IRC logs of #boinc for Sunday, 2011-11-20

03:25 *** efc has quit IRC

03:31 *** efc has joined #boinc

04:39 *** efc has quit IRC

04:53 *** Haldrik has joined #boinc

04:58 *** Aeternus has joined #boinc

04:58 *** Aeternus has quit IRC

04:58 *** Aeternus has joined #boinc

05:16 *** desti_T2 has quit IRC

05:30 *** desti_T2 has joined #boinc

05:36 *** Caterpillar has joined #boinc

06:09 <CoderForLife>

06:09 <Romulus> <> (at

06:29 *** Aeternus has quit IRC

07:39 *** quail has quit IRC

07:46 *** quail has joined #boinc

08:36 *** mweltin has quit IRC

08:36 *** mweltin has joined #boinc

08:38 *** trogdog has joined #boinc

08:40 *** Marcofe has joined #boinc

08:40 *** Marcofe has quit IRC

08:40 *** Marcofe has joined #boinc

09:16 *** freakazoid0223 has joined #boinc

09:56 *** virtualdirt has joined #boinc

09:56 <virtualdirt> hi all

10:22 *** trogdog has quit IRC

10:37 <PovAddict> CoderForLife: should have asked who is on second

10:37 <PovAddict> "no, he's on first, what's on second"

12:16 <Romulus> New news from boinc: Welcome Mersenne@home

12:50 *** Marcofe has quit IRC

13:30 *** yoyo[RKN] has joined #boinc

13:42 *** MTughan has joined #boinc

13:44 <MTughan> So BOINC won't seem to request work from PG anymore. I got the communications issue fixed (it was my firewall blocking, haven't tried to find an exception rule for it yet), and finished off and reported tasks I had. But now the CC won't request any work.

13:44 <MTughan> I'm trying to find debts for the projects, which I thought were in client_state.xml, but I don't seem them there. Any ideas what might be causing the non-fetch?

13:45 <MTughan> The problem now is that E@H and WCG together are keeping my CPU busy with enough work, but my GPU will run out of work in about 1.5-2 hours.

13:59 *** PovAddict has quit IRC

13:59 *** PovAddict has joined #boinc

14:04 <PovAddict> MTughan: debts are indeed in client_state.xml, just search for the word 'debt' :P

14:04 <RomW> I don't think the new scheduler uses the debt fields anymore

14:04 <RomW> They may no longer be in the client_state.xml file

14:05 <PovAddict> the scheduler needs a version number, the terms "old scheduler" and "new scheduler" change meaning every year

14:05 <RomW> 6.13

14:05 <RomW> :)

14:05 <RomW> or REC based scheduling

14:16 <RomW> On the plus side, MTughan, you can use the client emulator ( to see what is going on.

14:16 <Romulus> Title: The BOINC Client Emulator (at

14:22 *** Saenger has joined #boinc

15:30 <RomW> In you all opinion, what are the most important pieces of information about an active task?

15:38 *** PovAddict has quit IRC

15:38 *** PovAddict has joined #boinc

15:39 <Saenger> reliable progress bar, deadline, dl-date, estimated left runtime, RAM-use, time since last checkpoint

15:39 *** efc has joined #boinc

15:45 <RomW> I'm not sure if I can make the pregress bar any more reliable than what the CC knowns...

15:46 <RomW> knows

15:51 <MTughan> RomW: Thanks for the link, but it appears since my last message that BOINC has indeed asked for PG work to keep the GPU filled.

15:52 <Saenger> I know, it's mainly how predictable the WUs are, and that's something even the projects themselves don't know most of the time ;)

15:52 <RomW> cool deal

15:53 *** yoyo[RKN] has quit IRC

16:03 *** Marcofe has joined #boinc

16:03 *** Marcofe has quit IRC

16:03 *** Marcofe has joined #boinc

16:15 <PovAddict> RomW: well sure, the "reliable" part is up to the project :)

16:17 <RomW> Well, I'm just trying to figure out what to put into an active task control

16:17 <RomW> I was hoping to keep it down to two or three items

16:17 <PovAddict> not sure if this is the place to ask

16:18 <RomW> I think I'm just going to have to make it configurable

16:18 <PovAddict> most people in this channel would prefer a panel full of information and knobs

16:18 <PovAddict> #boinc is definitely not a good statistical sample of the BOINC user population :)

16:18 <RomW> I wasn't looking for the overall population

16:18 <RomW> :)

16:19 <RomW> Just the segment that is likely to use alternative UIs

16:19 <PovAddict> ah, you're not making a new *default* UI :)

16:19 <RomW> No

16:20 <RomW> Although, some of the new technology might find its way into BOINC

16:20 <RomW> Although, that is up to David

16:20 <RomW> I'm just fooling around right now...

