|
http://www.aircrack-ng.org/
|
# ? Sep 6, 2013 17:07 |
|
|
# ? Jun 10, 2024 14:33 |
|
Socracheese posted:i want to dabble with the nvidia parallel computing library but i dont have any big processing problems to solve there are two "popular" lovely wifi protocols. one is WEP, attacks on which are bottlenecked by how fast you can gather relevant packets. the other is WPA w/ WPS, and that's limited by how powerful/stupid the router you're trying break is. the only relevant-to-your-interests, computationally-limited attack is good ol' brute forcing the pw. for that, you've been beaten to the punch: pyrit does it on the gpu already, which for WPA tops out at 110k pw/sec on latest-gen AMD hardware. even then, you'll only get lucky if your neighbour's have picked a lovely dictionary-based password. don't let this put you off though, it'd still be hella instructive for you to write your own. coffeetable fucked around with this message at 17:25 on Sep 6, 2013 |
# ? Sep 6, 2013 17:22 |
|
Rude. posted:patenting this atm the other thing is that when theyre not charging they will be programmed to seek out common flight paths via gps and try to kamikaze themselves into jet intakes around airports
|
# ? Sep 6, 2013 19:09 |
|
Jonny 290 posted:the other thing is that when theyre not charging they will be programmed to seek out common flight paths via gps and try to kamikaze themselves into jet intakes around airports goddammit jonny, now the nsa is double-watching
|
# ? Sep 6, 2013 19:10 |
|
haha
|
# ? Sep 6, 2013 19:10 |
|
ol qwerty bastard posted:
you'll never get great performances with an intepreted language but python is particularly awful
|
# ? Sep 6, 2013 19:16 |
|
|
# ? Sep 6, 2013 19:55 |
|
yeah boiiiiii(tch)
|
# ? Sep 6, 2013 19:55 |
|
made an anroid app that runs an analyzer+voiceprint for whatever audio is playing on the device it's p much a rewrite of a thing i'd made for linux a while ago doesnt work on sarnsung s3's lol Progressive JPEG fucked around with this message at 08:41 on Jan 16, 2017 |
# ? Sep 7, 2013 18:51 |
|
this is my reaction seeing this
|
# ? Sep 7, 2013 18:58 |
|
Symbolic Butt posted:
qfb
|
# ? Sep 7, 2013 19:38 |
|
Jonny 290 posted:yeah the cores of uav's and little flyin things are commodity now. at this point its how you apply them disguise them as sneakers
|
# ? Sep 7, 2013 22:29 |
Symbolic Butt posted:this is my reaction seeing this rip
|
|
# ? Sep 7, 2013 22:30 |
|
ol qwerty bastard posted:i mean granted half of that is the fault of the python interpreter which by my measurements seems to be running at around 2.5 megaflops; my laptop is old but i'm prrreeeeeettttttyyyyy sure it can go faster than that Sounds like it should go faster. is the code online anywhere?
|
# ? Sep 8, 2013 23:28 |
|
oqb try pypy
|
# ? Sep 9, 2013 01:50 |
|
alright i got it to render triangles! lit triangles even! here's an icosahedron wooooooooooo okay this poo poo is making me crazy i need to go for a walk outside
|
# ? Sep 9, 2013 18:56 |
|
peepsalot posted:oqb try pypy i'll look into it, thanks the wizards beard posted:Sounds like it should go faster. is the code online anywhere? my code? no not yet i'm still writing some of the 3d stuff; the 2.5 megaflops estimate was just based on making it do a floating point addition 10 million times which took about 4 seconds anyway here the icosahedron renders a lot faster than spheres; here's a light source orbiting around it as it recedes into the distance: https://www.youtube.com/watch?v=EFSXuVuWqIY yeah it's pretty basic but it took a long time to get all the vector stuff really working properly still have to write a z buffer so i can have triangles occluding each other, a "rotate object" function, a thing that handles how they should be drawn if the camera is really close, etc.
