

Endgame124

Senior member
Feb 11, 2008
955
669
136
Hmm, I created an account through the SRBase website, but I still can't add SRBase through boinc add project. It just says:
Failed to add project. Try again later.
Any ideas?
 

mmonnin03

Senior member
Nov 7, 2006
214
213
116
I think the Sprint page is missing a total credit number. TSBT last 3hr used to say 12mil, now its at 9.5m as the 1st hour dropped off. Not what the delta really is between teams.
 

Endgame124

Senior member
Feb 11, 2008
955
669
136
Which client version are you using?
I updated my client - it was at .11 and I updated to .20 and now its running. SRBase seems to be rather insane usage wise - junction temps on my 6700XT immediately up to 102C and the fans kicked up to an amazing (and loud) 4k rpm. I've been putting off moving around hardware so I can watercool my 3080ti and put it in this machine, but I may need to do that sooner than later if I'm going to run projects like this!