IRC log for #brlcad on 20120707

01:21.07*** join/#brlcad jbschw (~jbschw@ool-4355ee10.dyn.optonline.net)
03:09.49CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r4065 10/wiki/User:Plussai/GSoC_2012_log: /* 7 May 2012 */
03:16.51CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r4066 10/wiki/User:Plussai/GSoC_2012_log: /* 23 May 2012 */
03:17.33CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r4067 10/wiki/User:Plussai/GSoC_2012_log: /* 21 May 2012 */
03:20.03CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r4068 10/wiki/User:Plussai/GSoC_2012_log: /* ''Week 1'' */
03:22.55CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r4069 10/wiki/User:Plussai/GSoC_2012_log: /* ''Week 1'' */
03:23.35CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r4070 10/wiki/User:Plussai/GSoC_2012_log: /* ''Week 1'' */
03:24.16CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r4071 10/wiki/User:Plussai/GSoC_2012_log: /* 29 May 2012 */
03:24.55CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r4072 10/wiki/User:Plussai/GSoC_2012_log: /* 31 May 2012 */
03:25.25CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r4073 10/wiki/User:Plussai/GSoC_2012_log: /* ''Week 2'' */
03:26.19CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r4074 10/wiki/User:Plussai/GSoC_2012_log: /* 4 June 2012 */
03:26.49CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r4075 10/wiki/User:Plussai/GSoC_2012_log: /* 12 June 2012 */
03:27.29CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r4076 10/wiki/User:Plussai/GSoC_2012_log: /* 18 June 2014 */
03:27.59CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r4077 10/wiki/User:Plussai/GSoC_2012_log: /* 26 June 2012 */
03:28.40CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r4078 10/wiki/User:Plussai/GSoC_2012_log: /* 2 July 2012 */
03:35.02CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r0 10/wiki/Special:Log/upload: uploaded "[[Image:Ell1.png]]"
03:36.33CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r4080 10/wiki/User:Plussai/GSoC_2012_log: /* 4 July 2012 */
03:37.29CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r0 10/wiki/Special:Log/upload: uploaded "[[Image:Ell2.png]]"
03:38.22CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r0 10/wiki/Special:Log/upload: uploaded "[[Image:ELL3.png]]"
03:39.09CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r4083 10/wiki/User:Plussai/GSoC_2012_log: /* 6 July 2012 */
03:39.51CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r0 10/wiki/Special:Log/upload: uploaded "[[Image:Ell3.png]]"
03:41.39CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r4085 10/wiki/User:Plussai/GSoC_2012_log: /* 6 July 2012 */
03:44.23CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r0 10/wiki/Special:Log/upload: uploaded a new version of "[[Image:Ell1.png]]"
03:45.07CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r4087 10/wiki/User:Plussai/GSoC_2012_log: /* 6 July 2012 */
03:46.40CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r0 10/wiki/Special:Log/upload: uploaded a new version of "[[Image:Ell2.png]]"
03:50.17CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r0 10/wiki/Special:Log/upload: uploaded a new version of "[[Image:Ell3.png]]"
03:51.37brlcadandrei_: try removing ssh from the script .. we're testing the behavior of libpkg, not your network setup
03:51.47brlcadthat might give you some better performance
03:54.03CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r0 10/wiki/Special:Log/upload: uploaded a new version of "[[Image:Ell3.png]]"
03:54.51brlcadalso, how long are we talking here?  .. if 1-2048 with 8mb would complete within 24 hours, then it's worth it so we can merge the graphs
03:55.41CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r4091 10/wiki/User:Plussai/GSoC_2012_log: /* 6 July 2012 */
03:55.49brlcadif it's longer, something is probabably wrong .. that's transferring 16GB, even 1byte at a time it shouldnt' be too bad
