61) Message boards : News : Problems sorted :) (Message 3304)
Posted 5 Feb 2019 by mmonnin
Post:
You mentioned new app available. Do we have to download anything?


Nope, everything will download from the server automatically. You can see the app version and date implemented here:
https://universeathome.pl/universe/apps.php
62) Message boards : News : Problems sorted :) (Message 3299)
Posted 5 Feb 2019 by mmonnin
Post:
Thanks for the quick response. Much appreciated!

Onwards to 100m and the next Badge :)
63) Message boards : Number crunching : cheating? (Message 3291)
Posted 5 Feb 2019 by mmonnin
Post:
I'd rather have a fixed credit system per task. Having ones own credit vary because the wingman has a slow/fast PC just blows my mind. Some tasks are longer than others and the credit can vary but please be fixed per task length. This also makes it much easier to compare systems between users, between OSs, and a good way to monitor clocks remotely.
64) Message boards : Number crunching : Linux new app version (Message 3279)
Posted 4 Feb 2019 by mmonnin
Post:
Seem some new versions come through today. Up to 0.14 which seems faster on my 1950x than the 0.13 version. From about 62min on 0.13 to 33-45min on 0.14. Back down to how it was with 0.10. :)
65) Message boards : Number crunching : cheating? (Message 3278)
Posted 4 Feb 2019 by mmonnin
Post:
I checked some statistics... this looks pretty bad. Some users gained massively (doubling and more) and some even lost. No clue what the hell is going on, but the statistics are scuffed to the max at the moment.

Some more background info would be nice at this stage.


The one user is def cheating. These can't be completed in several seconds.

My guess is tasks from the past several days/weeks are being sent back through the validator and we're getting extra credit for tasks we've already done. That assumption is based off of the much higher pending tasks I've seen for myself AND the cheating user having thousands of invalids now.

For the rest of us its like extra credit, even if its much lower than the 667 credit we were getting.
66) Message boards : Number crunching : cheating? (Message 3273)
Posted 4 Feb 2019 by mmonnin
Post:
All those tasks would have to be re-validated too since it's just bogus results.

Oh, Is that whats happening right now? Are old results being sent back through the validator? I have 10x the normal pending tasks and credit is several times normal today even though credit per task is like 1/8 to 1/6 of yesterday.
67) Message boards : Number crunching : cheating? (Message 3270)
Posted 4 Feb 2019 by mmonnin
Post:
Looks like there's a new credit system. Boinc's Credit screw New?

Credit today is higher for everyone yet credit per task is much lower. And many times its zero credit for valid tasks. :(
https://universeathome.pl/universe/workunit.php?wuid=23329615

No one likes this credit system. 69.1 credit for this task at 2,634.6 seconds
https://universeathome.pl/universe/workunit.php?wuid=23330497

Same PC just 9 seconds difference in CPU time (2,643.46) yet 84 credit just because the wingman was faster.
https://universeathome.pl/universe/workunit.php?wuid=23330422
68) Message boards : Number crunching : Missing Team (Message 3263)
Posted 4 Feb 2019 by mmonnin
Post:
My guess would be this. Looks like team founders should be on the lookout for shenanigans.

http://www.cosmologyathome.org/forum_thread.php?id=7591


Wow I hadn't heard of that before. I guess its been at least Nov since I've head Cosmo forum.

Thanks for restoring it Krzysztof.
69) Message boards : Number crunching : Missing Team (Message 3256)
Posted 3 Feb 2019 by mmonnin
Post:
For some reason the team Overclock.net disappeared and we are all on a team called INTERNATION instead. What happened?
70) Message boards : Number crunching : Linux new app version (Message 3254)
Posted 3 Feb 2019 by mmonnin
Post:
I see the same thing. On my i7-3770 (Ubuntu 16.04.5), the 0.10 took 50 minutes, while the 0.11 takes 73 minutes.

I now see some 0.11 work units completing in 45 minutes or less. So it was probably just the normal variation in the work unit length that I was seeing, not any significant difference between 0.11 and 0.10 (or 0.09 either).

