| 01:24.10 | *** join/#brlcad LordOfBikes_ (~armin@dslb-092-074-248-217.092.074.pools.vodafone-ip.de) |
| 01:36.25 | Notify | 03BRL-CAD:vasco_costa * 66116 brlcad/trunk/src/librt/bool.c: remove all gotos from rt_boolweave. |
| 01:37.10 | Notify | 03BRL-CAD Wiki:Ejno------ * 9486 /wiki/GCV_Plugin_Tutorial: Created page with "==Introduction== This tutorial will demonstrate the process of integrating conversion plugins into the Geometry Conversion Library (GCV). ==Build Integration== In order to ..." |
| 02:35.39 | *** join/#brlcad kintel (~kintel@unaffiliated/kintel) |
| 02:38.33 | Notify | 03BRL-CAD:vasco_costa * 66117 (brlcad/branches/opencl/HACKING brlcad/branches/opencl/doc/docbook/CMakeLists.txt and 44 others): merge from trunk. |
| 03:19.33 | Notify | 03BRL-CAD:vasco_costa * 66118 (brlcad/branches/opencl/include/rt/boolweave.h brlcad/branches/opencl/include/rt/region.h and 5 others): apply patch to replace rt_booleval and other performance critical functions to use a compiled postfix (RPN) bool tree. |
| 04:03.01 | *** join/#brlcad kintel (~kintel@unaffiliated/kintel) |
| 04:18.09 | *** join/#brlcad kintel (~kintel@unaffiliated/kintel) |
| 04:53.20 | *** join/#brlcad sofat (~androirc@101.213.15.36) |
| 05:14.30 | *** join/#brlcad kintel (~kintel@unaffiliated/kintel) |
| 06:29.32 | *** join/#brlcad Boquete (~piotr@bwa86.neoplus.adsl.tpnet.pl) |
| 06:52.00 | *** join/#brlcad ries (~ries@D979C47E.cm-3-2d.dynamic.ziggo.nl) |
| 08:52.45 | *** join/#brlcad merzo (~merzo@user-94-45-58-141.skif.com.ua) |
| 09:19.15 | *** join/#brlcad konrado (~konro@41.205.22.34) |
| 09:26.15 | *** join/#brlcad teepee-- (bc5c2134@gateway/web/freenode/ip.188.92.33.52) |
| 09:59.16 | *** join/#brlcad LordOfBikes (~armin@dslb-092-074-248-217.092.074.pools.vodafone-ip.de) |
| 10:51.17 | *** join/#brlcad ries_nicked (~ries@217.121.196.126) |
| 10:54.01 | *** join/#brlcad konrado (~konro@41.205.22.45) |
| 11:33.20 | *** join/#brlcad vasc (~vasc@bl12-1-27.dsl.telepac.pt) |
| 12:05.48 | vasc | brlcad, checked out the goto removal i did on src/librt/bool.c? |
| 12:06.16 | vasc | brlcad, i also have a patch to replace the rt_booleval function that i assigned to you |
| 12:08.32 | vasc | brlcad, afaik it didn't break anything or lose performance from the goto removal. at least from the limited testing i did. i don't have complex csg models to test with. |
| 12:09.13 | vasc | brlcad, but other than an extra branch or two an optimized building with compiler branch optimizations should compute the gotos 100% exact same |
| 13:34.31 | *** join/#brlcad d_rossberg (~rossberg@66-118-151-70.static.sagonet.net) |
| 13:40.18 | Notify | 03BRL-CAD:d_rossberg * 66119 brlcad/trunk/src/librt/primitives/bot/gct_decimation/auxiliary/mm.c: local variables need to be declared at the beginning of a block for C89/MSVC before 2013 |
| 13:49.54 | Notify | 03BRL-CAD:starseeker * 66120 brlcad/trunk/src/libbrep/shape_recognition_cylinder.cpp: Not sure if this is handled correctly, but make sure the axis and the planes line up for whole cylinders... |
| 13:56.53 | *** join/#brlcad sofat (~androirc@101.208.159.250) |
| 14:07.37 | *** join/#brlcad kintel (~kintel@unaffiliated/kintel) |
| 14:18.26 | Notify | 03BRL-CAD:starseeker * 66121 (brlcad/trunk/src/libbrep/tests/brep_csg_1.g brlcad/trunk/src/libbrep/tests/brep_csg_2.g): rename objects for convenience. |
| 14:28.03 | Notify | 03BRL-CAD:vasco_costa * 66122 brlcad/trunk/src/librt/primitives/primitive_util.c: simplify basic ocl material lookup code. |
| 14:41.10 | *** join/#brlcad Shubham (6719e766@gateway/web/freenode/ip.103.25.231.102) |
