Formula-BOINC 2017

ChristianVirtual said:
I'm a bit power-constraint
Click to expand...

I'm running 1080 at 99W and it is at 57°C. Still get decent run time, about 8 mins 30 secs per task. What's yours?

upload_2017-6-29_6-54-55.png



upload_2017-6-29_6-58-24.png
 
longer runtimes ....

1) MSI SeaHawk 980Ti (AIO Watercooled) -> 15:17
2) "unnamed" in SMI: 1080Ti FounderEdition at stock; only fixed the fan on higher speed -> 9:30
3) 1080 (no idea what model) -> 12:50

Centos ...

I'm shocked that is quite slower compared to yours ... what have you done with your card ... overclocked ?
 
I'm using this config file for 1080
One task per GPU.
Windows 7, driver 373.06


verbose=1
kernels_per_reduction=32
threads=8
lut_size=17
sieve_size=30
sleep=1
reduce_cpu=0


To run the GPU cool, set Power Limit (%)= 50. Keep the GPU voltage at 0.78V. Nice and low.
For more details on GPU crunching efficiency, see my earlier post.
upload_2017-6-29_7-30-50.png
 
Seems like a very good compromise between performance and power/heat.
 
Crosshairs said:
this is problematic...
Click to expand...

I think you misunderstood me. You can join the sprint at anytime. That deadline was just for bunkering if you chose to do so. Otherwise, anytime between 4PM EST today and 4PM EST on the 2nd is when points count towards the sprint.
 
Ok, 5.5 minutes a task on Collatz on my 1080 is what I'm seeing right now.

Why do I have non GPU Collatz tasks? They are projected to take 3 days. Hopefully they aren't the same exact thing as the GPU tasks because if they are that is a waste of power. Can I limit it to only GPU tasks?
 
Go to the Collatz site. Log in. Change your project preferences to not use CPU.
 
Skillz said:
Do this on your 1080s.

Some recent reading and playing around for 1080/1080 Ti has me currently at the following settings:


collatz_sieve_1.21_windows_x86_64__opencl_nvidia_gpu.config

verbose=1
kernels_per_reduction=48
sleep=1
threads=9
lut_size=17
reduce_CPU=0
sieve_size=30
cache_sieve=1

Watch out for heat at these settings as the load moves from 70% to 95-100%. Not really a problem if you only have one gpu per rig.


app_config

<app_config>
<app>
<name>collatz_sieve</name>
<gpu_versions>
<gpu_usage>1.0</gpu_usage>
<cpu_usage>0.5</cpu_usage>
</gpu_versions>
</app>
</app_config>

Unlike Einstein, I haven’t tried running two tasks on each GPU yet as I’m pretty much at 100% load and the consensus seems to be to leave it be.

The above seems to be working well (Free-DC project # 1 in points so far today:)) but I’d like to experiment some more when time allows or alternate suggestions arise.

Edit: Removed max_concurrent line from app_config as suggested.​
Click to expand...

collatz_sieve_1.21_windows_x86_64__opencl_nvidia_gpu.config is located here:
C:\ProgramData\BOINC\projects\boinc.thesonntags.com_collatz

app_config is located here
C:\ProgramData\BOINC
Click to expand...

What would one modify to load the GPU to 80 to 90%? I'm cool with running at 100% when I'm gone, but when I'm here using the computer I'd like to be able to watch videos etc. Right now I can't even watch a lowres youtube video without it stuttering. I can see that I should be able to switch in a different file and tell Boinc Manager to reread the config files.
 
It would be better to just suspend the work unit as the client doesn't support multiple profiles like that.
 
An alternative, albeit more complicated, approach would be to remove/rename the collatz_sieve_1.21_windows_x86_64__opencl_nvidia_gpu.config file whilst you were using the computer.

The client will move back to standard usage when the next work unit begins. I just tested this and a youtube HD video ran fine whilst collatz was still crunching.

EDIT: Yeah, don't do this without replacing the file with a empty file. I think it could have caused the 100 unit upload failure I just experienced. That's 3M points I'll never see again.
 
