00:04.20 | IriX64 | wish you could see my screen, I attached to X and it came up and havoc is there, but ogl still gives me a segmentation fault when i attach to it, sigh. |
00:06.02 | IriX64 | hah it rt'ed |
00:06.38 | IriX64 | where does it go? nothing on the graphics screen. |
00:19.54 | IriX64 | ill get back to you. |
00:23.46 | ``Erik | heh, I tend to ignore ogl and use Xl ... |
00:24.02 | IriX64 | man it....should work |
00:24.25 | ``Erik | on modern equipment, X stomps the living crap out of ogl for raytrace results... |
00:24.45 | IriX64 | based on what hardware? |
00:25.22 | IriX64 | my geForce runs at 1g |
00:26.10 | IriX64 | well 1.2 but the screen can't keep up (flickers) |
00:30.49 | IriX64 | what did you guys change? |
00:30.59 | IriX64 | in 7.6.2 it comes up fine. |
00:38.38 | deltazap | IriX64: is that apache one of the 'hush hush' files or an example file? |
00:40.32 | IriX64 | huh? just something i was compiling |
00:40.42 | deltazap | ah |
00:41.05 | IriX64 | whats a hush hush file? |
00:42.07 | deltazap | well, i hadn't seen that before |
00:42.17 | IriX64 | ah |
00:42.25 | deltazap | but then again, i still have difficulty looking at the included examples :P |
00:43.00 | IriX64 | as they told me, the usual method is to do a tops to examine the database, then do an e "something" |
00:48.28 | deltazap | oh great, i need to middle click in this tutorial :( |
00:48.37 | deltazap | i've only have one button |
00:49.07 | deltazap | :P |
00:50.50 | IriX64 | cut it in two :) |
00:51.03 | IriX64 | err 3 |
00:51.40 | deltazap | :P |
00:51.40 | IriX64 | which tutorial are you at? |
00:51.49 | deltazap | lesson 6 |
00:51.54 | deltazap | i just started the other night |
00:52.48 | deltazap | oh, volume 2 |
00:52.56 | IriX64 | what does that one show you how to produce (im way behind) |
00:53.01 | deltazap | the goblet |
00:53.30 | IriX64 | be sure to specify bill of materials that Gold must be used :P |
00:53.50 | deltazap | haha |
00:55.04 | deltazap | after working with mged for a while, i realized tab completion would be nice :) |
00:55.44 | IriX64 | put it in ... be a contributor (unlike me) |
00:56.11 | deltazap | i'd have no clue how to implement that |
00:56.30 | IriX64 | examine tops20 |
00:57.09 | IriX64 | up arrow works tho |
00:58.02 | deltazap | if i make an object (primative), how can i change it's properties later |
01:08.54 | ``Erik | one button? a ma |
01:08.55 | ``Erik | mac? |
01:09.38 | ``Erik | every bit of ARL data irix64 has access to is an "open source", nothing "hush hush" |
01:10.23 | deltazap | ``Erik: yes |
01:11.35 | deltazap | i'm guessing the little bit of problems that i have with some of the compositing of the framebuffer with the wireframe is because of apple's x11 as well |
01:12.35 | ``Erik | the ctl/option/apple keys modify clicks so you can get 3 button activity with a 1 button mouse |
01:13.22 | ``Erik | oddly enough, mac is probably the MOST supported os... the 4 heaviest developers at the moment all use macs as their destop |
01:13.41 | deltazap | i kept trying ctrl-click, but t kept zooming out on me so i thought it wasn't working in x11 |
01:13.44 | deltazap | oh, that's funny :) |
01:14.02 | ``Erik | apple+click will zoom in option+click will rotate |
01:14.05 | ``Erik | iirc |
03:21.33 | brlcad | there are no hush hush files floating around anywhere |
03:22.05 | brlcad | no geometry, no images. period. |
03:22.48 | brlcad | deltazap: tab-completion would be nice .. i.e. usual readline support |
03:23.22 | brlcad | you can edit geometry after creating by a couple ways.. there's the Edit->primitive selection option in the menu as well as several ways on the command line |
03:23.57 | deltazap | ah, ok |
03:24.05 | deltazap | one more question |
03:24.17 | deltazap | how can i render an image so i can use it elsewhere |
03:31.07 | brlcad | deltazap: on the mged command line, rt -sSIZE -o filename.pix or on the raytrace control panel under "Destination" you can put a filename |
03:31.28 | deltazap | ah, i see :) |