| 15:05.14 | *** join/#brlcad sofat_ (~androirc@101.216.174.25) |
| 15:19.51 | *** join/#brlcad kintel (~kintel@unaffiliated/kintel) |
| 15:20.48 | Notify | 03BRL-CAD:starseeker * 66123 brlcad/trunk/src/libbrep/shape_recognition_cylinder.cpp: If an arc is also a circle, need to handle a bit differently. |
| 15:23.30 | *** join/#brlcad ih8sum3r (~ih8sum3r@106.192.42.160) |
| 15:24.12 | ih8sum3r | ``Erik, brlcad : Is that kernel driver problem regarding Vbox solved? |
| 15:33.15 | Notify | 03BRL-CAD:brlcad * 66124 brlcad/trunk/src/librt/primitives/rec/rec.c: after much pinpointing, revert the specific portion of rweiss' change in r56016 that made perfectly slicing through the ends of a tgc/rec a miss. this subtle change causes geometry to be missed which cascades into different analysis results. more inspection warranted, but this restores behavior for a patch release. |
| 15:34.24 | Notify | 03BRL-CAD:brlcad * 66125 brlcad/trunk/src/librt/primitives/rec/rec.c: eliminate equality test |
| 15:43.34 | Notify | 03BRL-CAD:brlcad * 66126 brlcad/trunk/src/librt/primitives/rec/rec.c: looks like vshot is now terribly out of sync, so leave a cautionary note for future travelers. vshot() will probably go away wholesale as the new coherent pipeline comes online. |
| 15:46.33 | *** join/#brlcad berserker (~berserker@122.15.200.115) |
| 16:00.03 | Notify | 03BRL-CAD:starseeker * 66127 (brlcad/trunk/src/libbrep/shape_recognition_cylinder.cpp brlcad/trunk/src/libbrep/tests/brep_csg_2.g): More circle handling. |
| 16:00.56 | berserker | Hi I am new here and would like to contribute to BRL-CAD. Could someone help me out. |
| 16:22.49 | Notify | 03BRL-CAD:starseeker * 66128 brlcad/trunk/src/libbrep/shape_recognition_cone.cpp: More circle handling. |
| 16:23.22 | starseeker | berserker: howdy! |
| 16:23.27 | starseeker | what are your general interests? |
| 16:23.40 | starseeker | coding, documentation, website work, etc... |
| 16:23.51 | berserker | Hi I am interested in writing code |
| 16:25.26 | starseeker | what languages are you familiar with? |
| 16:32.03 | *** join/#brlcad Stragus (~alexis@modemcable090.29-19-135.mc.videotron.ca) |
| 16:36.53 | *** join/#brlcad Boquete (~piotr@83.27.242.191) |
| 16:50.32 | Notify | 03BRL-CAD:brlcad * 66129 brlcad/branches/RELEASE/src/librt/primitives/rec/rec.c: merge r66124 through r66126 which restores behavior for a subtle floating point issue in tgc/rec shotlining. not the end-state, but consistent behavior for patch release. |
| 16:56.17 | *** join/#brlcad yiyus (1242712427@je.je.je) |
| 17:10.29 | brlcad | vasc: I noticed and started yesterday, it'll take a while to test |
| 17:10.40 | brlcad | first simple test passed, so that was good ;) |
| 17:10.46 | *** join/#brlcad ih8sum3r_ (~ih8sum3r@223.225.158.67) |
| 17:11.54 | vasc | it took several tries to eliminate the gotos best way |
| 17:12.16 | vasc | make sure to test with -O or -O2 on or branch optimization won't be done by compiler and may be slower than gotos |
| 17:12.46 | vasc | i think -O does it but might need more to enable branch opts |
| 17:13.41 | vasc | modern optimizing C compilers usually choke on gotos a lot. |
| 17:13.46 | vasc | it disables a lot of optimizations |
| 17:14.51 | vasc | it makes it harder to define basic blocks |
| 17:15.09 | vasc | or to predict code flow even |
| 17:24.54 | vasc | an optimizing compiler turns those cascading ifs into direct jumps to end of cascade if you enter one block |
| 17:25.55 | vasc | i think it also makes it more easy to analyze the code |