|
# ? Sep 10, 2013 02:10 |
|
i was thinking about your project old qwerty bastard and i realized the next step. you can't get good performance from an interpreted language the way you are using it. but what if you turned an fpga into a python-specific dedicated cpu?
|
# ? Sep 10, 2013 02:36 |
|
what if you just coded it all in assembly
|
# ? Sep 10, 2013 02:40 |
|
cps: started learning about CUDA C. highlight so far:quote:However, it is the responsibility of the programmer not to dereference the pointer now i've been explicitly warned against it, i am guaranteed to do it. coffeetable fucked around with this message at 03:06 on Sep 10, 2013 |
# ? Sep 10, 2013 02:54 |
|
Sagebrush posted:i was thinking about your project old qwerty bastard and i realized the next step. you can't get good performance from an interpreted language the way you are using it. but what if you turned an fpga into a python-specific dedicated cpu? that sounds awful
|
# ? Sep 10, 2013 03:30 |
|
that's what she said
|
# ? Sep 10, 2013 04:47 |
|
coffeetable posted:cps: started learning about CUDA C. highlight so far: Wrap it in an opaque type or something and interact with it through a limited set of known safe functions. Or buy amd and wait for unified memory I guess
|
# ? Sep 10, 2013 04:52 |
|
Malcolm XML posted:Or buy amd
|
# ? Sep 10, 2013 04:54 |
|
oqb don't bother with speed right now, doing what your doing is a v good way to learn how 3d rendering works and when you reach the point where you want it to be fast just switch to using opengl and shaders. with what you learned by doing it all by hand it shouldn't be hard
|
# ? Sep 10, 2013 09:27 |
|
Malcolm XML posted:Wrap it in an opaque type or something and interact with it through a limited set of known safe functions. yeah i shouldn't have much issue w/ it if im not an idiot. it's just that after years of p-langs, not having the compiler scream every time i screw up makes me distinctly uncomfortable cps: "just instantiate a thread for every pixel n let the gpu sort them out" two million threads and counting i like this gpgpu thing coffeetable fucked around with this message at 14:34 on Sep 10, 2013 |
# ? Sep 10, 2013 14:32 |
|
Sagebrush posted:i was thinking about your project old qwerty bastard and i realized the next step. you can't get good performance from an interpreted language the way you are using it. but what if you turned an fpga into a python-specific dedicated cpu? http://pycpu.wordpress.com/
|
# ? Sep 10, 2013 14:52 |
|
I've been doing some image processing with numpy these days and it rules so loving much. I'll try doing one of those sudoku solvers.
|
# ? Sep 10, 2013 15:47 |
|
Hi i just wanted to say that you are extremely stupid, and bad, and that you should commit suicide for the good of the forums and the world
|
# ? Sep 10, 2013 16:11 |
|
Yaoi Mandel posted:Hi i just wanted to say that you are extremely stupid, and bad, and that you should commit suicide for the good of the forums and the world How it's done, folks.