It would take a much longer average to detect any real differences, if there are any.


One other thing is that on my Zen systems some tasks take 2-3x processing time on 0.09 and 0.11 but I didn't see that on 0.10.

BOINCTask ETA is 50min for 0.11, much slower than the 36min average on 0.10.
71) Message boards : Number crunching : Linux new app version (Message 3250)
Posted 2 Feb 2019 by mmonnin
Post:
Looks like windows is already on v0.12
72) Message boards : Number crunching : Linux new app version (Message 3245)
Posted 2 Feb 2019 by mmonnin
Post:
Another version on server.
Should work on kernels 3.2 and above.


It works. For some reason the 0.10 version was much faster. Like 55min to 36min. 0.11 is back to the 0.09 crunch times if not slower.

On a system where 0.10 did not work, v 0.11 is much slower than 0.09. From 70min to 110min.
73) Message boards : Number crunching : Linux new app version (Message 3240)
Posted 31 Jan 2019 by mmonnin
Post:
Correction, the Mint 17 2P 2670v1 system also has the issue. So many aborted that they never reported for me to see until I arrived home.
74) Message boards : Number crunching : Linux new app version (Message 3238)
Posted 31 Jan 2019 by mmonnin
Post:
Most are fine for me. One PC has all errors
https://universeathome.pl/universe/results.php?hostid=39977
Ubuntu 14.04, 3570k

Mint 17 and 2x Ubuntu 18.04 systems are ok.
75) Message boards : News : Patreon (Message 3204)
Posted 9 Jan 2019 by mmonnin
Post:
Patreon is basically a GoFundMe in another format.


It's like a monthly donation, not a one time donation. Or so is my understanding of it. To generate a monthly salary income. Monthly I cannot do. A one time donation here or there, sure, but I'd like to know exactly what it was for as mentioned above. I've got 5 computers and a cell phone crunching this project and my electric bill is way above average. As someone said above, I do it for science and am paying though burning out my equipment like hdds/ssds cpus etc as well as a whopping electric bill. I don't mind donating something, but doing it every month gave me that sour taste in my mouth.

I live near a very large national laboratory. Job descriptions for job openings include knowing how to be able to apply for grants to get projects funded. That seems to be what should be done here. Or, setup a way to make one time donations towards a set goal.

Just my 2 cents.


Make a donation and cancel it. 1 time donation. Done. There is no contract. One can start and stop donations whenever.
76) Message boards : Number crunching : Project has no tasks available (Message 3180)
Posted 29 Dec 2018 by mmonnin
Post:
It was fine for me yesterday during the time of the previous posts. I still have some available but I am getting the no tasks available message. 4153927 ready to send and server status page is current.
77) Message boards : News : Patreon (Message 3173)
Posted 28 Dec 2018 by mmonnin
Post:
Patreon is basically a GoFundMe in another format.
78) Message boards : Number crunching : Threadripper for Universe@Home (Message 3040)
Posted 23 Sep 2018 by mmonnin
Post:
Clocks of any Ryzen CPU should be enough. Clocks will vary output more than the difference between Zen and Zen+.

Any anyway its not hard to find the top computers list that is available for every BOINC project.
https://universeathome.pl/universe/show_host_detail.php?hostid=502490

That PC isn't running Universe full time though. Only 21 tasks atm. Find the average completion time and multiply for # of cores. RAC is useless.
79) Message boards : Number crunching : neverending wus (3-4-8-9-days) (Message 3029)
Posted 26 Aug 2018 by mmonnin
Post:
The issue goes back several years. Just abort them.
80) Message boards : Number crunching : 0.06 Universe BHspin v2 "Long Runner" (Message 2961)
Posted 12 Jul 2018 by mmonnin
Post:
That's been an issue for several years.


Previous 20 · Next 20




Copyright © 2024 Copernicus Astronomical Centre of the Polish Academy of Sciences
Project server and website managed by Krzysztof 'krzyszp' Piszczek