03:56.06CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r4092 10/wiki/User:Plussai/GSoC_2012_log: /* 6 July 2012 */
03:56.52CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r4093 10/wiki/User:Plussai/GSoC_2012_log: /* 6 July 2012 */
03:56.53brlcadandrei_: 1024 should be the average time, so you can run just that size once and multiply by 2048 to get a time estimate for all of them
03:57.41CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r4094 10/wiki/User:Plussai/GSoC_2012_log: /* 6 July 2012 */
03:57.47brlcadif it's going to be more than 24 hours, just drop the file size down to 4096 bytes instead of 8mb
03:58.37CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r4095 10/wiki/User:Plussai/GSoC_2012_log: /* 6 July 2012 */
03:59.35CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r0 10/wiki/Special:Log/upload: uploaded "[[Image:Ell4.png]]"
04:00.53CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r4097 10/wiki/User:Plussai/GSoC_2012_log: /* 6 July 2012 */
04:01.11CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r4098 10/wiki/User:Plussai/GSoC_2012_log: /* 6 July 2012 */
04:01.54CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r4099 10/wiki/User:Plussai/GSoC_2012_log: /* 6 July 2012 */
04:02.22CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r4100 10/wiki/User:Plussai/GSoC_2012_log: /* 6 July 2012 */
04:03.28CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r4101 10/wiki/User:Plussai/GSoC_2012_log: /* 6 July 2012 */
04:04.04CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r0 10/wiki/Special:Log/upload: uploaded "[[Image:HYP1.png]]"
04:41.29CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r0 10/wiki/Special:Log/upload: uploaded a new version of "[[Image:HYP1.png]]"
04:43.26andrei_it should take 15 min.
04:45.03CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r0 10/wiki/Special:Log/upload: uploaded a new version of "[[Image:HYP1.png]]"
04:47.47CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r0 10/wiki/Special:Log/upload: uploaded a new version of "[[Image:HYP1.png]]"
04:48.21CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r0 10/wiki/Special:Log/upload: uploaded "[[Image:HYP2.png]]"
04:49.11CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r4107 10/wiki/User:Plussai/GSoC_2012_log: /* 6 July 2012 */
04:49.27CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r0 10/wiki/Special:Log/upload: uploaded "[[Image:HYP3.png]]"
04:52.15CIA-50BRL-CAD: 03Plussai 07http://brlcad.org * r0 10/wiki/Special:Log/upload: uploaded "[[Image:HYP4.png]]"
05:05.46*** join/#brlcad Stattrav (u3131@gateway/web/irccloud.com/x-ndydjenkhynfjzta)
08:10.40*** join/#brlcad CIA-23 (~CIA@cia.atheme.org)
09:04.28*** join/#brlcad stas__ (~stas@188.24.47.249)
10:26.37*** join/#brlcad stas (~stas@188.24.46.176)
10:44.58*** join/#brlcad stas (~stas@188.24.35.127)
10:58.47*** join/#brlcad stas (~stas@188.24.33.157)
11:06.30CIA-23BRL-CAD: 03Phoenix 07http://brlcad.org * r4110 10/wiki/User:Phoenix/GSoc2012/Reports: /* Week 7 */
11:32.49CIA-23BRL-CAD: 03Phoenix 07http://brlcad.org * r4111 10/wiki/User:Phoenix/GSoc2012/Reports: /* Week 7 */
11:33.33CIA-23BRL-CAD: 03Phoenix 07http://brlcad.org * r4112 10/wiki/User:Phoenix/GSoc2012/Reports: /* Week 7 */
11:34.30CIA-23BRL-CAD: 03Phoenix 07http://brlcad.org * r4113 10/wiki/User:Phoenix/GSoc2012/Reports: /* Week 6 */
11:34.48CIA-23BRL-CAD: 03Phoenix 07http://brlcad.org * r4114 10/wiki/User:Phoenix/GSoc2012/Reports: /* Week 7 */
13:26.09brlcadandrei_: great, then can't wait to see the graph :)
13:29.31andrei_hm
13:29.43andrei_1024 took 0.5 seconds
13:30.00andrei_I have been running the script for approximatively  8 hours
13:30.06andrei_it currently is on 35 bytes
13:30.34andrei_running the 2048 => 2048 * 2048 aswell. That one is like 70%
13:34.12andrei_I want to stop the small one at some point ( probably at 100 bytes)
13:34.33andrei_and then increase  distance between tests up to 100
13:35.57andrei_does this seem like a good ide?
