| 01:21.07 | *** join/#brlcad jbschw (~jbschw@ool-4355ee10.dyn.optonline.net) | |
| 03:09.49 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r4065 10/wiki/User:Plussai/GSoC_2012_log: /* 7 May 2012 */ |
| 03:16.51 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r4066 10/wiki/User:Plussai/GSoC_2012_log: /* 23 May 2012 */ |
| 03:17.33 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r4067 10/wiki/User:Plussai/GSoC_2012_log: /* 21 May 2012 */ |
| 03:20.03 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r4068 10/wiki/User:Plussai/GSoC_2012_log: /* ''Week 1'' */ |
| 03:22.55 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r4069 10/wiki/User:Plussai/GSoC_2012_log: /* ''Week 1'' */ |
| 03:23.35 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r4070 10/wiki/User:Plussai/GSoC_2012_log: /* ''Week 1'' */ |
| 03:24.16 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r4071 10/wiki/User:Plussai/GSoC_2012_log: /* 29 May 2012 */ |
| 03:24.55 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r4072 10/wiki/User:Plussai/GSoC_2012_log: /* 31 May 2012 */ |
| 03:25.25 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r4073 10/wiki/User:Plussai/GSoC_2012_log: /* ''Week 2'' */ |
| 03:26.19 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r4074 10/wiki/User:Plussai/GSoC_2012_log: /* 4 June 2012 */ |
| 03:26.49 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r4075 10/wiki/User:Plussai/GSoC_2012_log: /* 12 June 2012 */ |
| 03:27.29 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r4076 10/wiki/User:Plussai/GSoC_2012_log: /* 18 June 2014 */ |
| 03:27.59 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r4077 10/wiki/User:Plussai/GSoC_2012_log: /* 26 June 2012 */ |
| 03:28.40 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r4078 10/wiki/User:Plussai/GSoC_2012_log: /* 2 July 2012 */ |
| 03:35.02 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r0 10/wiki/Special:Log/upload: uploaded "[[Image:Ell1.png]]" |
| 03:36.33 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r4080 10/wiki/User:Plussai/GSoC_2012_log: /* 4 July 2012 */ |
| 03:37.29 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r0 10/wiki/Special:Log/upload: uploaded "[[Image:Ell2.png]]" |
| 03:38.22 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r0 10/wiki/Special:Log/upload: uploaded "[[Image:ELL3.png]]" |
| 03:39.09 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r4083 10/wiki/User:Plussai/GSoC_2012_log: /* 6 July 2012 */ |
| 03:39.51 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r0 10/wiki/Special:Log/upload: uploaded "[[Image:Ell3.png]]" |
| 03:41.39 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r4085 10/wiki/User:Plussai/GSoC_2012_log: /* 6 July 2012 */ |
| 03:44.23 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r0 10/wiki/Special:Log/upload: uploaded a new version of "[[Image:Ell1.png]]" |
| 03:45.07 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r4087 10/wiki/User:Plussai/GSoC_2012_log: /* 6 July 2012 */ |
| 03:46.40 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r0 10/wiki/Special:Log/upload: uploaded a new version of "[[Image:Ell2.png]]" |
| 03:50.17 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r0 10/wiki/Special:Log/upload: uploaded a new version of "[[Image:Ell3.png]]" |
| 03:51.37 | brlcad | andrei_: try removing ssh from the script .. we're testing the behavior of libpkg, not your network setup |
| 03:51.47 | brlcad | that might give you some better performance |
| 03:54.03 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r0 10/wiki/Special:Log/upload: uploaded a new version of "[[Image:Ell3.png]]" |
| 03:54.51 | brlcad | also, 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 |
| 03:55.41 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r4091 10/wiki/User:Plussai/GSoC_2012_log: /* 6 July 2012 */ |
| 03:55.49 | brlcad | if it's longer, something is probabably wrong .. that's transferring 16GB, even 1byte at a time it shouldnt' be too bad |
| 03:56.06 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r4092 10/wiki/User:Plussai/GSoC_2012_log: /* 6 July 2012 */ |
| 03:56.52 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r4093 10/wiki/User:Plussai/GSoC_2012_log: /* 6 July 2012 */ |
| 03:56.53 | brlcad | andrei_: 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 |
| 03:57.41 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r4094 10/wiki/User:Plussai/GSoC_2012_log: /* 6 July 2012 */ |