16:20 <PovAddict> yeah but that's different

16:21 <RomW> The first part of the project was re-inventing the RPC mechinism

16:21 <PovAddict> you could make a UI that is completely unsuitable to be the default, and yet the underlying protocol code could still be suitable for BOINC

16:21 <RomW> Which exceeded my expectations

16:22 <RomW>

16:22 <Romulus> <> (at

16:22 <PovAddict> the "RPC client" mechanism, or end-to-end (new protocol)?

16:22 <RomW> Just the manager-side

16:22 <RomW> It consumes the same information as the current manager

16:22 <PovAddict> yeah, terminology is unfortunately confusing there

16:22 <RomW> just in a multi-threaded way

16:23 <RomW> using a thread pool

16:23 <PovAddict> the BOINC client has the server-side of the RPC stuff :)

16:24 <RomW> My initial crack at this on my computer gave me a 30x perf boost

16:24 <RomW> on one of the big machines (8 core with hyperthreading, and 3 GPUs) it went up to 50x

16:25 <PovAddict> did you test its scalability?

16:25 <PovAddict> it wouldn't surprise me if the original code gets more-than-linearly slower with lots of tasks, and your new one doesn't

16:26 <RomW> Well, I only have two plot points

16:26 <PovAddict> ah, lol

16:26 <RomW> So, it would need to be tested on a bunch of different machines

16:27 <RomW> I figured the validation of the APIs could be handled by playing around with a new UI

16:27 <PovAddict> indeed

16:27 <RomW> So, I'm playing around writing a new UI

16:28 <RomW> Although, I might re-invent the advanced GUI just to compare CPU loads

16:28 * PovAddict dives back into conic sections, bbl

16:28 <RomW> relative to the stock implementation

16:29 <PovAddict> high school: an ellipse is defined like this, and its equation is this

16:30 <PovAddict> university: an ellipse is defined like this; now let's deduce the equation from that definition alone

16:30 <RomW> lol

16:30 <RomW> fun fun fun

16:30 <PovAddict> then three blackboards of algebra

16:30 <PovAddict> and finally "oh btw this deduction is a likely exam question"

16:31 <PovAddict> and a smartass student asked if at least that can be the *only* exam question, but got a negative reply

16:32 <RomW> ouch

16:33 <PovAddict> a friend had that class last year, and told me at one point the prof drew a diagram and labeled a point 'G'

16:34 <PovAddict> and the students couldn't hold their laughter

16:34 <PovAddict> especially when some conic section, given some parameters, "degenerated" into the G point

16:35 <PovAddict> ( see where it redirects ;)

16:35 <Romulus> Title: G-Spot - Wikipedia, the free encyclopedia (at

16:36 <RomW> lol

16:37 <Tank_Master> hiya RomW

16:38 <RomW> Howdy TM

17:09 *** Saenger has quit IRC

17:18 *** Marcofe has quit IRC

17:31 *** Caterpillar has quit IRC

17:33 *** Marcofe has joined #boinc

17:41 *** whynot has joined #boinc

17:43 *** Marcofe has quit IRC

18:29 *** Haldrik has quit IRC

19:00 <CoderForLife> somebody call me?

19:01 * CoderForLife thinks it probably wasn't G-spot that lit up his indicator

19:20 <desti_T2>

19:31 *** whynot has quit IRC

19:39 *** freakazoid0223 has quit IRC

20:09 <PovAddict> [08:09] <CoderForLife>

20:09 <Romulus> <> (at

20:10 <PovAddict> [12:37] <PovAddict> CoderForLife: should have asked who is on second

20:10 <PovAddict> [12:37] <PovAddict> "no, he's on first, what's on second"

20:11 * efc lights up CFL's indicators

20:33 *** efc has quit IRC

21:50 *** virtualdirt has quit IRC

22:58 *** tolkien has joined #boinc

23:32 <RomW>

23:32 <Romulus> <> (at

Generated by 2.4 by Marius Gedminas - find it at!