IRC logs of #boinc for Tuesday, 2011-06-21

00:12 *** freakazoid0223 has quit IRC

00:25 *** Simone1 has joined #boinc

00:28 *** SimoneCnt has quit IRC

00:30 *** zombie67 has quit IRC

01:15 *** up365 has quit IRC

01:17 *** up365 has joined #boinc

01:50 *** Tank_Master has quit IRC

01:52 *** Tank_Master has joined #boinc

02:05 *** SimoneCnt has joined #boinc

02:05 *** SimoneCnt has quit IRC

02:05 *** SimoneCnt has joined #boinc

02:08 *** Simone1 has quit IRC

02:27 *** SimoneCnt has quit IRC

02:52 *** desti_T2 has joined #boinc

02:55 *** desti has quit IRC

04:22 *** Tank_Master has quit IRC

04:22 *** Tank_Master has joined #boinc

04:53 *** Tank_Master has quit IRC

04:53 *** Tank-Master has joined #boinc

05:20 *** Caterpillar has joined #boinc

05:20 *** Caterpillar has quit IRC

05:20 *** Caterpillar has joined #boinc

05:23 *** Tank_Master has joined #boinc

05:26 *** Tank-Master has quit IRC

05:40 *** synapt has quit IRC

05:41 *** synapt has joined #boinc

05:54 *** Tank-Master has joined #boinc

05:54 *** Tank_Master has quit IRC

06:25 *** Caterpillar has quit IRC

06:25 *** Tank_Master has joined #boinc

06:28 *** Tank-Master has quit IRC

06:37 *** Caterpillar has joined #boinc

06:37 *** Caterpillar has quit IRC

06:37 *** Caterpillar has joined #boinc

06:56 *** Tank_Master has quit IRC

06:56 *** Tank-Master has joined #boinc

07:17 *** efc has quit IRC

07:18 *** Aeternus has joined #boinc

07:18 *** Aeternus has quit IRC

07:18 *** Aeternus has joined #boinc

07:26 *** Tank_Master has joined #boinc

07:30 *** Tank-Master has quit IRC

07:43 *** SimoneCnt has joined #boinc

07:43 *** SimoneCnt has quit IRC

07:43 *** SimoneCnt has joined #boinc

07:57 *** Tank_Master has quit IRC

07:57 *** Tank_Master has joined #boinc

08:28 *** Tank_Master has quit IRC

08:28 *** Tank_Master has joined #boinc

08:31 *** SimoneCnt has quit IRC

08:39 *** zombie67 has joined #boinc

08:48 *** up365 has quit IRC

08:55 *** os2mac has joined #boinc

08:59 *** Tank_Master has quit IRC

08:59 *** Tank_Master has joined #boinc

09:06 *** up365 has joined #boinc

09:30 *** Tank_Master has quit IRC

09:30 *** Tank_Master has joined #boinc

09:52 *** SimoneCnt has joined #boinc

09:52 *** SimoneCnt has quit IRC

09:52 *** SimoneCnt has joined #boinc

09:58 *** Marcofe has joined #boinc

09:58 *** Marcofe has quit IRC

09:58 *** Marcofe has joined #boinc

10:36 *** zombie67 has quit IRC

10:44 *** Simone1 has joined #boinc

10:47 *** SimoneCnt has quit IRC

11:34 *** SimoneCnt has joined #boinc

11:34 *** SimoneCnt has quit IRC

11:34 *** SimoneCnt has joined #boinc

11:37 *** Simone1 has quit IRC

12:12 *** Marcofe has quit IRC

12:18 *** Simone1 has joined #boinc

12:20 *** SimoneCnt has quit IRC

13:22 *** Aeternus has quit IRC

13:50 *** DerMeister has joined #boinc

13:51 *** Tank_Master has quit IRC

13:51 *** Tank_Master has joined #boinc

14:27 *** [B^S]renemayer has joined #boinc

15:07 *** yoyo[RKN] has joined #boinc

15:42 *** yoyo_RKN has joined #boinc

15:44 *** synapt has quit IRC

15:44 *** yoyo[RKN] has quit IRC

15:44 *** synapt has joined #boinc

15:44 *** synapt has quit IRC

15:44 *** synapt has joined #boinc

15:49 *** licensed has joined #boinc

15:49 *** licensed has quit IRC

15:49 *** licensed has joined #boinc

15:49 *** [B^S]renemayer has quit IRC

15:52 *** Aeternus has joined #boinc

16:13 *** SimoneCnt has joined #boinc

16:17 *** Simone1 has quit IRC

16:41 *** Caterpillar has quit IRC

16:42 *** Aeternus has quit IRC

16:55 *** DerMeister has quit IRC

17:21 *** licensed has quit IRC

17:24 *** yoyo_RKN has quit IRC

17:53 *** Simone1 has joined #boinc

17:55 *** SimoneCnt has quit IRC

18:21 *** Aeternus has joined #boinc

18:21 *** Aeternus has quit IRC

18:21 *** Aeternus has joined #boinc

18:37 <PovAddict> http://boinc.berkeley.edu/trac/wiki/DevProcess?action=diff&version=13 still losing more and more devs...

