#jogamp @ irc.freenode.net - 20150314 05:05:37 (UTC)
20150314 05:05:37 -jogamp- Previous @ http://jogamp.org/log/irc/jogamp_20150313050537.html
20150314 05:05:37 -jogamp- This channel is logged @ http://jogamp.org/log/irc/jogamp_20150314050537.html
20150314 05:12:15 * hija (~hija@anon) Quit (Quit: hija)
20150314 08:32:35 * monsieur_max (~maxime@anon) has joined #jogamp
20150314 10:45:49 * monsieur_max (~maxime@anon) Quit (Quit: Leaving.)
20150314 10:46:45 * monsieur_max (~maxime@anon) has joined #jogamp
20150314 10:48:10 * Eclesia (~eclesia@anon) has joined #jogamp
20150314 10:48:14 <Eclesia> hi
20150314 11:33:45 * hija (~hija@anon) has joined #jogamp
20150314 13:17:55 * hija (~hija@anon) Quit (Quit: hija)
20150314 13:41:46 <Eclesia> I have something odd : my layout is like this :
20150314 13:41:48 <Eclesia> layout(location = 0) in mat4x3 l_insttrs;
20150314 13:41:48 <Eclesia> layout(location = 3) in vec3 l_position;
20150314 13:41:48 <Eclesia> layout(location = 4) in vec3 l_normal;
20150314 13:42:11 <Eclesia> But he still complain : two vertex attribute variables (named l_position and l_insttrs) were assigned to the same generic vertex attribute
20150314 13:42:30 <Eclesia> mat4x3 requieres 3 slots isn't that right ?
20150314 13:42:49 <sgothel> that is the question .. sorry not familiar w/ layout
20150314 13:42:55 <sgothel> what is 1 slot?
20150314 13:43:17 <Eclesia> seems like the base vertex attribute size : size of 4 floats
20150314 13:43:29 <sgothel> yes, defaults to 4
20150314 13:43:29 <Eclesia> vec2/3/4 take 1 slot
20150314 13:43:37 <sgothel> so you can use mat4x4
20150314 13:43:49 <sgothel> 4th defaults to 1
20150314 13:44:02 <sgothel> (AFAIK)
20150314 13:59:47 * hija (~hija@anon) has joined #jogamp
20150314 14:00:32 * hija (~hija@anon) Quit (Client Quit)
20150314 14:04:11 * hija (~hija@anon) has joined #jogamp
20150314 14:10:10 <Eclesia> erf, that's an ugly way to bind datas : http://pastebin.com/kMvHehDh
20150314 14:11:37 <sgothel> interleaved .. technically looks good :)
20150314 14:11:50 <sgothel> I know .. beauty .. look at it like manual linker code :)
20150314 14:12:31 <sgothel> glEnableVertexAttribArray ? still used for core ?
20150314 14:19:27 <Eclesia> don't know
20150314 14:36:02 <rmk0> is in core, yes
20150314 14:36:15 <Eclesia> so still need it ?
20150314 14:36:21 <Eclesia> rmk0: o/
20150314 14:36:24 <rmk0> lo
20150314 14:36:30 <rmk0> still need it to do what it does, yes!
20150314 14:36:55 <sgothel> ah .. hehe, this time me so lazy looking things up :)
20150314 14:38:14 <Eclesia> instancing works now... damn that required many changes. added VertexAttribute class, automatic layout index (based on data type) and properly binding datas by 4floats blocks :/
20150314 14:38:39 <rmk0> will be so nice not to have all this nonsense with vulkan
20150314 14:38:59 <sgothel> oh .. you need to link your data there as well ofc
20150314 14:39:44 <sgothel> yesterday was looking more into Graham Sellers talk .. looks like we could map Vulkan .. a bit like an voluntary VKContext,
20150314 14:39:58 <sgothel> which may hold associated VKQueue ..
20150314 14:40:19 <sgothel> at least one queue is required anyways and shall be operated thread safe
20150314 14:40:39 <sgothel> (so the queue <-> queue interop)
20150314 14:40:51 <rmk0> the slides suggested that initialization of vulkan was more like opencl than opengl
20150314 14:41:02 <sgothel> sure ..
20150314 14:41:19 <sgothel> i.e. selecting devices first
20150314 14:41:24 <rmk0> i'd not personally stray too far from that model
20150314 14:41:31 <sgothel> based on caps .. so its a bit like +devices
20150314 14:41:49 <sgothel> i.e. today we select by caps only .. assuming one device
20150314 14:42:07 <sgothel> it is not thaaat different
20150314 14:42:09 <rmk0> obviously you can't even open an opengl context on x11 without hundreds of code, so it pays to have all that crap hidden behind jogl
20150314 14:42:19 <sgothel> hehe
20150314 14:42:35 <rmk0> it doesn't look anywhere near as painful to open a vulkan context, so it might not be a problem to have less of an abstraction over it
20150314 14:42:49 <sgothel> the windowing system buffer interop will change a bit as well
20150314 14:43:06 <sgothel> oh .. yes it will be more complex, since it adds one more variable
20150314 14:43:10 <sgothel> (the devices ..)