13:35.59andrei_idea*
15:16.22*** join/#brlcad cristina (~quassel@188.24.33.157)
15:16.36*** join/#brlcad cristina (~quassel@unaffiliated/cristina)
15:42.21*** join/#brlcad jbschw (~jbschw@ool-4355ee10.dyn.optonline.net)
15:47.23brlcadandrei_: actually testing every byte between 1 and 2048 is intentional to see if there are any anamolies
16:21.35andrei_ah, it s at 65 now.  The good part is it speeds up as it goes.
16:28.30andrei_the "large" one is almost done tho.
16:31.07andrei_what if I let the small one finish ( even if it takes " ages " ) and do  remaining tests instead?
16:55.57*** join/#brlcad xth1 (~thiago@187-40-100-224.user.veloxzone.com.br)
17:34.56brlcadandrei_: at the rate it's increasing, it should be done really soon
17:35.06brlcadapparently non-linear
17:35.16brlcadsee, that's actually rather informative
17:36.01andrei_the large one is done. Currently I m processing info to be used by octave
17:36.37brlcadnot sure what you're asking though -- yes, please let 1-2048 finish, then do remaining 2048-4194304 in 2048 increments (with the same 8mb file)
17:36.59brlcadthe order they're done certainly really matter
17:37.05brlcadcertainly *doesn't* matter
17:37.27andrei_the 2048-4194304 is done
17:37.55andrei_I m " drawing " the graph soon
17:38.49brlcadawesome
17:39.16andrei_and the small one hit 80
17:39.36brlcadso maybe 2-3 hours more
17:39.56brlcadcan you graph what you already have so far with 1-80?
17:40.14andrei_I should be able to, yes.
17:47.59andrei_for some really wierd reason netbeans refuses to open
18:26.01andrei_brlcad, I have the graphic for 2048 -> 2048^2
18:26.15andrei_it looks linear
18:26.24andrei_uploading a picture now
18:29.50andrei_http://imgur.com/Ot7vb
18:29.58andrei_Sorry for not adding legend to axis
18:30.38andrei_no, I made a mistake. Ignore that, please.
18:36.43andrei_this is the valid graphic : http://i.imgur.com/H5Q2u.jpg
18:37.10andrei_on X axis is package size  ( 2048 -> 2048 * 2048 )
18:37.16andrei_on y is speed.
18:41.52andrei_1 - 2048 hit 100.
18:41.59andrei_how does that graphic look?
18:52.27CIA-23BRL-CAD: 03brlcad * r51441 10/brlcad/trunk/TODO: looks like the rendering of part may be wrong when the pill squishes in on itself.
18:53.51brlcadx is package size??
18:54.01brlcadpackage size should be 1-2048...
18:54.02andrei_yes
18:54.05andrei_no no
18:54.17andrei_I said that s the graphic for 2048 - > 2048 * 2048
18:54.54brlcadah, so 2048-4M
18:55.17andrei_yes
18:56.02brlcadso for a 8M transfer, it flatlines around 3M packet size
18:56.38brlcadmore interestingly, is terrible after 16k or so to 1M
18:57.02brlcador is smaller good -- is that time?
18:57.28andrei_X axis is package size and Y Axis is speed ( 8mb / time )
18:58.41brlcadso yeah, does terrible for large packets
18:59.09brlcadhere's where it'll be interesting to combine the two graphs
18:59.25brlcadthere will undoubtedly be a "sweet spot"
18:59.57brlcadeither way, that looks MUCH better .. nice noise-free curve where you can really see some behavior going on
19:00.43andrei_the first script is on 95
19:00.52andrei_105*
19:01.05brlcadvery good
19:01.11brlcadfind out tomorrow :)
19:01.30andrei_I have some 1 - 95 data that I can process and draw some graph
19:01.34brlcadin the meantime, you could try to set up the next one (but not run it or it might mess up 1-2048's results)
19:01.45brlcadnext up, the full 3D case with different file/data sizes
19:01.52brlcadyeah, graph would be good
19:02.07brlcadfor different file/data sizes, I'd suggest just doing powers of 2
19:02.26brlcadfrom all the way down to one byte
19:02.40andrei_while I was searching I found something called Shanon's theorem
19:02.48andrei_should I investigate more in that aspect?