03:31.53 | brlcad | pix is a raw image format, not useful in most image tools, so you'd then want to run pix-png outside mged (e.g. pix-png myimage.pix > myimage.png) |
03:31.54 | deltazap | and i just found the export RT Script |
03:32.10 | deltazap | i've messed with pix-png once before |
03:32.27 | brlcad | er, pix-png -sSIZE myimage.pix > myimage.png |
03:33.08 | brlcad | when/if you set a size, the size has to be used on all the pix* tools, else tool defaults to 512x512 |
03:33.32 | deltazap | ah, ok |
03:33.46 | brlcad | fyi by default, rt will make pix files read-only so you have to rm -f filename to delete or chmod them |
03:46.12 | deltazap | ah, thanks :) |
03:46.48 | brlcad | (common newbie issue) ;) |
08:23.45 | *** join/#brlcad clock_ (n=clock@zux221-122-143.adsl.green.ch) |
10:34.07 | *** join/#brlcad Twingy (n=justin@74.92.144.217) [NETSPLIT VICTIM] |
10:55.50 | CIA-7 | BRL-CAD: 03d_rossberg * 10brlcad/include/config_win.h: POSIX's fmax added |
10:58.09 | CIA-7 | BRL-CAD: 03d_rossberg * 10brlcad/src/librt/g_brep.cpp: missing return statement added |
11:07.25 | ``Erik | blargh |
11:11.24 | *** join/#brlcad jack- (i=jack@dub.minimal.nl) |
14:23.20 | CIA-7 | BRL-CAD: 03erikgreenwald * 10brlcad/src/librt/g_metaball.c: disable the "big bounding sphere" in plot as it confuses users. |
14:24.06 | *** join/#brlcad Elperion (n=Elperion@p54874ee6.dip.t-dialin.net) |
14:52.25 | CIA-7 | BRL-CAD: 03erikgreenwald * 10brlcad/src/mged/edsol.c: |
14:52.25 | CIA-7 | BRL-CAD: Fix bug that caused a segfault when the last metaball point was deleted. |
14:52.25 | CIA-7 | BRL-CAD: Fixed display when no point is selected. Removed debugging bu_log. |
15:35.45 | CIA-7 | BRL-CAD: 03erikgreenwald * 10brlcad/src/mged/typein.c: fix an off by one error in metaball prompting. |
17:07.33 | *** join/#brlcad IriX64 (n=mario_du@bas3-sudbury98-1168048679.dsl.bell.ca) |
17:16.51 | IriX64 | http://www.pastebin.ca/397655 benchmark, anybody interested? |
17:20.40 | brlcad | those numbers are looking much better |
17:29.05 | Maloeran | What's the hardware? |
17:33.22 | IriX64 | amd64 |
17:33.30 | IriX64 | model 3800+ |
17:33.37 | IriX64 | 1g mem |
17:35.29 | Maloeran | My 8 cores Xeon only seems to be 3.5 times faster, though I guess it does not scale too well |
17:36.43 | Maloeran | Oh the 3800+ is a X2 of course, dual-core |
17:38.23 | Maloeran | Actually, my build is not optimized |
17:38.51 | IriX64 | my is but debug is in there |
17:40.21 | IriX64 | i did nothing to your data, all is as you gave me. |
17:40.35 | brlcad | Maloeran: not the same comparison either |
17:41.05 | Maloeran | brlcad, it depends what you mean by "shader". It's presently an user callback |
17:41.14 | brlcad | the benchmark images have other characteristics like scene-lights, various shaders, recursion, that all affect the rtfm number |
17:41.51 | brlcad | i mean like for actual image generation -- rfdemo level support for making pretty pictures with various settable material properties |
17:42.04 | brlcad | not just the libraries capacity to have them, but their existence :) |
17:42.24 | Maloeran | Ah, so you mean actually writing the shaders :) |
17:42.55 | Maloeran | Yes, I really plan to do that since... about an year. I presently feel I terribly need to take a break from raytracing |
17:43.52 | Maloeran | The old demo had dancing lights code that could be imported without too much trouble, although it wouldn't support distributed processing or frame buffering |
17:47.13 | Maloeran | Working alone for 8 months on something without cooperation or collaboration is really demotivating, I guess that's why it's called "work"... It's difficult to maintain focus against other projects that involve cooperation and teamwork |
17:54.51 | brlcad | Maloeran: yep, to be expected some -- variety is the spice of life! |
17:55.30 | brlcad | why do you think I also work on BZFlag? :) |
17:55.42 | brlcad | and other open source projects for similar reasons.. |
17:55.53 | Maloeran | :) What is BZFlag? I have other projects as well |