20150314 14:43:14 <rmk0> right
20150314 14:43:33 <sgothel> and the rest will be similar .. buffer caps etc
20150314 14:43:48 <rmk0> i have to buy new hardware this year
20150314 14:43:53 <sgothel> but yeah, the visible onscreen default framebuffer is optional
20150314 14:43:59 <rmk0> putting off buying a gpu until i know i'll get vulkan support with open drivers
20150314 14:45:28 <sgothel> the latest Intel will be a good choice :)
20150314 14:45:42 <sgothel> i.e. open source driver will be released when they are ready
20150314 14:45:44 <rmk0> don't really want to buy another intel cpu
20150314 14:46:03 <rmk0> dislike their approach to socket compatibility (in other words, a new socket every few months to guarantee no compatibility)
20150314 14:46:20 <sgothel> in jogl today .. the GLDrawable and notion of FBObject are a bit redundant ..
20150314 14:46:36 <sgothel> i.e. there shall be no reason to differentiate them ..
20150314 14:46:57 <sgothel> so our FBO based GLDrawable removes that differentiation ..
20150314 14:47:15 <sgothel> in vulkan .. its all drawables or [frame]buffers
20150314 14:47:26 <rmk0> right
20150314 14:48:03 <sgothel> so for JOGL 3.0 .. I guess I will try to mimick our stuff w/ Vulkan .. and see where we have similarities for OO abstraction
20150314 14:48:18 <rmk0> 3.0 \o/
20150314 14:48:19 <sgothel> but it could be ofc - that the letters SCREAMS: stop it
20150314 14:48:30 <sgothel> *letter*
20150314 14:48:44 <sgothel> i.e. where it makes no sense .. really to find common ground
20150314 14:48:48 <sgothel> so this has to be seen
20150314 14:49:13 <sgothel> i.e. the common ground would be just hollow interfaces .. and the like
20150314 14:49:46 <sgothel> the basic binding ofc shall expose all of Vulkan 1:1 - granted
20150314 14:49:59 <sgothel> just talking about our managed code stuff ..
20150314 14:50:31 * hija (~hija@anon) Quit (Quit: hija)
20150314 14:51:23 <sgothel> I like the new memory management very much .. and its implicit multithreading undefined behavior, i.e. up to the coder to deal w/ it - using semaphores/sync etc
20150314 14:51:54 <rmk0> is good
20150314 14:52:45 <sgothel> so the diff to opencl .. maybe: rendering pipeline definition, and framebuffer semantics
20150314 14:52:57 <sgothel> and some other pixel attributes/caps ..
20150314 14:53:09 <sgothel> not much :)
20150314 14:53:45 <sgothel> maybe .. they missed the opportunity to melt it together ..
20150314 14:53:56 <sgothel> oh .. dinner, run run :)
20150314 14:54:02 * rmk0 waves
20150314 15:08:24 * Eclesia notes : never forget to set glVertexAttribDivisor value
20150314 15:08:40 <Eclesia> had wasting 30minutes on such little nasty bugs
20150314 15:13:29 <Eclesia> hate*
20150314 16:01:53 * monsieur_max (~maxime@anon) Quit (Ping timeout: 240 seconds)
20150314 16:02:28 * monsieur_max (~maxime@anon) has joined #jogamp
20150314 18:18:46 * hija (~hija@anon) has joined #jogamp
20150314 19:09:15 * DannyB (6ced813b@anon) has joined #jogamp
20150314 19:09:48 * Red_ (8e81728d@anon) has joined #jogamp
20150314 19:17:48 * DannyB (6ced813b@anon) Quit (Quit: Page closed)
20150314 19:30:58 * Red_ (8e81728d@anon) Quit (Ping timeout: 246 seconds)
20150314 20:40:34 * d (6ced813b@anon) has joined #jogamp
20150314 20:40:58 * d is now known as Guest58512
20150314 20:41:06 * Red_ (8e81728d@anon) has joined #jogamp
20150314 20:46:37 <Red_> Hello everyone, I am trying to reach out Demoscene Passivist. Can someone point me in the right direction please?
20150314 20:50:11 <Red_> He goes by Lutsche on this forum I beleive
20150314 21:05:53 * Red_ (8e81728d@anon) has left #jogamp
20150314 21:08:36 * Red__ (8e81728d@anon) has joined #jogamp
20150314 21:12:34 <Eclesia> never seen this name before
20150314 21:24:44 * Eclesia (~eclesia@anon) Quit (Quit: Leaving.)
20150314 21:25:12 <rmk0> Red__: Dominik Ströhlein
20150314 21:25:22 <rmk0> i don't have an address for him but sgothel will when he appears, i think
20150314 21:26:28 <Red__> thank you, i hope to hear from sgothel then
20150315 02:02:23 * monsieur_max (~maxime@anon) Quit (Quit: Leaving.)
20150315 04:27:58 * Guest58512 (6ced813b@anon) Quit (Quit: Page closed)
20150315 05:05:37 -jogamp- Continue @ http://jogamp.org/log/irc/jogamp_20150315050537.html