#jogamp @ irc.freenode.net - 20150303 05:05:35 (UTC)

20150303 05:05:35 -jogamp- Previous @ http://jogamp.org/log/irc/jogamp_20150302050535.html
20150303 05:05:35 -jogamp- This channel is logged @ http://jogamp.org/log/irc/jogamp_20150303050535.html
20150303 07:18:51 * jvanek (jvanek@anon) has joined #jogamp
20150303 07:47:06 * monsieur_max (~maxime@anon) has joined #jogamp
20150303 08:23:35 * zubzub (~zubzub@anon) Quit (Remote host closed the connection)
20150303 08:26:47 * zubzub (~zubzub@anon) has joined #jogamp
20150303 08:31:14 * eclesia (~husky@anon) has joined #jogamp
20150303 08:31:22 <eclesia> good morning
20150303 09:22:43 * xranby (~xranby@anon) has joined #jogamp
20150303 09:24:12 <xranby> https://www.khronos.org/vulkan <- the next generation graphics API from Khronos.
20150303 09:24:39 <xranby> Initial specs and implementations expected this year Will work on any platform that supports OpenGL ES 3.1 and up
20150303 09:24:48 <zubzub> wasn't it called glNext?
20150303 09:24:56 <zubzub> and hi xranby, long time no see
20150303 09:25:09 <xranby> i guess they renamed it
20150303 09:26:18 <xranby> zubzub: yes, it was a long time, i am obsessed with raising the lowest bar... that is trying to get mesa opencl working
20150303 09:26:39 <xranby> to have opencl wokring everywhere
20150303 09:28:47 <xranby> currently thinkering with a raspberry pi 2 + eric anholts foss mesa 3d gallium gpu driver
20150303 09:29:30 <zubzub> I'm mostly working on getting a fully funciton java wayland compositor
20150303 09:29:32 <zubzub> https://github.com/Zubnix/westmalle
20150303 09:29:49 <xranby> opencl implementation on top of mesa3d using llvm and galliumcompute http://dri.freedesktop.org/wiki/GalliumCompute/
20150303 09:30:07 <zubzub> sounds complicated
20150303 09:31:18 <xranby> the trickiest part is to build yes... distributions like debian have not yet enabled opencl in mesa
20150303 09:31:29 <xranby> so i am using buildroot
20150303 09:31:37 <xranby> to cross compile the parts i want to test
20150303 09:32:49 <xranby> joal using openal already work on the pi
20150303 09:33:27 <xranby> jocl require some implementation that can use the GPU and mesa3d may be the first to do that on the pi
20150303 09:34:45 <xranby> jogl currently supports opengl es w/o swing using the broadcom gpu blob,,, but will support opengl & ES & X11 using the mesa 3d implementation
20150303 09:39:23 <xranby> http://www.pcworld.com/article/2891613/meet-vulkan-the-powerful-platform-agnostic-gaming-tech-taking-aim-at-directx-12.html "Khronos Group is throwing out nearly 25 years of OpenGL and starting fresh with a powerful new gaming-centric API, Vulkan (previously known as glNext or the "Next Generation OpenGL Initiative") that’s built for the next generation of games."
20150303 09:40:12 <xranby> http://blog.imgtec.com/powervr/trying-out-the-new-vulkan-graphics-api-on-powervr-gpus
20150303 09:41:43 <xranby> more news: OpenCL 2.1 released today https://www.khronos.org/opencl/
20150303 09:46:20 <xranby> https://www.khronos.org/news/press/khronos-reveals-vulkan-api-for-high-efficiency-graphics-and-compute-on-gpus press release: Khronos Reveals Vulkan API for High-efficiency Graphics and Compute on GPUs "New SPIR-V intermediate language shared by Vulkan and OpenCL 2.1"
20150303 09:58:36 * zzuegg (~zzuegg___@anon) has joined #jogamp
20150303 09:59:49 * xranby (~xranby@anon) Quit (Quit: Leaving.)
20150303 10:01:01 * xranby (~xranby@anon) has joined #jogamp
20150303 10:03:20 <rmk0> \o/
20150303 10:05:28 <xranby> zubzub: https://github.com/Zubnix/westmalle/commits/master <- kudos for publishing your work!
20150303 10:20:26 <zubzub> np
20150303 10:20:46 <zubzub> resizing works as well actually, I just have to adjust my test client to properly resize
20150303 10:21:14 <monsieur_max> nice :)
20150303 10:32:11 * zzuegg (~zzuegg___@anon) Quit (Remote host closed the connection)
20150303 10:38:30 * zzuegg (~zzuegg___@anon) has joined #jogamp
20150303 10:59:08 <xranby> Vulkan: " there are no glUniform*() equivalent entry points in Vulkan; instead, writing to GPU memory is the only way to pass data to shaders. When you call glUniform*(), the OpenGL ES driver typically needs to allocate a driver managed buffer and copy data to it, the management of which incurs CPU overhead. In Vulkan, you simply map the memory address and write to that memory location directly."