| 03:57.47 | brlcad | if it's going to be more than 24 hours, just drop the file size down to 4096 bytes instead of 8mb |
| 03:58.37 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r4095 10/wiki/User:Plussai/GSoC_2012_log: /* 6 July 2012 */ |
| 03:59.35 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r0 10/wiki/Special:Log/upload: uploaded "[[Image:Ell4.png]]" |
| 04:00.53 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r4097 10/wiki/User:Plussai/GSoC_2012_log: /* 6 July 2012 */ |
| 04:01.11 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r4098 10/wiki/User:Plussai/GSoC_2012_log: /* 6 July 2012 */ |
| 04:01.54 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r4099 10/wiki/User:Plussai/GSoC_2012_log: /* 6 July 2012 */ |
| 04:02.22 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r4100 10/wiki/User:Plussai/GSoC_2012_log: /* 6 July 2012 */ |
| 04:03.28 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r4101 10/wiki/User:Plussai/GSoC_2012_log: /* 6 July 2012 */ |
| 04:04.04 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r0 10/wiki/Special:Log/upload: uploaded "[[Image:HYP1.png]]" |
| 04:41.29 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r0 10/wiki/Special:Log/upload: uploaded a new version of "[[Image:HYP1.png]]" |
| 04:43.26 | andrei_ | it should take 15 min. |
| 04:45.03 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r0 10/wiki/Special:Log/upload: uploaded a new version of "[[Image:HYP1.png]]" |
| 04:47.47 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r0 10/wiki/Special:Log/upload: uploaded a new version of "[[Image:HYP1.png]]" |
| 04:48.21 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r0 10/wiki/Special:Log/upload: uploaded "[[Image:HYP2.png]]" |
| 04:49.11 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r4107 10/wiki/User:Plussai/GSoC_2012_log: /* 6 July 2012 */ |
| 04:49.27 | CIA-50 | BRL-CAD: 03Plussai 07http://brlcad.org * r0 10/wiki/Special:Log/upload: uploaded "[[Image:HYP3.png]]" |
| 04:52.15 | CIA-50 | BRL-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) | |
| 11:06.30 | CIA-23 | BRL-CAD: 03Phoenix 07http://brlcad.org * r4110 10/wiki/User:Phoenix/GSoc2012/Reports: /* Week 7 */ |
| 11:32.49 | CIA-23 | BRL-CAD: 03Phoenix 07http://brlcad.org * r4111 10/wiki/User:Phoenix/GSoc2012/Reports: /* Week 7 */ |
| 11:33.33 | CIA-23 | BRL-CAD: 03Phoenix 07http://brlcad.org * r4112 10/wiki/User:Phoenix/GSoc2012/Reports: /* Week 7 */ |
| 11:34.30 | CIA-23 | BRL-CAD: 03Phoenix 07http://brlcad.org * r4113 10/wiki/User:Phoenix/GSoc2012/Reports: /* Week 6 */ |
| 11:34.48 | CIA-23 | BRL-CAD: 03Phoenix 07http://brlcad.org * r4114 10/wiki/User:Phoenix/GSoc2012/Reports: /* Week 7 */ |
| 13:26.09 | brlcad | andrei_: great, then can't wait to see the graph :) |
| 13:29.31 | andrei_ | hm |
| 13:29.43 | andrei_ | 1024 took 0.5 seconds |
| 13:30.00 | andrei_ | I have been running the script for approximatively 8 hours |
| 13:30.06 | andrei_ | it currently is on 35 bytes |
| 13:30.34 | andrei_ | running the 2048 => 2048 * 2048 aswell. That one is like 70% |
| 13:34.12 | andrei_ | I want to stop the small one at some point ( probably at 100 bytes) |
| 13:34.33 | andrei_ | and then increase distance between tests up to 100 |
| 13:35.57 | andrei_ | does this seem like a good ide? |
| 13:35.59 | andrei_ | 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) | |
| 15:47.23 | brlcad | andrei_: actually testing every byte between 1 and 2048 is intentional to see if there are any anamolies |
| 16:21.35 | andrei_ | ah, it s at 65 now. The good part is it speeds up as it goes. |
| 16:28.30 | andrei_ | the "large" one is almost done tho. |
| 16:31.07 | andrei_ | 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) | |
| 17:34.56 | brlcad | andrei_: at the rate it's increasing, it should be done really soon |
| 17:35.06 | brlcad | apparently non-linear |
| 17:35.16 | brlcad | see, that's actually rather informative |
| 17:36.01 | andrei_ | the large one is done. Currently I m processing info to be used by octave |
| 17:36.37 | brlcad | not 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) |
| 17:36.59 | brlcad | the order they're done certainly really matter |
| 17:37.05 | brlcad | certainly *doesn't* matter |
| 17:37.27 | andrei_ | the 2048-4194304 is done |
| 17:37.55 | andrei_ | I m " drawing " the graph soon |
| 17:38.49 | brlcad | awesome |
| 17:39.16 | andrei_ | and the small one hit 80 |
| 17:39.36 | brlcad | so maybe 2-3 hours more |
| 17:39.56 | brlcad | can you graph what you already have so far with 1-80? |
