non-CPU intensive??
log in

Advanced search

Message boards : Number crunching : non-CPU intensive??

Author Message
Honza1616
Send message
Joined: 14 Dec 15
Posts: 3
Credit: 405,610
RAC: 956
Message 1643 - Posted: 29 Mar 2017, 23:54:38 UTC

my Skylake 6700k CPU with HT commonly employed for 8 WCG WU + 4 for Goofyx
but now only works 4WCG and 4 Goofyx WU
Goofyx but fails to exploit the full power of the CPU, and half the CPU does not work.

If Goofyx pause for a moment, I will start another 4 units on another project and re-counting the 8 + 4, but after a while these units pauses and again counting only 4 other +4 Goofyx
before Goofyx was simple application that does not burden the CPU as
WUProp@Home ....WTF?
sry for my Eng.

mmonnin
Send message
Joined: 28 Oct 16
Posts: 7
Credit: 1,618,800
RAC: 7,467
Message 1644 - Posted: 30 Mar 2017, 0:13:26 UTC

See some of the other threads. The other most recent one has app_config.xml settings to allow for 8 threads to crunch again.

Chris Skull
Send message
Joined: 28 Jan 17
Posts: 8
Credit: 539,810
RAC: 4,745
Message 1649 - Posted: 30 Mar 2017, 10:29:22 UTC

after a while this problem was over i have it back also since yesterday...
Solution: suspend Goofyx from affected machines :)
____________
Greetz,
Chris

Honza1616
Send message
Joined: 14 Dec 15
Posts: 3
Credit: 405,610
RAC: 956
Message 1650 - Posted: 30 Mar 2017, 10:47:17 UTC - in response to Message 1649.

...Solution: suspend Goofyx from affected machines :)

Yes, I already tried. But the suspension of the project is not a long term solution

Profile Tarmo Ilves
Avatar
Send message
Joined: 18 Jan 16
Posts: 22
Credit: 23,841,120
RAC: 102,500
Message 1651 - Posted: 30 Mar 2017, 11:20:25 UTC

Looks like this mysterious problem is gone.
All my Windows boxes behaving OK again.
____________

Profile [boinc@poland]goofyx
Project administrator
Project developer
Project scientist
Help desk expert
Send message
Joined: 8 Nov 15
Posts: 540
Credit: 9,002,054
RAC: 8,283
Message 1658 - Posted: 31 Mar 2017, 6:36:29 UTC - in response to Message 1651.

problem solved

jamezz
Send message
Joined: 31 Mar 17
Posts: 3
Credit: 127,250
RAC: 4,222
Message 1740 - Posted: 4 Jun 2017, 17:24:48 UTC
Last modified: 4 Jun 2017, 17:58:29 UTC

Maybe solved for your windows users but on every single Linux ( Gentoo using BILD the Boinc Italy Linux Disro latest release) 16 core and 8 core blades I run at home along with all arm+android devices we have issues.
Its not properly a NCI application , they are running as if 1 WU = 1 Core . They do not start that way , but after boinc manages cpu usage and memory it see's your half nci application transforms itself to taking a core and then its cpu intensive using 100% of a core. So 8 core crunching a Gridcoin whitelisted project I get paid money to crunch and your project just cut out 50% pay for that machine , my blades with dual xeon x5550's suddenly 12 cores getting paid to crunch and your project steals cash from me via 4 threads each machine.
Now I have no problem crunching a NCI project like quake catcher , wuprop and GG if it functions correctly but this seems rushed , sadly you took your time 2 weeks turned 2 months almost and still issues?
Kill dd@home if your having issues running multiple projects on 1 server or get more servers since you have users that invest in you how about investing a little more in us and your research/project?
If you are still part of dd and sodumb and its failboat promise to pay users as if trying to replace Gridcoin with dd|coin and be a Curecoin copy project ran by a bunch of nerds/geeks with hardware for hire $$$ and needy scientists and doctors that need data crunched and do not have the resources , so the nerds to crunch it! zomg then use a free distributed computer system... sever your ties man that left a bad taste in many boincer's mouths.
Your windows executables are fine quad core machine 4 gg processes and 4 of my chosen Gridcoin whitelisted projects workunits to get paid for on those machines , but then we go over to Linux and 50% and 20% my resources and its not NCI. You can put NCI in an application name and ghettofy it all to hell rigging it to work almost correctly , but where is the integrity?
Also , how about adding SSL to your project so your crunchers are secure especially if your going to be tied into the DD@home bull especially as the trust level is extremely low due to that bs and your 2 weeks mia turning 2 months. Yes your data , but you obviously need us and you need your project to work correct.

Screen Shots:
HP BL460c G6 2 x X5550 48gb BILD OS loaded in RAM = https://ibb.co/fwAHfv GG stealing 20% pay per my credit to crunch srbase per device
Dual Core ARM7 Tablet https://ibb.co/fBquSa 1/4 GG stealing 50% my pay for credit to crunch seti@home per device

" added- Now same dual core ARM7 tablet and GG has taken both CPU cores 100%
and push my actual CPU project aside do to GG's needed resources. You are
stealing full cpu claiming its NCI but its bunk code and not limiting it is
allowing 100% and no user should need to modify an .xml plan class or ncpu
cpu% etc it should work out of box. This is not NCI , look it kicked
seti@home to the side and took over 100% both cores. "

PS: suspend for a min and letting those cores crunch as they should without gg mucking up stuff just lasts a few minutes and is an asinine fix.

kuckoo
Send message
Joined: 5 Jun 17
Posts: 1
Credit: 0
RAC: 0
Message 1741 - Posted: 5 Jun 2017, 13:39:41 UTC - in response to Message 1740.

Thanks for the lolz


There was me thinking ignorance was bliss when it clearly isn't for you

HAL9000
Avatar
Send message
Joined: 19 Nov 15
Posts: 88
Credit: 66,085,300
RAC: 146,392
Message 1742 - Posted: 6 Jun 2017, 1:36:01 UTC - in response to Message 1741.
Last modified: 6 Jun 2017, 1:39:38 UTC

Thanks for the lolz


There was me thinking ignorance was bliss when it clearly isn't for you

It is more amusing that they post the message on the old site instead of the new one. They clearly know the new address given grid computing center is displayed in their screenshots.
Also to bad they don't seem to visit there. Since it is where I posted a workaround/fix for the apps not being setup as nci correctly.
____________

Join the BP6/VP6 User Group today!

Profile [boinc@poland]goofyx
Project administrator
Project developer
Project scientist
Help desk expert
Send message
Joined: 8 Nov 15
Posts: 540
Credit: 9,002,054
RAC: 8,283
Message 1744 - Posted: 15 Jun 2017, 23:10:17 UTC

It's old problem.
At this moment gcc is set up as nci project so monkeys will work as NCI in boinc clients

Message boards : Number crunching : non-CPU intensive??


Main page · Your account · Message boards


Copyright © 2018 Goofyx

Generated 26 Sep 2018, 12:13:59 UTC