Last edited:
Some of you guys should switch to the 10 teams below us, but skip Heidelberg. Make sure they don't get any points from this. LOL
 
We have a new cruncher, Adam J Bavier (AJB?) who just joined this challenge. If you happen to read this post, just drop us a line. We welcome you to our team(y)


upload_2017-6-29_21-8-18.png
 

Attachments

  • upload_2017-6-29_21-7-22.png
    upload_2017-6-29_21-7-22.png
    23.8 KB · Views: 14
pututu said:
We have a new cruncher, Adam J Bavier (AJB?) who just joined this challenge. If you happen to read this post, just drop us a line. We welcome you to our team(y)


View attachment 29097
Click to expand...

Thanks for the welcome! Those points are from 9 hours of processing on the 1080. This project certainly dishes out MILLIONS of points! :playful:

I'll stay on for this challenge. Not sure about long term as it is the summer and my AC bills are going to be :greedy:. Plus hopefully EVGA will contact me and tell me my stepup is ready and I'll have to pull it out. :happy:
 
the points are insane ! can't wait until the sprint is over and switch back to my beloved proteins; this stuff feels not right for me ... :oops:
 
Ah I didn't know that MN Scout is you with different alias. Great to see you here. Understand about the AC bills. I'm running 1080 with Power Limit to reduce AC bills since we already have a huge lead. We might need you to call of duty in next year 2018 BOINC pentathlon.
 
Yeah I got that alias first in life :D.

Yesterday my GPU was at 1950mhz, but today it is at 1809 not sure what the deal is. Haven't changed anything that I know of. Must be the boost doing random things, or maybe MSI Afterburner and EVGA Precision XOC are competing with each other in some way. I need to uninstall one of them....

Feel free to PM me a week before the pentathlon, and I can see what I can do.
 
Well you lot certainly have things under control! :D All my systems are down for the long holiday; but we'll see about the next challenge when it comes. :)
 
now that the sprint comes to an end I thought: switch back to GPUgrid; but seems they run dry ... (n):cry:

Put collate on 0 resource share as fall-back project; but seems it will keep busy.

Any other bio/pharma/non-math GPU project around ? SETI count as math ... :blackalien:
 
Asteroids - astronomy
Einstein - astro physics
BURP (if it has work) - video rendering
Enigma (just released a CUDA app) - Cracking the Enigma code from WW2
Milkyway - astronomy
SETI - astronomy?
Moo! Wrapper - cracking the RSA labs encryption

Then there is SETI BETA and Albert which are the test sites for SETI and Einstein.
 
Finished the 9th sprint event with gold again with almost 198M points ahead of the 2nd team. Well done [H]ordes!

Current medal tally:
G - S - B - None
5 - 2 - 1 - 1

Next sprint event: Thu, Jul 13 @ 4pm / 7pm (PST/EST)
 
What me really drives nuts that now miss 1 or 2 days in GPUGrid to reach 200M and they run dry with no response on any channel. Grrrrrr.
 
CPU only. The project admin mentioned about creating GPU applications but so far no news. I ran about a dozen tasks this morning to get a feel of it. The WU run time varies considerably (from about 1 hr to 7 hrs and still crunching). On my 2695v2, I'm getting about 83 credit points per hour per task with 12 threads running.

We are closing the gap with the top team in CSG. Since we have good turnout in CSG, I'll continue with CSG for a while. May revisit NF later on.
 
That shows I'm 3rd on the team for CSG. That's pretty good for the ol' G34 rig.

Since the gap is larger for CSG. Ill switch back and help try to close the gap.
 
Started running a few NFS@Home tasks to bump us up some, however some of these tasks are taking weeks to months to complete with a deadline that's only a few days. Something I am doing wrong on these?
 
Skillz Nope. Read that others reported that issue on the forum. They will "eventually" complete and restarting Boinc may help them complete.
 
Well I can't let these tasks run for 100+ days before they complete. Was gonna try to gain some positions in NFS, but looks like I'll have to move my resources else where. Already got us to the number 2 position in Wanless Mersenne +2 and #1 will happen in a week or so.
 
You must log in or register to reply here.
Back
Top