| 17:26.31 | vasc | the number of partitions is between N and 1+2*(N-1) |
| 17:26.44 | vasc | where N is number of segments |
| 17:27.41 | vasc | i couldn't see that with the gotos |
| 17:28.43 | vasc | actually surprised because i expected max bound to be larger |
| 17:29.26 | vasc | i hope the performance is also to expectations |
| 17:30.15 | vasc | if you turn the compiler optimizations all on a modern compiler can analyze code like this a lot better than old one |
| 17:30.27 | Notify | 03BRL-CAD:brlcad * 66130 (brlcad/branches/STABLE/CMakeLists.txt brlcad/branches/STABLE/NEWS and 45 others): merge r61075 through r66074 from RELEASE to STABLE, in preparation for 7.24.4. this pulls datums, build fixes, and tgc/rec grazing behavior reversion. |
| 17:42.09 | *** join/#brlcad sofat (~androirc@101.214.216.126) |
| 17:46.58 | brlcad | on a big smp parallel test I did yesterday (64 cores iirc), there was essentially no difference unoptimized (which was a good sign) |
| 17:47.15 | brlcad | we'll have to explore the lower and higher end performance profile before this goes live |
| 17:47.37 | vasc | cool |
| 17:47.44 | brlcad | both optimized and un just to understand the impact |
| 17:47.59 | brlcad | but obviously optimized is what matters |
| 17:48.01 | vasc | you saw the RPN tree patch? |
| 17:48.34 | brlcad | saw it, but didn't inspect |
| 17:48.42 | vasc | https://sourceforge.net/p/brlcad/patches/417/ |
| 17:48.57 | vasc | this one would make future porting to opencl or some other architecture a lot easier as well |
| 17:49.03 | vasc | and it should be slightly faster. i hope. |
| 17:49.05 | vasc | not sure. |
| 17:49.23 | vasc | but this one might not be 100% formally equivalent to old code |
| 17:49.31 | vasc | so this is why i did not commit to svn trunk like other |
| 17:50.13 | vasc | the major concern is how it handles XOR and its little friends |
| 17:50.24 | vasc | the current code for XOR is seriously too strange for me to contemplate. |
| 17:50.30 | vasc | i put my version in that patch. |
| 17:50.42 | vasc | in fact i tried doing a xor in mged and it failed |
| 17:50.50 | vasc | it didn't accept the comb |
| 17:51.06 | vasc | c sph1 ^ sph2 and blam it crashed |
| 17:51.14 | vasc | c sph1 + sph2 works fine |
| 17:51.49 | vasc | well c csg sph1 ^ sph2 |
| 17:51.57 | vasc | and c csg sph1 + sph2 |
| 17:52.03 | vasc | something like that |
| 17:57.35 | *** join/#brlcad kintel (~kintel@unaffiliated/kintel) |
| 17:58.05 | Notify | 03BRL-CAD:vasco_costa * 66131 (brlcad/trunk/src/librt/primitives/arb8/arb8_shot.cl brlcad/trunk/src/librt/primitives/bot/bot_shot.cl and 9 others): process ocl segments instead of hit points. use registers to store segments in single pass. new single pass lightmodes: "full" and "multi-hit" transparent. |
| 18:01.36 | *** join/#brlcad sofat (~androirc@101.214.216.126) |
| 18:04.17 | *** join/#brlcad sofat (~androirc@101.214.216.126) |
| 19:13.42 | *** join/#brlcad Nicop06 (~nicolas@htr06-1-82-227-229-91.fbx.proxad.net) |
| 19:18.11 | *** join/#brlcad Nicop06 (~nicolas@htr06-1-82-227-229-91.fbx.proxad.net) |
| 19:27.57 | *** mode/#brlcad [+o brlcad] by ChanServ |
| 19:28.54 | *** join/#brlcad vasc (~vasc@bl12-1-27.dsl.telepac.pt) |
| 19:32.17 | *** join/#brlcad ries (~ries@D979C47E.cm-3-2d.dynamic.ziggo.nl) |
| 19:38.42 | vasc | back |
| 19:45.22 | *** join/#brlcad konrado (~konro@41.205.22.2) |
| 20:45.08 | Notify | 03BRL-CAD:vasco_costa * 66132 (brlcad/trunk/include/rt/shoot.h brlcad/trunk/src/librt/primitives/common.cl and 2 others): updated multiple-kernel launch renderer to work with whole segment lists. |