18:37 <Romulus> <http://tinyurl.com/3lxjn3s> (at boinc.berkeley.edu)

19:06 *** licensed has joined #boinc

19:06 *** licensed has quit IRC

19:06 *** licensed has joined #boinc

19:08 *** SimoneCnt has joined #boinc

19:08 *** SimoneCnt has quit IRC

19:08 *** SimoneCnt has joined #boinc

19:11 *** Simone1 has quit IRC

19:20 *** Tank_Master has quit IRC

19:20 *** Tank_Master has joined #boinc

19:33 *** Simone1 has joined #boinc

19:36 *** SimoneCnt has quit IRC

19:42 *** Aeternus has quit IRC

19:42 *** zombie67 has joined #boinc

19:47 *** zombie67 has quit IRC

19:58 *** SimoneCnt has joined #boinc

19:58 *** SimoneCnt has quit IRC

19:58 *** SimoneCnt has joined #boinc

20:01 *** Simone1 has quit IRC

20:10 <desti_T2> http://www.moonviews.com/

20:10 <Romulus> Title: MoonViews - Providing Imagery and Data For Lunar Exploration (at www.moonviews.com)

20:23 *** Simone1 has joined #boinc

20:26 *** SimoneCnt has quit IRC

20:47 *** SimoneCnt has joined #boinc

20:47 *** SimoneCnt has quit IRC

20:47 *** SimoneCnt has joined #boinc

20:51 *** Simone1 has quit IRC

21:03 *** efc has joined #boinc

21:37 *** Simone1 has joined #boinc

21:41 *** sadfgdf has joined #boinc

21:41 *** SimoneCnt has quit IRC

21:41 <sadfgdf> does anyone here know anything about app_info.xml files?

21:42 <Tank_Master> someone does yeah

21:43 <sadfgdf> or an idea of where I could ask?  My main GPU is currently running at 64 degrees C with fan speed at 44%.  I want to see if I can run two instances of a GPU app on my GPU.

21:43 <PovAddict> that sounds to me like three totally unrelated things

21:43 <Tank_Master> most apps use all the GPU power

21:44 <sadfgdf> I could be mis-stating

21:44 <efc> You can probably switch between them, but probably not run them at the same time..

21:44 <PovAddict> I mean app info is unrelated to running multiple instances of an app, and that's unrelated to GPU speed

21:44 <PovAddict> GPU temperature, I mean

21:44 <PovAddict> >.>

21:45 <sadfgdf> the anonymous-platform, if I'm correct, should be able to accomplish that....I'm clearly not running at 100%, as I see others with my same GPU putting out twice the amount of work units per amount of time

21:45 <sadfgdf> it was suggested

21:45 <sadfgdf> that under <coproc>

21:45 <sadfgdf> it shows <count>

21:46 *** licensed has quit IRC

21:46 <Tank_Master> i think early on MW had a way to run more than one app at once, but that was mainly do to them being so short, the GPU was idle more frim switching tasks than dong work

21:46 <Tank_Master> not true now

21:46 <sadfgdf> one site suggested that you could change 1.00000 under <count) to 0.50000 and that would make two instances run

21:46 <PovAddict> and why would you want to run two?

21:46 <PovAddict> to let them have some higher memory and GPU-bus contention and run even slower? :)

21:47 <sadfgdf> more work would get done, possible....

21:47 <sadfgdf> some people with load indicators said that some of their einstein stuff was only using 40% of their GPU

21:47 <sadfgdf> if that's so, then maybe if I could run 2 instances and it would be pulling 0.2CPUs for each instance

21:47 <desti_T2> it's using gpu only for some parts of the calculations

21:48 <PovAddict> which doesn't necessarily mean that running two will get it to 80%

21:48 <sadfgdf> instead of relegating itself to what it can do off of 0.2

21:48 <PovAddict> if it's not using all your GPU, there is a bottleneck somewhere

21:48 <sadfgdf> hmmm

21:48 <sadfgdf> okay

21:48 <sadfgdf> well

21:48 <sadfgdf> I am rather out of my depth

21:49 <sadfgdf> I'll just state facts

21:49 <sadfgdf> it took around 1,500 seconds of CPU time and about 6,000 seconds total run time to complete a taks

21:50 <sadfgdf> I stopped running two of my six cores and this changed these numbers to around 1,900 seconds of CPU time and around 4,000 total run-time seconds