20150303 11:00:57 <zubzub> almost like you pass data from java<->c with a directbytebuffer ;)
20150303 11:01:09 <xranby> lets hope so!
20150303 11:19:56 * jvanek (jvanek@anon) Quit (Quit: Leaving)
20150303 13:34:14 <rmk0> \o/
20150303 13:34:34 <rmk0> i'm hoping valve get an implementation into all the open drivers
20150303 13:46:05 <rmk0> presumably this stuff would involve a lot less of an implementation burden than any of the existing opengl APIs
20150303 13:46:15 <rmk0> not that there's... actually a specification yet
20150303 13:50:29 <xranby> the hardest part of vulcan is to implement the spir-v compiler https://www.khronos.org/registry/spir-v/papers/WhitePaper.pdf
20150303 13:51:02 <xranby> both vulcan and opencl 2.1 uses spir-v intermediate format
20150303 13:52:14 <rmk0> compiler?
20150303 13:52:24 <rmk0> as in turning a language that you define into spir-v?
20150303 13:53:47 <xranby> spir-v to GPU compiler implemented by the driver
20150303 13:53:53 <rmk0> oh, right
20150303 13:55:06 <rmk0> would expect that to be fairly easy
20150303 13:55:40 <xranby> also known as the "Vulkan Runtume"
20150303 13:55:51 * jvanek (jvanek@anon) has joined #jogamp
20150303 13:57:28 <xranby> The Vulkan Runtime may be a SPIR-v to OpenGL ES 3.1 shader compiler
20150303 13:57:43 <xranby> making use of the promise : "Initial specs and implementations expected this year Will work on any platform that supports OpenGL ES 3.1 and up "
20150303 13:58:37 <xranby> it may also be an OpenCL runtime https://www.khronos.org/spir
20150303 13:58:48 <xranby> OpenCL 2.1 runtime
20150303 13:59:52 <rmk0> my main concern is implementations of this stuff in the free drivers
20150303 14:00:01 <rmk0> i can't target any of it if it only works on binary drivers
20150303 14:01:08 <xranby> rmk0: i put my bets on GalliumCompute to implement OpenCL support into Mesa3D
20150303 14:01:28 <xranby> http://dri.freedesktop.org/wiki/GalliumCompute/
20150303 14:02:41 <xranby> getting a SPIR-v parser into Mesa3D may be possible using GalliumCOmpute (that is allready using llvm)
20150303 14:04:41 <xranby> http://www.phoronix.com/scan.php?page=article&item=khronos-vulcan-spirv&num=1 "the Khronos Group has a work-in-progress item of putting out a SPIR-V to LLVM IR pass that would convert this OpenCL and Vulkan code representation into LLVM IR."
20150303 14:06:09 <xranby> thus by adding this LLVM IR pass (hopefully opensource) completes the Vulcan implementation into Mesa3D GalliumCompute architecture
20150303 14:09:02 * sgothel (~sgothel@anon) Quit (Ping timeout: 245 seconds)
20150303 14:10:05 <rmk0> \o/
20150303 14:13:29 <xranby> linux distributions only need to pass --enable-opencl to the mesa3d compile to actually use GalliumCompute and they have acknowledge it on their todo list
20150303 14:14:38 <rmk0> i think it's enabled on arch, but my hardware can't do it
20150303 14:15:27 <rmk0> i have a 4670 and a 4870, both of which have "partial" support for opencl 1.0
20150303 14:15:35 <rmk0> is about time i upgraded...
20150303 14:29:51 <rmk0> i'd love to have an engine that did real-time GI
20150303 14:30:10 <rmk0> not even particularly interested in it for realism's sake, i just want shadow casting on all light sources
20150303 14:30:52 <rmk0> i've heard that it's possible to have hundreds of shadow casting lights on opengl 4.5 + extensions, due to the lack of draw call overhead
20150303 14:36:39 <rmk0> er... point was that vulcan should enable that kind of rendering
20150303 14:36:54 <rmk0> no draw call overhead, easy instancing, etc
20150303 14:36:58 <xranby> \o/
20150303 15:26:38 * doev (~doev@anon) has joined #jogamp
20150303 16:02:06 * jvanek (jvanek@anon) Quit (Quit: Leaving)
20150303 16:03:07 * jvanek (jvanek@anon) has joined #jogamp
20150303 16:04:57 * jvanek (jvanek@anon) Quit (Client Quit)
20150303 16:48:58 * doev (~doev@anon) Quit (Quit: Verlassend)
20150303 16:52:54 * monsieur_max (~maxime@anon) Quit (Quit: Leaving.)