17:56.14 | brlcad | you've never played bzflag? |
17:56.33 | Maloeran | But I think I'm really tired of raytracing at this point, I solved all problems on paper 1-2 years ago and it now has no appeal whatsoever. There are times I think I feel physical pain when I force myself to work on it |
17:56.50 | IriX64 | BZFlag=get to work, the boss is on the way? ;) |
17:57.06 | brlcad | it's a game .. one of the most popular open source games ever actually, been around for more than a decade, continues to be developed and continues to gain popularity actually unlike a lot of games |
17:57.12 | Maloeran | Oh BZFlag, I read about it, I haven't tried |
17:57.25 | Maloeran | Very nice project |
17:58.11 | IriX64 | reminds me of quake:) |
17:58.22 | brlcad | that's where the other half of my time goes |
17:58.30 | IriX64 | heh |
17:58.39 | brlcad | when the commits aren't flowing in here, it's probably because i'm working on bzflag |
17:58.42 | Maloeran | I have asked Survice if I could only work part-time for a little while, put my mind on new problems and stuff I like |
17:58.46 | brlcad | or i'm out rowing ;) |
17:58.46 | Maloeran | Eheh great |
17:59.57 | Maloeran | I'm starting to really like climbing, even in a huge interior gymnasium |
18:00.33 | IriX64 | good thing rounds a shape or i would be out of shape :) |
18:01.29 | IriX64 | ill be back when the builds finished. |
18:02.31 | Maloeran | And my unsafe ( -ffast-math, etc. ) hand-picked optimisation for the BRL-CAD benchmark is now 9 times faster than IriX64's, that's more like it :} |
18:03.38 | brlcad | yeah, it would be nice if some more aggressive flags were added during configure |
18:04.42 | brlcad | -ffast-math is particularly problematic on some platforms, but there are other useful ones that should be added like the alignment flags |
18:05.06 | brlcad | and arch-specific -mcpu style flags |
18:15.19 | *** join/#brlcad ``Erik (i=erik@c-69-250-155-85.hsd1.md.comcast.net) |
18:20.00 | Maloeran | -mcpu has been deprecated long ago, it's -mtune these days |
18:20.57 | brlcad | yep, hence style |
18:21.25 | Maloeran | I like how it says I'm 19594 times faster than the reference, it shows the age of BRL-CAD :) |
18:23.57 | brlcad | yeah, one of the very first actually reported 1 ;) |
18:24.23 | brlcad | running a ray-trace could take days |
18:24.38 | dtidrow_work | isn't VGR dead now? |
18:24.46 | brlcad | yeah |
18:24.55 | brlcad | chuck has her platters sitting in his office |
18:25.32 | dtidrow_work | thought it had gotten retired some time ago |
18:26.40 | IriX64 | Maloeran: does rt discriminate between colors? |
18:27.09 | IriX64 | heh a red thread a green thread ... |
18:28.43 | IriX64 | rgb all you need? wait... maybe 256 colors, = 256 threads? |
18:35.18 | IriX64 | just spectulating. |
18:36.25 | Maloeran | You could theorically shoot different wavelenghts independantely, for even greater realism, but I never heard anyone doing that and BRL-CAD sure doesn't |
18:37.02 | Maloeran | independently, rather |
18:37.11 | IriX64 | was wondering where it would be usefull |
18:38.58 | IriX64 | would be difficult to co-ordinate the rays, unless you had the thread report hit/miss or something |
18:39.07 | Maloeran | For photo-realistic rendering of scenes with complex transparent mediums with non-parallel surfaces ( glass prisms ) |
18:39.25 | brlcad | brl-cad does shoot ranges of wavelengths in libmultispectral |
18:39.32 | IriX64 | heh true color :) |
18:39.34 | Maloeran | Oh? Neat |
18:40.30 | brlcad | more of the overall everything including the kitchen sink approach |
18:40.59 | IriX64 | heh the doitall approach, as opposed to the doitoit :) |
18:47.05 | IriX64 | mmm rt glitches every 20 seconds to 50% cpu, you're good. |
18:47.56 | IriX64 | ahh now its got something to do iguess its at 99% |
18:53.40 | IriX64 | 430mb ram, dunno how much of that is rt |
18:53.51 | IriX64 | wait i will |
18:54.05 | IriX64 | 15.984k |
18:54.14 | IriX64 | 15,984k |
18:54.28 | IriX64 | steady good job. |
