#jogamp @ irc.freenode.net - 20150310 05:05:36 (UTC)

20150310 05:05:36 -jogamp- Previous @ http://jogamp.org/log/irc/jogamp_20150309050536.html
20150310 05:05:36 -jogamp- This channel is logged @ http://jogamp.org/log/irc/jogamp_20150310050536.html
20150310 05:25:21 <sgothel> https://github.com/CMU-SAFARI/rowhammer .. Kim et al. Flipping Bits in Memory Without Accessing Them. ISCA 2014.
20150310 05:49:47 <sgothel> http://googleprojectzero.blogspot.de/2015/03/exploiting-dram-rowhammer-bug-to-gain.html
20150310 05:51:03 <sgothel> ^^^ see, 'Java' can be more secure :)
20150310 05:51:25 <sgothel> (or any other VM tech .. for that matter)
20150310 08:02:28 * doev (~doev@anon) has joined #jogamp
20150310 08:23:23 * monsieur_max (~maxime@anon) has joined #jogamp
20150310 11:05:54 * jvanek (jvanek@anon) has joined #jogamp
20150310 13:48:34 * doev (~doev@anon) Quit (Ping timeout: 245 seconds)
20150310 14:12:39 * doev (~doev@anon) has joined #jogamp
20150310 14:52:34 <rmk0> it's almost as if having a small number of memory-safe primitives with which to produce programs is a good idea
20150310 14:52:53 <zubzub> it's performant yes
20150310 14:53:15 <zubzub> and flexible
20150310 14:53:34 <zubzub> but you have to know exactly what you're doing
20150310 14:53:58 <zubzub> which you can expect from vulcan/opengl programmers
20150310 14:54:03 <zubzub> so yeah
20150310 14:54:06 <zubzub> it's a good idea
20150310 14:54:34 <rmk0> i was only referring to bytecode
20150310 14:54:51 <rmk0> jvm bytecode
20150310 15:02:58 <zubzub> oh in that case
20150310 15:03:02 <zubzub> :p
20150310 15:03:07 <rmk0> hehe
20150310 15:03:38 <zubzub> I saw a presentation a while back (online slides) about the jvm architecture itself
20150310 15:03:39 <rmk0> i extend that to any other bytecode that comes with a verifier that actually proves properties
20150310 15:03:52 <zubzub> the conclussion is that it's becoming horribly out dated
20150310 15:03:59 <rmk0> urhur
20150310 15:04:46 <rmk0> they're doing heavy work on it, as far as i'm aware
20150310 15:05:39 <rmk0> i think i posted presentation slides in here a while back
20150310 15:06:42 <rmk0> http://cr.openjdk.java.net/~jrose/pres/201502-JVMChallenges.pdf
20150310 15:07:50 <rmk0> refreshing lack of, er... delusion
20150310 15:08:08 <rmk0> "we know what's bad, we know what's good, and we're working on making it good"
20150310 15:08:24 <zubzub> sounds good :p
20150310 15:09:36 <rmk0> value types look good
20150310 15:09:46 <rmk0> if nothing else, they should give cheap newtypes as in haskell
20150310 15:10:05 <rmk0> newtype Name = Name String deriving (Eq, Ord, Show)
20150310 15:10:19 <rmk0> Name is type-incompatible with String, but is erased to String at compile-time
20150310 15:10:33 <rmk0> can't do that in java without wrapping the class in another class
20150310 15:10:38 <rmk0> more garbage on the heap
20150310 16:00:52 * doev (~doev@anon) Quit (Ping timeout: 255 seconds)
20150310 16:01:08 * jvanek (jvanek@anon) Quit (Quit: Leaving)
20150310 16:53:12 * Eclesia (~eclesia@anon) has joined #jogamp
20150310 16:53:20 <Eclesia> hi
20150310 16:53:35 <rmk0> lo
20150310 16:54:01 <sgothel> ho
20150310 16:54:40 <sgothel> ^^^ compute demo ? .. :)
20150310 16:55:19 <Eclesia> I missed something ?
20150310 16:55:32 <sgothel> (07:42:23 PM) sgothel: "Add compute shader support in shader utils?" <- if anybody likes to do that demo - would be nice !
20150310 16:56:31 <sgothel> the memory intrinsics .. (unsafe fast array .. whatever) .. I remember, AFAIK plain arrays are now still fastest ..
20150310 16:56:33 <Eclesia> ho right, there is 4.3-4.5 support now :)
20150310 16:56:53 <sgothel> plus using the ole arraycopy .. hehe
20150310 16:57:09 <sgothel> yup .. would be nice to merge in a compute demo for todays release :)
20150310 16:57:36 <sgothel> well .. will search for one myself .. later, if not presented on a silver platter :)
20150310 16:57:44 * Eclesia buzy with 3D N-gons triangulation
20150310 16:58:10 <sgothel> good ole triangles :)
20150310 17:28:59 <Eclesia> triangles everywhere !
20150310 18:12:01 * monsieur_max (~maxime@anon) Quit (Quit: Leaving.)
20150310 18:51:45 * monsieur_max (~maxime@anon) has joined #jogamp
20150310 19:09:11 * mbien (~mbien@anon) has joined #jogamp
20150310 19:12:51 * mbien (~mbien@anon) Quit (Read error: Connection reset by peer)
20150310 19:14:52 * mbien (~mbien@anon) has joined #jogamp
20150310 22:32:54 * monsieur_max (~maxime@anon) Quit (Quit: Leaving.)
20150310 23:34:32 <Eclesia> what a waste ... lost nearly a full day on those Ngons.
20150310 23:36:04 <Eclesia> here is some free feedback : when you triangulate an Ngon, project the 3d points in 2d on the plan defined by the normalized sum on all vertice normals. this is what given the best result.
20150310 23:37:34 <Eclesia> hope what I'm saying makes sense ^^
20150311 00:25:25 * Eclesia (~eclesia@anon) Quit (Quit: Leaving.)
20150311 03:31:17 * mbien1 (~mbien@anon) has joined #jogamp
20150311 03:32:43 * mbien (~mbien@anon) Quit (Ping timeout: 250 seconds)
20150311 04:38:38 * mbien1 (~mbien@anon) Quit (Read error: Connection reset by peer)
20150311 05:05:36 -jogamp- Continue @ http://jogamp.org/log/irc/jogamp_20150311050536.html