So I did some more tryings, here what I found out so far:
jit.freenect with 32bits only camera works (means no depth map possible), same with 64bits (it is also very slow) (help file use 106% of CPU!)
jit.openni with 32 bits works (camera, depth, skeleton) but it is very slow (help file use 106% of CPU!)
synapse we do not have yet, we will try asap
isight just opens with 64bits, with 32 does not
ps3eye webcam with 32bits opens but freeze immediately the image, with 64 bits does not read it (probably the madam driver is not supporting the 64)
jit.net.send and jit.net.recv works with localhost but not between two computers (I try new macbook pro with new macbook pro, new macbook pro with old macbook with maverick installed, new macbook pro with old macbook pro with mountain lion)… the result is always an error message in the max window saying “CORRUPT DATA”
Also, the local host from the help file does not work in the one with mountain lion and in the old macbook with maverick installed
no, i tried between 2 totally identical new 15" MacBooks and "corrupt data" message come up.
does anybody have any idea what could be the problem with these new haswell cpu?
and if it is solvable in short time? i know there is just been just an update for the maverick situation, but it's really depressing seeing a new computer running slower than an old 13" inch with no dedicated gpu
yes i'm well aware that the cpu drops when patcher is unlock, it's not that what i wanted to point out, what i wanted to do is to compare the new 15" macbook (with dedicated graphic card) with an old 13" macbook 2010 (with NO dedicated graphic card) and just to be sure i checked the cpu both with lock and unlock patcher. The point is that the patch attached above run faster on the old laptop, the old 13" waste 8% less of cpu compared to the new 15"
yesterday (and also 5 min ago just to be sure) i did the same thing with an example patch for gpu testing (phys.multiple.mesh) and also in this case the old 13" was faster (55fps) compared to the new 15" (29fps)
i was totally sure it was just a late retina problem, i have a friend with old retina and he don't have this problem and he probably just forgot to have switched on the low resolution option while we were making the comparison.
the last curious behavior that max still has is with the pays.multiple.mesh patch:
the patch runs 22 fps on old retina, 30 on new retina, 52 on old 2010 13" MBP,
Don't know if in this case has sense but i will try a SMC reset as suggested by Karl Kliem in the http://cycling74.com/forums/topic/retina-performance/
@stefano or your friend was perhaps using an older version of Max which had no retina support and thus always in low resolution mode. This is a very recent feature.
As for the performance differences with the phys.multiple.mesh.maxpat, it looks like there are some OpenGL problems with performance of displaylists on those machines under 10.9.
I was able to speed up that patch significantly to match older machine/OS performance or better with jit.gl.model @cache_mode vbo @displaylist 0. I'm not sure if it is an OpenGL bug which Apple will address at this point, but your problem for that patch is definitely the displaylist issue.
I know this thread is really old, but I wanted to chime in with something I discovered in case others are still running into issues (as I was, yesterday). I upgraded my computer and was having problems using the jit.qt.grab object in Max5--worked just fine in 6 and 7, but nothing but errors in 5. After trying a lot of the suggestions here, I was still coming up short--until I found this command to allow un-checking the application's low resolution box:
Open App’s Info.plist file via TextEdit or any other text editing application. (You can find Info.plist file in “/Applications/ApplicationName.app/Contents/Info.plist”).
Insert this code at the end. (Just before ).
I can't recommend this solution officially because I don't want to be responsible for breaking anyone's stuff, but it worked for me!