19:08.01 | IriX64 | 41,624, looks like it grabbed another chunk. |
19:08.37 | IriX64 | its climbing :( |
19:10.15 | IriX64 | 4k chunks |
19:22.45 | CIA-7 | BRL-CAD: 03erikgreenwald * 10rtcmp/ (tri.c tri.h Makefile.am adrt/adrt.c): beginning of triangle caching |
19:41.09 | *** join/#brlcad dli (n=dli@wireless-194-246.uchicago.edu) |
19:41.51 | dli | can I specify a nonbending start point with pipe? |
19:43.48 | *** join/#brlcad ^Albe^ (n=alberto@host150-54-dynamic.56-82-r.retail.telecomitalia.it) |
19:46.02 | *** part/#brlcad ^Albe^ (n=alberto@host150-54-dynamic.56-82-r.retail.telecomitalia.it) |
20:39.34 | *** join/#brlcad clock_ (i=clock@84-72-60-230.dclient.hispeed.ch) |
21:44.27 | IriX64 | http://www.pastebin.ca/398002 < === optimizide, all debug off bench |
21:44.43 | IriX64 | optimized too |
21:45.04 | IriX64 | really? |
21:45.19 | ``Erik | alignment issues |
21:45.27 | IriX64 | i usually leave it on for the protection it affords as said |
21:46.07 | IriX64 | err help it affords in tracking stuff. |
21:46.22 | IriX64 | protects me against my self. |
21:46.28 | ``Erik | the further you get form "./configure", the less we can help you |
21:46.32 | ``Erik | from |
21:46.43 | IriX64 | heh ive learned that lesson |
21:47.25 | ``Erik | (now people who really know their shit, the further they get from ./configure, the more they can help us... but...) |
21:48.08 | IriX64 | configur has a plethora of options, sometimes i play with them ... just to see. |
21:49.02 | ``Erik | slews of options, many don't show up with --help... I tend to call the script two different ways... so those two ways... on those arch/os platforms I use... tend to work |
21:50.09 | ``Erik | I'd imagine brlcad has a couple different ways he calls the script, and on different arch/os platforms |
22:01.26 | clock_ | ``Erik: now the wiggly outlines in rt-edge are fixed, having been caused by a race condition, aren't they? |
22:01.56 | ``Erik | yeah, how many cpu's is your p3? |
22:02.34 | IriX64 | ``Erik, just to set the record straight i'm not one of the guys you mentioned, I'm just a hacker, i do best starting with something that already exists. |
22:03.36 | ``Erik | heh, I think I have a pretty good feel for what your abilities are, irix64... I would be highly surprised if you seriously schooled brlcad or me... :) but it's all good *shrug* |
22:05.04 | ``Erik | clock: the patch and comments are pretty straight forward... in painting to the frame buffer, it used the correct y value to "seek", where writing to a file just blasted the next scanline done, whether or not it was out of order... |
22:05.16 | ``Erik | entire scanlines were swapped :) |
22:06.08 | ``Erik | tweaking things so I had 400 threads on a 4 core box made teh issue a lot more consistent, and doing both display framebuffer (-F/dev/Xl) and pix-fb on the output file side by side made it obvious |
22:22.18 | Maloeran | Ah, seems I'll be in Maryland around next Friday |
22:22.27 | ``Erik | O.o eh? |
22:22.43 | ``Erik | the 'completed contract' meeting? or setup for something new? :D |
22:25.01 | Maloeran | Apparently, a demonstration of RF to Justin and friends |
22:26.06 | IriX64 | for some reason i think it affects X too, not entirely sure yet. |
22:26.39 | ``Erik | aight, interesting.... I haven't heard of that O.o heh |
22:27.03 | IriX64 | :) |
22:38.00 | clock_ | brlcad: and when will a version with the fix come out? |
23:25.32 | IriX64 | That math library is intense. |
23:30.37 | IriX64 | brlcad: how big is your mged.exe? |
23:32.42 | IriX64 | heh ``Erik pipes up and says give it a rest dude, there's always somebody with a bigger one :) |
23:37.24 | IriX64 | http://www.pastebin.ca/398133 <-- just so you know i put the magic thing back. |
23:37.36 | IriX64 | now to fix it. |
23:46.58 | IriX64 | dudes... your comparing your magic number to the *pointer? |
23:50.14 | IriX64 | :) |
23:52.50 | IriX64 | I'll simply use if( (_magic) != (_magic)) error out ;) |
23:56.11 | IriX64 | gotta go walk the k9 bbiab. |