| 17:40.14 | andrei_ | I should be able to, yes. |
| 17:47.59 | andrei_ | for some really wierd reason netbeans refuses to open |
| 18:26.01 | andrei_ | brlcad, I have the graphic for 2048 -> 2048^2 |
| 18:26.15 | andrei_ | it looks linear |
| 18:26.24 | andrei_ | uploading a picture now |
| 18:29.50 | andrei_ | http://imgur.com/Ot7vb |
| 18:29.58 | andrei_ | Sorry for not adding legend to axis |
| 18:30.38 | andrei_ | no, I made a mistake. Ignore that, please. |
| 18:36.43 | andrei_ | this is the valid graphic : http://i.imgur.com/H5Q2u.jpg |
| 18:37.10 | andrei_ | on X axis is package size ( 2048 -> 2048 * 2048 ) |
| 18:37.16 | andrei_ | on y is speed. |
| 18:41.52 | andrei_ | 1 - 2048 hit 100. |
| 18:41.59 | andrei_ | how does that graphic look? |
| 18:52.27 | CIA-23 | BRL-CAD: 03brlcad * r51441 10/brlcad/trunk/TODO: looks like the rendering of part may be wrong when the pill squishes in on itself. |
| 18:53.51 | brlcad | x is package size?? |
| 18:54.01 | brlcad | package size should be 1-2048... |
| 18:54.02 | andrei_ | yes |
| 18:54.05 | andrei_ | no no |
| 18:54.17 | andrei_ | I said that s the graphic for 2048 - > 2048 * 2048 |
| 18:54.54 | brlcad | ah, so 2048-4M |
| 18:55.17 | andrei_ | yes |
| 18:56.02 | brlcad | so for a 8M transfer, it flatlines around 3M packet size |
| 18:56.38 | brlcad | more interestingly, is terrible after 16k or so to 1M |
| 18:57.02 | brlcad | or is smaller good -- is that time? |
| 18:57.28 | andrei_ | X axis is package size and Y Axis is speed ( 8mb / time ) |
| 18:58.41 | brlcad | so yeah, does terrible for large packets |
| 18:59.09 | brlcad | here's where it'll be interesting to combine the two graphs |
| 18:59.25 | brlcad | there will undoubtedly be a "sweet spot" |
| 18:59.57 | brlcad | either way, that looks MUCH better .. nice noise-free curve where you can really see some behavior going on |
| 19:00.43 | andrei_ | the first script is on 95 |
| 19:00.52 | andrei_ | 105* |
| 19:01.05 | brlcad | very good |
| 19:01.11 | brlcad | find out tomorrow :) |
| 19:01.30 | andrei_ | I have some 1 - 95 data that I can process and draw some graph |
| 19:01.34 | brlcad | in the meantime, you could try to set up the next one (but not run it or it might mess up 1-2048's results) |
| 19:01.45 | brlcad | next up, the full 3D case with different file/data sizes |
| 19:01.52 | brlcad | yeah, graph would be good |
| 19:02.07 | brlcad | for different file/data sizes, I'd suggest just doing powers of 2 |
| 19:02.26 | brlcad | from all the way down to one byte |
| 19:02.40 | andrei_ | while I was searching I found something called Shanon's theorem |
| 19:02.48 | andrei_ | should I investigate more in that aspect? |
| 19:03.27 | andrei_ | from what I could have found it wasn't used for exactly this kind of tests |
| 19:03.45 | andrei_ | from what I read* |
| 19:03.47 | brlcad | 1 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 |
| 19:03.47 | andrei_ | sorry |
| 19:04.27 | andrei_ | so I ll stop the file / package size at 4M? |
| 19:04.57 | brlcad | you could read into shannon's theorem and nyquist limits, but it's not really relevant to this |
| 19:05.30 | brlcad | tcp 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 |
| 19:06.24 | brlcad | yes, file size stops at 4M, but not the package sizes being tested are still 1-2048,2048-4M for all files |
| 19:07.41 | brlcad | so you basically just got half of the 2D graph for an 8M data transfer |
| 19:08.09 | brlcad | eventually, you'll have the whole dataset for 1-2048,2048-4M to complete the graph |
| 19:08.33 | brlcad | then 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 |
| 19:09.15 | brlcad | should make a very useful 3D graph (or individually useful 2D graphs) |
| 19:10.49 | andrei_ | I doubt the answer would be that simple, but shouldn't the " sweet" spot be situated somewhere near the tcp payload ? |
| 19:39.58 | andrei_ | brlcad, you were right, the graphic is exponential |
| 19:40.19 | *** join/#brlcad jbschw (~jbschw@ool-4355ee10.dyn.optonline.net) | |
| 19:40.23 | andrei_ | 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) | |
| 20:31.19 | CIA-23 | BRL-CAD: 03louipc * r51442 10/brlcad/trunk/misc/archlinux/brlcad.sh: archlinux: Remove path setting script, as it's generated in the PKGBUILD. |
| 20:40.46 | CIA-23 | BRL-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) | |
| 21:37.06 | CIA-23 | BRL-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. |
| 21:40.45 | CIA-23 | BRL-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) | |