19:03.27andrei_from what I could have found it wasn't used for exactly this kind of tests
19:03.45andrei_from what I read*
19:03.47brlcad1 2 4 8 16 32 64 128 256 512 1024 2048 4k 8k 16k 32k 64k 128k 256k 512k 1M 2M 4M .. should be good for now
19:03.47andrei_sorry
19:04.27andrei_so I ll stop the file / package size at 4M?
19:04.57brlcadyou could read into shannon's theorem and nyquist limits, but it's not really relevant to this
19:05.30brlcadtcp takes care of data transfer noise and you're not dealing with a known value trying to characterize the wall-clock timing in a stable way
19:06.24brlcadyes, file size stops at 4M, but not the package sizes being tested are still 1-2048,2048-4M for all files
19:07.41brlcadso you basically just got half of the 2D graph for an 8M data transfer
19:08.09brlcadeventually, you'll have the whole dataset for 1-2048,2048-4M to complete the graph
19:08.33brlcadthen you're going to repeat that graph test for 1 2 4 8 16 32 64 128 256 512 1024 2048 4k 8k 16k 32k 64k 128k 256k 512k 1M 2M 4M data sets
19:09.15brlcadshould make a very useful 3D graph (or individually useful 2D graphs)
19:10.49andrei_I doubt the answer would be that simple, but shouldn't the " sweet" spot  be situated somewhere near the tcp payload ?
19:39.58andrei_brlcad, you were right, the graphic is exponential
19:40.19*** join/#brlcad jbschw (~jbschw@ool-4355ee10.dyn.optonline.net)
19:40.23andrei_here is the graphic : http://i.imgur.com/wQz08.png ( from 2 to 94 , apparently 1 is broken, will have to rerun it after script is done )
20:19.51*** join/#brlcad jbschw_ (~jbschw@ool-4355ee10.dyn.optonline.net)
20:30.58*** join/#brlcad jbschw (~jbschw@ool-4355ee10.dyn.optonline.net)
20:31.19CIA-23BRL-CAD: 03louipc * r51442 10/brlcad/trunk/misc/archlinux/brlcad.sh: archlinux: Remove path setting script, as it's generated in the PKGBUILD.
20:40.46CIA-23BRL-CAD: 03louipc * r51443 10/brlcad/trunk/misc/archlinux/brlcad.install: archlinux: Clean up old ld.so.conf, new configuration provided in separate file.
20:48.32*** join/#brlcad jbschw_ (~jbschw@ool-4355ee10.dyn.optonline.net)
21:09.06*** join/#brlcad jbschw (~jbschw@ool-4355ee10.dyn.optonline.net)
21:09.51*** join/#brlcad juanman (~quassel@unaffiliated/juanman)
21:35.16*** join/#brlcad jbschw_ (~jbschw@ool-4355ee10.dyn.optonline.net)
21:37.06CIA-23BRL-CAD: 03Cprecup 07http://brlcad.org * r0 10/wiki/Special:Log/upload: uploaded "[[Image:Graph link nodes no duplications.png]]": A second version of the links between a COMB object and its subtree nodes without node duplications.
21:40.45CIA-23BRL-CAD: 03Cprecup 07http://brlcad.org * r4116 10/wiki/User:Cprecup/GSoC2012_progress: 06-07/07/2012 - data stored in hash tables + node duplications in the graph eliminated
22:35.49*** join/#brlcad jbschw (~jbschw@ool-4355ee10.dyn.optonline.net)
22:36.49*** join/#brlcad jbschw (~jbschw@ool-4355ee10.dyn.optonline.net)
22:37.49*** part/#brlcad jbschw (~jbschw@ool-4355ee10.dyn.optonline.net)
23:19.57*** join/#brlcad juanman (~quassel@unaffiliated/juanman)
23:32.39*** join/#brlcad juanman (~quassel@unaffiliated/juanman)
23:41.46*** join/#brlcad starseek1r (~starseeke@BZ.BZFLAG.BZ)
23:45.42*** join/#brlcad brlcad (~sean@BZ.BZFLAG.BZ)

Generated by irclog2html.pl Modified by Tim Riker to work with infobot.