| 20:53.17 | Notify | 03BRL-CAD Wiki:Vasco.costa * 9487 /wiki/User:Vasco.costa/GSoC15/logs: |
| 21:20.18 | Notify | 03BRL-CAD Wiki:Vasco.costa * 9488 /wiki/User:Vasco.costa/GSoC15/logs: /* Post Development Phase */ |
| 21:20.48 | Notify | 03BRL-CAD Wiki:Vasco.costa * 9489 /wiki/User:Vasco.costa/GSoC15/logs: /* Week: 24-30 Aug */ |
| 21:21.07 | Notify | 03BRL-CAD Wiki:Vasco.costa * 9490 /wiki/User:Vasco.costa/GSoC15/logs: /* Week: 24-30 Aug */ |
| 21:21.46 | Notify | 03BRL-CAD Wiki:Vasco.costa * 9491 /wiki/User:Vasco.costa/GSoC15/logs: /* Week: 24-30 Aug */ |
| 21:24.31 | Notify | 03BRL-CAD Wiki:Vasco.costa * 9492 /wiki/User:Vasco.costa/GSoC15/logs: /* Week: 24-30 Aug */ |
| 21:25.29 | Notify | 03BRL-CAD Wiki:Vasco.costa * 9493 /wiki/User:Vasco.costa/GSoC15/logs: /* Week: 24-30 Aug */ |
| 21:34.00 | Notify | 03BRL-CAD Wiki:Vasco.costa * 9494 /wiki/User:Vasco.costa/GSoC15/logs: /* Week: 24-30 Aug */ |
| 21:34.54 | Notify | 03BRL-CAD Wiki:Vasco.costa * 9495 /wiki/User:Vasco.costa/GSoC15/logs: /* Week: 24-30 Aug */ |
| 21:35.18 | Notify | 03BRL-CAD Wiki:Vasco.costa * 9496 /wiki/User:Vasco.costa/GSoC15/logs: /* Week: 24-30 Aug */ |
| 21:37.42 | Notify | 03BRL-CAD Wiki:Vasco.costa * 9497 /wiki/User:Vasco.costa/GSoC15/logs: /* Week: 24-30 Aug */ |
| 21:40.43 | Notify | 03BRL-CAD Wiki:Vasco.costa * 9498 /wiki/User:Vasco.costa/GSoC15/logs: /* Week: 24-30 Aug */ |
| 21:42.37 | Notify | 03BRL-CAD Wiki:Vasco.costa * 9499 /wiki/User:Vasco.costa/GSoC15/logs: /* Week: 24-30 Aug */ |
| 21:44.53 | Notify | 03BRL-CAD:brlcad * 66133 brlcad/trunk/src/librt/primitives/rec/rec.c: this fixes where the previous attempt at eliminating a floating point equality test went wrong. '1.0' equals '1.0 + epsilon' for really tiny values, like our SMALL_FASTF. that meant we lost the equals case (creating an open set) and got incorrect grazing behavior on tgc. you can do comparisons with values that small but not math, so here we shift |
| 21:44.55 | Notify | the test and get the correct behavior. on quick review, a lot of places throughout the code are getting this wrong. |
| 21:44.57 | Notify | ... |
| 21:53.21 | Notify | 03BRL-CAD:vasco_costa * 66134 (brlcad/branches/opencl/include/rt/shoot.h brlcad/branches/opencl/src/libbrep/shape_recognition_cone.cpp and 19 others): merge changes from trunk. |
| 21:54.24 | *** join/#brlcad ries (~ries@D979C47E.cm-3-2d.dynamic.ziggo.nl) |
| 21:55.02 | Notify | 03BRL-CAD:brlcad * 66135 brlcad/branches/RELEASE/src/librt/primitives/rec/rec.c: merge c66133 from trunk |
| 22:21.25 | Notify | 03BRL-CAD:vasco_costa * 66137 brlcad/branches/opencl/src/librt/primitives/primitive_util.c: fix ocl material colors for good. |
| 22:21.48 | Notify | 03BRL-CAD:brlcad * 66136 (brlcad/branches/STABLE/src/librt/primitives/rec/rec.c Property Changed: and 2 others): merge c66135 from RELEASE so we get the right LOS values when grazing through an endcap |
| 22:22.43 | Notify | 03BRL-CAD:brlcad * 66138 (svn:mergeinfo ## -1,3 +1,3 ## and 11 others): oops, don't forget mergeinfoProperty Changed:----------------brlcad/branches/RELEASE/ |
| 22:25.27 | Notify | 03BRL-CAD:vasco_costa * 66139 (brlcad/trunk/src/librt/primitives/primitive_util.c Property Changed: and 2 others): fix ocl solid material colors. for good i hope. |
| 22:26.36 | Notify | 03BRL-CAD Wiki:Vasco.costa * 9500 /wiki/User:Vasco.costa/GSoC15/logs: /* Week: 24-30 Aug */ |
| 22:26.55 | Notify | 03BRL-CAD Wiki:Vasco.costa * 9501 /wiki/User:Vasco.costa/GSoC15/logs: /* Week: 24-30 Aug */ |