| 00:00.31 | vasc | overgeneralization is the root of all evil. so hardcoding it is. |
| 00:00.49 | starseeker | thought it was premature optimization :-P |
| 00:01.38 | vasc | sure |
| 00:01.51 | vasc | but this maxim i said is to prevent the second system effect. |
| 00:03.18 | starseeker | nods |
| 00:07.59 | ``Erik | heh, sf borked since '01? sounds about right O:-) |
| 00:08.45 | ``Erik | if foobar isn't adequate, what about qix? |
| 00:09.32 | Stragus | vasc, is scene preparation even taking so long to be worth building on GPU? |
| 00:10.00 | vasc | well not with the scenes i'm testing with |
| 00:10.01 | Stragus | I have no idea really, but I would assume CSG scene prep to be a lot faster than for meshes with millions of triangles |
| 00:10.38 | vasc | its just that i plan to reuse this for the bot |
| 00:10.42 | vasc | later |
| 00:19.17 | vasc | with the test scenes i've seen so far we wouldn't even need a spatial acceleration structure at all... |
| 00:20.01 | vasc | at most they have like two dozen solids in them |
| 00:21.15 | vasc | i probably goofed up something in the traversal code |
| 00:21.28 | vasc | i'm getting all these messages about OVERLAP of whatever in the mged console |
| 00:21.29 | Stragus | I would rather suggest to focus on the traversal and raytracing |
| 00:21.34 | Stragus | Hum. :) |
| 00:21.52 | vasc | i'm not checking if the intersection with the primitive is actually inside the cell |
| 00:21.55 | vasc | that's probably it |
| 00:22.25 | vasc | man this is slow like molasses |
| 00:22.34 | vasc | all these opencl kernel calls |
| 00:22.42 | vasc | i'm testing with the operators.g |
| 00:22.50 | vasc | it has a bunch of spheres and cones |
| 00:22.55 | vasc | these are accelerated in opencl |
| 00:23.00 | vasc | or should i say decelerated |
| 00:23.36 | vasc | its little better than watching paint dry |
| 00:23.57 | vasc | well whatever |
| 00:24.43 | Stragus | Ahah |
| 00:25.25 | Stragus | brlcad said the OpenCL kernels for each object type was to test the validity/accuracy of the results, not real use |
| 00:25.56 | Stragus | It really should be a single OpenCL kernel launch to trace a million rays |
| 00:27.57 | vasc | yes |
| 00:28.08 | vasc | it should but its not gonna be easy to refactor it into that |
| 00:28.31 | vasc | that's what i'm working on |
| 00:28.39 | Stragus | :) Right, it doesn't look easy indeed |
| 00:28.45 | vasc | dunno if i'll finish though |
| 00:28.51 | vasc | i hope so |
| 00:29.07 | Stragus | If you can just get the traversal and a few objects, I think that would be a great start |
| 00:29.13 | Stragus | Scene building on GPU can wait |
| 00:32.36 | vasc | i'm nearly done with it already |
| 00:32.48 | vasc | i just need to get this prefix sum working |
| 00:32.56 | vasc | the grid bit at least |
| 00:33.08 | vasc | the solid database is gonna take more time |
| 00:47.17 | vasc | hmm |
| 00:51.15 | vasc | oh snap |
| 00:51.29 | vasc | i did an exclusive scan and it's supposed to be an inclusive scan |
| 00:51.29 | vasc | duh |
| 00:54.24 | vasc | bingo. done. |
| 00:56.01 | vasc | cleanup time |
| 01:02.41 | vasc | i'll do the error checking some other day |
| 01:03.54 | vasc | it seems to build the grid on the gpu fine |
| 01:10.50 | Notify | 03BRL-CAD Wiki:85.246.107.116 * 9044 /wiki/User:Vasco.costa/GSoC15/logs: |
| 01:11.18 | vasc | i'll debug the traversal tomorrow |
| 01:11.33 | vasc | oh its the weekend |
| 01:11.34 | vasc | well |
| 01:13.28 | vasc | the plan said GPU solid database next |
| 01:13.42 | vasc | i guess i'll do that then |
| 01:13.51 | vasc | but its too late today |
| 01:15.29 | vasc | i guess i can do the api design |
| 01:18.08 | vasc | done. have a nice weekend folks. |
| 02:41.05 | *** join/#brlcad infobot (ibot@69-58-76-73.ut.vivintwireless.net) | |
| 02:41.05 | *** topic/#brlcad is BRL-CAD || http://brlcad.org || logs: http://ibot.rikers.org/%23brlcad/ || Congrats to all GCI 2014 winners Peter & Marc! || Congratulations to our 12 GSoC students! || Don't ask if someone is here, just ask your questions and wait for a response. ;-) | |
| 05:01.54 | *** join/#brlcad bhollister2 (~brad@2600:1010:b065:7fe:6923:a7ad:7815:65a2) | |
| 11:30.08 | *** join/#brlcad cox (~quassel@188.226.208.53) | |
| 12:35.36 | *** join/#brlcad ries_nicked (~ries@D979C47E.cm-3-2d.dynamic.ziggo.nl) | |
| 13:47.16 | *** join/#brlcad sofat (~sofat@202.164.45.212) | |
| 14:32.19 | *** join/#brlcad sofat (~sofat@202.164.45.212) | |
| 14:45.45 | *** join/#brlcad sofat (~sofat@202.164.45.212) | |
| 16:04.02 | *** join/#brlcad Izakey_ (~Izakey@154.70.98.151) | |
| 16:05.19 | *** join/#brlcad bhollister2 (~brad@c-73-189-140-51.hsd1.ca.comcast.net) | |
| 16:07.46 | *** join/#brlcad Izakey_ (~Izakey@154.70.98.151) | |
| 16:43.19 | *** join/#brlcad sofat (~sofat@202.164.45.212) | |
| 17:12.48 | sofat | brlcad, hello |
| 17:13.01 | sofat | I need some discussion |
| 17:13.12 | sofat | if you free please reply |
| 18:34.29 | *** join/#brlcad konrado (~konro@41.205.22.39) | |
| 18:38.48 | *** join/#brlcad ries_nicked (~ries@D979C47E.cm-3-2d.dynamic.ziggo.nl) | |
| 19:07.44 | *** join/#brlcad Izakey (~Isaac@41.205.22.55) | |
| 19:27.02 | Notify | 03BRL-CAD Wiki:Konrado DJ * 9045 /wiki/User:Konrado_DJ/GSoc2015/logs: /* 18 JULY 2015 */ |
| 20:12.14 | *** join/#brlcad sofat (~sofat@202.164.45.204) | |