• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Info 12th BOINC Pentathlon 2021

Page 4 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
StefanR5R said:
There lie dangers in unattended reporting of bunkered results though:
  • The contest site may experience delays when pulling initial stats from the project server. Results which are validated before the initial stats were obtained don't count, obviously.
  • Congestion of the project server during the first hours may lead to transfer failures between server and client, which the client responds to with increasing retry backoff periods if left alone.
(And at my home specifically, the cheap cable internet connection works reliably according to Murphy's principle: It breaks up when I'm not there.)
Click to expand...
Thanks for highlighting this. I tried to mitigate risks a little, a) by only allowing the network usage way after the official pentathlon start time (+4h offset) and b) by getting up way before the earliest WU deadline anyhow 😉 However, a bad connection is out of our control... I'd say that's almost "force majeure" 😉
 
I think we are stuck at 2nd place. Every hour we loose 60-70k to Planet3DNow, 1.3m down now total.
 
There's still a long way to go, and anything can happen in terms of team positions and everything. That's what's great about the pentathlon, with all the different events and everything, based on how each team approaches it, the standings shouldn't stay static. Depending on how much various teams focus on the Javelin Throws, City Run etc. we can plan on either going up or going down in position throughout the competition 🙂
 
The SiDock admins sure love to change things around, at the drop of a hat.
– When the "cmdock + zipped input" application launched, they gave its tasks a 3 days deadline, unlike cmdock which was down to 2 days since Friday. But now the zipped input application is down to 2 days deadline too.
– Today they removed the server-side limit of 64 CPUs per host, which had capped the 2*active_cpus limit on tasks in progress at 128 tasks.
 
StefanR5R said:
The SiDock admins sure love to change things around, at the drop of a hat.
– When the "cmdock + zipped input" application launched, they gave its tasks a 3 days deadline, unlike cmdock which was down to 2 days since Friday. But now the zipped input application is down to 2 days deadline too.
– Today they removed the server-side limit of 64 CPUs per host, which had capped the 2*active_cpus limit on tasks in progress at 128 tasks.
Click to expand...
I now have about 100 per host staged in addition t the 128 running.
 
Post #1 updated with Cross Country: Einstein@Home dates.
I haven't written up the project details yet, but put a link to SG's Pentathlon Projects thread there.
 
I would do the 1080ti on Primegrid, and then move the 6700XT over to Einstein whenever it's convenient (the tasks times are long enough to bunker now, if you plan on bunkering)
 
Icecold said:
I would do the 1080ti on Primegrid, and then move the 6700XT over to Einstein whenever it's convenient (the tasks times are long enough to bunker now, if you plan on bunkering)
Click to expand...
PPS for prime grid on the 1080ti, or switch to a different project?
 
Yeah, I don't have a 1080ti but everything I've read says PPS Sieve is the highest PPD project on those. On some Turing and Ampere cards WW may or may not be higher PPD AFAIK.
 
Great! 1080ti is going on primegrid and the 6700XT is moving to Einstein. If I’m really feeling motivated, I’ll move the whole raspberry pi farm to Einstein too
 
Are those running one task at a time or 2? I seem to get the best PPD running 2 at a time. This is 2 at a time on a 2080ti with a 70% power limit on Windows-

5 May 2021 | 23:46:18 UTC​
6 May 2021 | 1:00:14 UTC​
Completed and validated (1st)​
200.7​
199.09​
3371​
PPS (Sieve) v1.42 (cudaPPSsieve)​
5 May 2021 | 23:46:18 UTC​
6 May 2021 | 1:07:44 UTC​
Completed and validated​
173.79​
172.17​
3371​
PPS (Sieve) v1.42 (cudaPPSsieve)​
5 May 2021 | 23:46:18 UTC​
6 May 2021 | 0:53:30 UTC​
Completed and validated​
200.91​
199.06​
3371​
PPS (Sieve) v1.42 (cudaPPSsieve)​
 
Trying out Einstein@Home and it's only giving me gravitation wave search work units despite having only selected gamma-ray pulsar binary search #1, how can I fix this? I've tried reverting the selections, saving, updating the client, re-selecting gamma-ray pulsar project, save, update client. Still only getting gravitational wave WUs.
 
It's frustrating how badly optimized these projects are for AMD hardware. I'm amazed that they're all still using OpenCL when HIP has been stable for over a year and basically gives you CUDA level speed with much less overhead.
 
bwoochowski said:
Trying out Einstein@Home and it's only giving me gravitation wave search work units despite having only selected gamma-ray pulsar binary search #1, how can I fix this? I've tried reverting the selections, saving, updating the client, re-selecting gamma-ray pulsar project, save, update client. Still only getting gravitational wave WUs.
Click to expand...
There are a few things I can think of-

Verify the host is in the correct "location" that you're changing. From the Einstein Page --> Account --> halfway down should be computers --> click the computer that is getting the incorrect tasks and see if it shows 'Generic' as the location, or something like school or work. It may be set to a different location than the preferences you're changing. When you're setting the preferences there is a drop down at the top to select the location but it defaults to generic.

Do you maybe not have it set to use the GPU type you're using? there should be 'Use Nvidia GPU', 'Use AMD GPU' etc. options. If you only want to run GPU tasks, set use CPU to no, AMD GPU or whatever is appropriate to yes, and then set 'Allow non-preferred apps: ' to no. If Allow non-preferred apps: is set to yes, it will grab stuff from other projects.

I'm not sure about the performance benefits with HIP, but Einstein using OpenCL should perform very well on the Vega 56 you mentioned earlier. I'm running it on a Vega 56 and it seems to be performing well compared to other Nvidia Hardware.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top