21:50 <sadfgdf> an efficiency increase of 50%ish

21:50 * PovAddict thinks even now, years after it was introduced, GPU computing is still not ready -.-

21:51 <sadfgdf> I looked at someone else who had a GPU on there of the same variety of mine.  He was using the anonymous platform and had CPU times of around 2,300 and a run time of around 2,400

21:52 <sadfgdf> http://einstein.phys.uwm.edu/results.php?hostid=3633014&offset=0&show_names=0&state=3

21:52 <Romulus> <http://tinyurl.com/663rjw4> (at einstein.phys.uwm.edu)

21:53 <sadfgdf> I have the special edition of that GPU and mine is using less CPU time and takes longer to complete

21:54 <sadfgdf> ....so I'm thinking he's probably configuring things differently than I am

21:54 <PovAddict> are you using the same OS as the other guy who does tasks faster than you?

21:56 <sadfgdf> we're both using linux, but no

21:56 <sadfgdf> not the same

21:56 <sadfgdf> would it really make that much of a difference?

21:56 <sadfgdf> I'm using, "2.6.38-8-generic"

21:57 <sadfgdf> and he's using, "2.6.18-194.32.1.el5"

22:00 <sadfgdf> I also note that those who are getting these great results from their GPUS are not completing any non-GPU apps, even though they have an excess of cores, though I don't know how this might play into it

22:03 <PovAddict> I think that's a BOINC flaw

22:03 <PovAddict> not allowing both apps to run

22:04 <sadfgdf> not allowing which apps to run?

22:05 <sadfgdf> oh

22:05 <sadfgdf> it does

22:05 <sadfgdf> I've run 6 instances of their CPU app on my six cores at the same time I was running 2 GPU apps (1 on each GPU)

22:09 <sadfgdf> which means that for some reason they're choosing not to....  I chose to only run 4 instances of the CPU app and 2 of the GPU app.  This made a CPU core completely available to each GPU and increased GPU efficiency by 50%  (not running the app on those CPU cores will decrease my daily output by 2,000 points, but the increase of efficiency will raise it by 5,000 points).  Using 0 and 3 instances of the CPU app didn't increase efficiency and using 5 decreased ef

22:11 <sadfgdf> Could this be a driver issue?

22:11 <sadfgdf> I'm using nvidia-current

22:11 <sadfgdf> in Ubuntu

22:12 <sadfgdf> driver version: 270.41.06

22:12 * PovAddict doesn't do GPU BOINCing

22:14 <sadfgdf> apparently there's a newer driver

22:14 <sadfgdf> I'll try it

22:15 <sadfgdf> I'm using 270 and there's a 275

22:16 <sadfgdf> ha!  I may actually be using the 32 bit variant....that could explain it....

22:24 <Tank_Master> its quite common for the GPU app to suffer if the CPU is at full load

22:24 <efc> Wasn't Intel's cancelled graphics thing going to run x86? That would have been nice for boinc.

22:24 <Tank_Master> yes

22:24 <efc> though, it seems like it couldn't be terribly efficient doing so

22:25 <Tank_Master> OCL, OGL, DX and x86 iirc

22:25 <PovAddict> efc: did you read the article on how much x86, and the way Intel&AMD manage it, sucks?

22:25 <efc> no, its been sucking since about 1979 though

22:25 <efc> the FPU instructions were the first big blunder

22:26 *** sadfgdf has left #boinc

22:26 <Tank_Master> the small amount of registers is another

22:27 <PovAddict> efc: http://www.agner.org/optimize/blog/read.php?i=82

22:27 <Romulus> Title: Agner`s CPU blog - Pentium Appendix H (at www.agner.org)

22:27 *** SimoneCnt has joined #boinc

22:27 *** SimoneCnt has quit IRC

22:27 *** SimoneCnt has joined #boinc

22:31 *** Simone1 has quit IRC

22:33 <efc> opcode space pollution

22:35 <PovAddict> http://www.youtube.com/watch?v=aP3gzee1cps

22:35 <Romulus> Title: YouTube - Cat gets caught barking by a human and resumes meowing (at www.youtube.com)

22:49 <Tank_Master> lol nice

22:52 *** Simone1 has joined #boinc

22:55 *** SimoneCnt has quit IRC

23:17 *** SimoneCnt has joined #boinc

23:17 *** SimoneCnt has quit IRC

23:17 *** SimoneCnt has joined #boinc

23:21 *** Simone1 has quit IRC

23:36 *** Celelibi has quit IRC

23:42 *** Simone1 has joined #boinc

23:44 *** Celelibi has joined #boinc

23:45 *** SimoneCnt has quit IRC

23:53 *** GaelicGrime has quit IRC

Generated by irclog2html.py 2.4 by Marius Gedminas - find it at mg.pov.lt!