20150303 16:58:20 * eclesia (~husky@anon) has left #jogamp
20150303 17:41:59 * monsieur_max (~maxime@anon) has joined #jogamp
20150303 19:19:55 * sgothel (~sgothel@anon) has joined #jogamp
20150303 19:19:56 * ChanServ sets mode +v sgothel
20150303 19:20:13 <sgothel> back from our local power outage ...
20150303 19:20:37 <sgothel> @Mark: http://jogamp.org/git/?p=jogamp.org.git;a=commit;h=0ee6ac9e221343928fedfeaf315cc08cf66be8e0
20150303 19:21:22 <sgothel> @All: always be alert if licensing issues
20150303 19:21:39 <rmk0> yow!
20150303 19:21:53 <sgothel> Question: File still in repo .. do we need rewrite history ? (in general)
20150303 19:22:52 <sgothel> I guess they will be fine for now .. but in general, maybe one needs a more thorough solution ..
20150303 19:23:02 <sgothel> @Xerxes: thx for the heads up
20150303 19:23:06 <rmk0> i don't know how this is handled in git
20150303 19:24:05 <sgothel> still 'massaging' GlueGen a bit to allow inclusion of more complex files (new GL headers etc) - hope it will be fit for 'Vulcan' then .. (what a fashion statement like name)
20150303 19:24:27 <rmk0> was i the one that put DIN Medium there?
20150303 19:24:36 <sgothel> I guess so :)
20150303 19:24:48 <rmk0> i don't remember doing that... i know i used it for the text in my slides
20150303 19:25:22 <rmk0> hrm
20150303 19:25:29 <sgothel> well .. I got it from you .. and dumped it in the git repo .. *beat me*
20150303 19:25:39 <rmk0> hehe
20150303 19:26:08 <sgothel> a new source of income maybe? :)
20150303 19:27:54 <sgothel> you have a new email address?
20150303 19:28:08 <rmk0> eh?
20150303 19:28:39 * rmk0 looks bewildered
20150303 19:28:46 <sgothel> mark at io7m.com .. failed ah .. "his message has been rejected because it has a potentially executable attachment "Trademark and copyright violations by jogamp_org.eml"
20150303 19:29:01 <sgothel> your bs3-dallas.accountservergroup.com rejected it .. well
20150303 19:29:04 <rmk0> ...
20150303 19:29:06 <rmk0> that's a new one
20150303 19:29:30 <rmk0> well, presumably it's a new one, noone's ever told me before
20150303 19:29:33 <sgothel> maybe the email was a trojan ? lol
20150303 19:29:37 <rmk0> hehe
20150303 19:29:52 <rmk0> i'm not responsible for administering that server... must be some new thing they've installed
20150303 19:30:28 <sgothel> the big io7m company w/ its many IT minions ..
20150303 19:30:36 <rmk0> there are so many of us
20150303 19:31:09 <sgothel> maybe on day you will open my eyes who/what it is :)
20150303 19:31:18 <rmk0> hehe, there's only me now
20150303 19:31:29 <sgothel> you have to say that :)
20150303 19:31:35 <rmk0> maybe!
20150303 19:31:50 <sgothel> probably a big new secret thingy
20150303 19:37:03 <rmk0> got the email
20150303 19:37:08 <rmk0> seems fine... assume it's all sorted out?
20150303 19:37:23 <sgothel> hope so
20150303 19:40:33 <rmk0> which repos was this? i'd like to find out if/where i pushed it
20150303 19:40:38 <rmk0> i'm usually more careful than that
20150303 19:40:51 <sgothel> linked above ..
20150303 19:41:51 <rmk0> ah, i see
20150303 19:42:11 <rmk0> i think i just sent you the directory full of stuff... didn't realize at the time that it was going into a publically accessible repos
20150303 19:42:29 <rmk0> included the font so that you could open the .odp to extract the slides
20150303 19:43:00 <sgothel> yup: my fault
20150303 19:43:15 * rmk0 hides under the carpet
20150303 19:43:23 <rmk0> tell me when the lawyers have gone
20150303 22:14:33 * monsieur_max (~maxime@anon) Quit (Quit: Leaving.)
20150304 05:05:35 -jogamp- Continue @ http://jogamp.org/log/irc/jogamp_20150304050535.html