|
# ? Sep 10, 2013 16:24 |
|
echinopsis posted:that's what she said LMFAO
|
# ? Sep 10, 2013 17:24 |
|
HAHAHAHa *wipes tear from eye*
|
# ? Sep 10, 2013 17:25 |
|
what i've learned today: -rotation is easy -subdividing a triangle is easy although i have to write some sort of "3d object" class with a list of vertices and triangles so that i can do it properly (i.e. subdividing every triangle into 4 instead of 3 - adjacent triangles need to share the new vertices so i can't do this as just a list of unrelated triangles) -z buffers are like black magic and i apparently have no idea how to code one (right now it's sorting by midpoint distance which kind of works a little bit but not really) https://www.youtube.com/watch?v=8AakDr0R7mI ol qwerty bastard fucked around with this message at 18:42 on Sep 10, 2013 |
# ? Sep 10, 2013 18:39 |
|
ol qwerty bastard posted:what i've learned today: you can get the normal vector by picking two vectors along two edges of the triangle and doing a cross product. however it means that you need to mind the vertex order of your triangles so that their normals face outwards. you'll want to do that when you decide to do lighting anyway but triangle sorting is kinda messy and you need to avoid having too big triangles to limit the cases where they end up sorted badly zbuffer is simple: for each pixel of the screen, you store the nearest distance (ie Z) at which you have previously drawn a point from another triangle on it. you initialize it with a max distance value before rendering your frame. when you rasterize a triangle, for each pixel you calculate its distance (simply by interpolating the z coordinates of the vertices, assuming they are already transformed into camera space), and if the pixel you want to draw is closer to the camera than the one you drawn previously, you draw it and store its z value in the z buffer, otherwise you ignore that pixel. you also ignore that pixel if its z value is out of range for the z buffer (which means you need to pick some arbitrary min and max draw distances)
|
# ? Sep 10, 2013 20:08 |
|
pagancow posted:also i can never export my poo poo with realistic looking film grain online because you can't see it on video sharing sites lol 6mbps get hosed die
|
# ? Sep 10, 2013 20:19 |
|
Zlodo posted:you can get better results with triangle sorting by doing backface culling: basically dont draw triangles whose normal vector is facing away from the camera (if your object is convex this will actually render it correctly without needing sorting or z buffering). ah yeah i'd forgotten to do the backface culling thing; i even already have the code set up so it automatically generates a unit normal vector when the triangle is instantiated. i was using it for lighting but it'd be straightforward to tell it to dot with the camera vector and then if it's negative don't render i was definitely finding that triangle sorting is messy; i dread the day when i try to write code that will allow two intersecting triangles to render correctly i guess i need to write my own draw_triangle() code to do proper zbuffering if i have to do a pixel-level thing on it though; currently i'm just using the pygame.draw.polygon thing. ah well, that shouldn't be too hard
|
# ? Sep 10, 2013 21:00 |
|
Cold on a Cob posted:i thought it was not really racist at all b/c they basically said the white man "won" via right place, right time. but i guess diamond also doesn't really talk about china, etc. i could be mixing it up with other books though like 1491 here is guns, germs and steel in a paragraph 'eurasian peoples were much better positioned to develop advanced civilizations, not because of any kind of racial differences, but because their continent was oriented east-west, allowing long-range transmission of domesticated crops & animals within a similar climate band. why did europe end up controlling the world instead of china/india? gently caress if I know lol' (he added a half-assed afterword in one edition suggesting 'china was centralized, india was decentralized, maybe europe was just the right mix of centralization & disparate small states to make progress????' it's the dumbest thing and so of course 'business leaders' ate it up) anyway I get the feeling that a lot of the people who criticize guns germs & steel are basically jealous of its success/popularity and/or his dismissal of touchy-feely 'intellectual factors' (i.e. either 'great men' theory or 'THE ESSENTIAL WESTERN VALUES OF FREEDOM'). most of his theses seem pretty solid - there are some issues with grains (domesticatable grains were more widely available & more easily spread than diamond thought), but the theory as a whole seems to hold up pretty well anyway that's me replying to week-old posts current spare time project status: considering open-sourcing it so people can look and tell me how bad my code is. not like I'm planning to make any money off it anyway. (lol @ trying to make money from games)
|
# ? Sep 10, 2013 21:58 |
|
holy poo poo https://www.youtube.com/watch?v=Oie1ZXWceqM
|
# ? Sep 11, 2013 01:04 |
|
i've been thinking about this and have come to the conclusion that there are literal wizards involved
|
# ? Sep 11, 2013 01:13 |
|
|
# ? Jun 10, 2024 14:33 |
|
ol qwerty bastard posted:i was definitely finding that triangle sorting is messy; i dread the day when i try to write code that will allow two intersecting triangles to render correctly what you need to do is split one of the triangles along the plane defined by the other one, and it's not actually that hard. but nobody bothers doing that kind of things because that's what the zbuffer is for
|
# ? Sep 11, 2013 01:53 |