#jogamp @ irc.freenode.net - 20130925 05:05:29 (UTC)


20130925 05:05:29 -jogamp- Previous @ http://jogamp.org/log/irc/jogamp_20130924050529.html
20130925 05:05:29 -jogamp- This channel is logged @ http://jogamp.org/log/irc/jogamp_20130925050529.html
20130925 06:46:27 * hharrison_ (~chatzilla@anon) Quit (Quit: ChatZilla 0.9.90.1 [Firefox 23.0.1/20130902133717])
20130925 07:39:51 * monsieur_max (~maxime@anon) has joined #jogamp
20130925 13:06:42 <xranby> http://mail.openjdk.java.net/pipermail/distro-pkg-dev/2013-September/024817.html - Chrome phasing out NPAPI [affects icedtea-web plug-in] (and the oracle plugin)
20130925 13:27:33 <monsieur_max> hey, i don't know if you've seen this news : http://blogs.nvidia.com/blog/2013/09/22/gpu-coming-to-java/
20130925 13:30:09 <xranby> monsieur_max: yes, that basically say IBM will optimize its java classes to use GPU with help of existing java gpu frameworks. it looks like they are quite arm in arm with nvidia
20130925 13:30:24 <xranby> and the cuda framework
20130925 13:31:38 <xranby> i got the impression from the IBM talk, part of the java one opening that this technology lands inside their IBM J9 offering.
20130925 13:31:52 <monsieur_max> yep, i think that CUDA is definitely the most important part... but i wonder how will this integrate... and what kind of application would be the best candidate for such optimization
20130925 13:33:05 <xranby> i think they target traditional workloads such as "unzip" that happens a lot during java execution
20130925 13:34:30 <xranby> oracle also talked a lot about gpu optimization, in the future, by painting the picture that by introducing lambas into the language will open up to off load tasks to the gpu
20130925 13:36:29 <monsieur_max> ... how could this could be even possible without threads ?
20130925 13:37:05 <monsieur_max> you know, dedicated thread off loaded to gpu , but not something automatic
20130925 13:38:17 <monsieur_max> ok, looks like i'm just making assumption here ...
20130925 13:39:05 <xranby> sure its possible,similar to how aparapi do it
20130925 13:39:12 <xranby> http://code.google.com/p/aparapi/
20130925 13:40:31 <xranby> thus the gpu is able to process code with loops that can be unrolled into parallel calculations
20130925 13:40:47 <xranby> sent to the gpu using opencl
20130925 13:43:18 <monsieur_max> nice api :)
20130925 13:45:12 <monsieur_max> gosh, it brings an incredible amount of calculation power ... for suitable cases, of course
20130925 13:55:48 <xranby> if you want to watch keynotes then they are all online: http://medianetwork.oracle.com/video/player/2685720528001
20130925 15:28:38 <hharrison> RE: NPAPI, I thought that chrome already moved to there own plugin api PPAPI (pepper)
20130925 16:00:43 * odin_ (~Odin@anon) Quit (Quit: Leaving)
20130925 16:00:48 * monsieur_max (~maxime@anon) Quit (Quit: Leaving.)
20130925 16:53:27 * monsieur_max (~maxime@anon) has joined #jogamp
20130925 17:04:21 * hharrison (~chatzilla@anon) Quit (Remote host closed the connection)
20130925 17:23:07 <sgothel> I guess the point here is that [_again_] google forces others to have a branch for their products
20130925 17:23:20 <sgothel> i.e. webkit, now chromse plugin, etc
20130925 17:24:02 <sgothel> instead of sitting together w/ firefox dudes and get a a NP3 API :)
20130925 20:13:42 * monsieur_max (~maxime@anon) has left #jogamp
20130925 21:02:20 * void256 (~chatzilla@anon) has joined #jogamp
20130925 21:35:45 * odin_ (~Odin@anon) has joined #jogamp
20130926 01:44:23 * void256 (~chatzilla@anon) Quit (Quit: ChatZilla 0.9.90.1 [Firefox 23.0.1/20130814063812])
20130926 05:05:29 -jogamp- Continue @ http://jogamp.org/log/irc/jogamp_20130926050529.html