| |
| |
| |
|
Page: 1 2
Comments:
<0> where can i find opengl 2.0 headers
<1> So.. how large is a typical object list memory in a graphic card?
<2> hi. i intend to develop an opengl application that can be compiled for qt or win32. in a visual studio 2005 solution, i have 3 projects. one is the core project, the other two platform specific ones - one a qt, one a win32 project. i a would like to depending on the project selected (either qt or win32) select in the core either one header file or the other (including the qt or windows headers). how would this be done in a visual studio?
<3> whats one efficient and portable way to render to a texture
<4> Hi
<4> Can anyone recommend a good C++ scene graph?
<5> hrmmmmm, ssg in plib?
<4> What are those?
<5> ... ssg is a scene graph
<5> plib is a library set for portable ogl shtuff... made for games
<4> URL?
<5> google.com ?
<4> No, that's for google.
<5> ...
<4> :)
<4> Sure, thanks!
<5> (but I'm not a c++ guy, so I d'no if that piece of software is crap or not, it's just the first one that jumps to mind)
<4> So how do you use ogl?
<5> via C or scheme
<4> scheme?!
<5> um, yes?
<4> Why not learn C++?
<5> um
<5> I know c++
<5> I used to use it a lot, and tell people that it was a good language
<5> but I got better about a decade ago
<5> now I'm recovered :)
<4> And scheme is better than C++?
<4> Or C?
<5> for some things, much
<4> Like? Recuresion?
<6> I wouldn't say, but C and Scheme are very good languages for different purposes
<6> say that*
<4> I wouldn't say that either!
<5> pick the language closest to your problem, it makes life easier *shrug*
<5> the main issue I have with c++ is that the problem is c++... :D *duck*
<4> ?
<6> C++ is a fairly horrible language, overall
<4> Why's that?
<4> I think C++ is a wonderful language.
<6> Do you know anything else?
<4> Yes.
<5> C is a very focused language for very specific reasons... c++ is a crude attempt to bolt something almost imitating OO onto C... a bad combo
<6> C, Lisp, Ruby, Smalltalk?
<4> How does that answer the question?
<5> try smalltalk
<4> What's wrong with the OO in C++?
<5> objc is nice, too
<5> uh, because it totally misses the main ****ing point of OO, and now we have eight zillion ****tards who don't understand principle OO :(
<6> C++ has serious processing and memory overhead too, completely inappropriate where performance matters, yet it's low-level
<4> Erik - Please elaborate.
<7> C++: where your friends have access to your private members
<4> Maloeran - What are you talking about?
<5> objects a a dandy way of holding a thing, but the point was complex messaging... which is totally missing in the faux oo that c++ and java implement
<5> http://steve-yegge.blogspot.com/2006/03/execution-in-kingdom-of-nouns.html is an interesting read :)
<6> yaron, I'm saying that C++, the way it is commonly used, produces code far slower than good C code would be. It's possible to use C++ properly of course, and that implies sometimes having to rely on the C subset
<4> Erik - So what's the point of OO that C++ is missing?
<5> messaging.
<4> Erik - ?
<5> ...
<7> you grok 'message' ?
<4> Maloeran - You can use pretty much any language improperly. That doesn't mean C++ is bad. Actually, you can optimize things much more easily and conveniently with C++.
<4> Erik - Never heard 'messaging' as part of OO. So please explain.
<6> I think one has to know ***embly in depth to realize the horrible mess that C++ code produces, and why
<5> yaron: not without writing a large article in chan. Try going to google. Try downloading 'squeak' and learning how to program on it.
<4> Malorean - What on earth are you talking about?
<6> I'm talking about code efficiency. When I code in C or C++, I care about performance ; otherwise I use Lisp
<4> Okay, you guys are just talking in clues... Up until now, I haven't understood one single argument you mentioned.
<5> C isn't exactly a shining example of elegant machine code generation when not on a pdp11.. :)
<4> Maloeran - And what's wrong with C++'s code efficiency?
<6> There's just serious overhead due to the language's design ( unless one agrees not to use cl***es when inappropriate, and frequently rely on the C subset )
<4> Like what overhead?
<4> Sorry.. but I can't understand one thing you're saying. You're not giving any examples.
<5> vtables, allocation hoops, etc
<5> gitcher self sum edjimucation :D
<7> Maloeran: *i* probably gotta brush up on my ***embly skills soon, gotta program an atmel microcontroller :)
<4> How would you implement virtualization then?
<5> why would you really need virtualization?
<4> And you can always write a good memory allocator.
<5> yeah, you can always reinvent wheels
<6> No GCC port for that chip, Kutulu?
<6> Most of these have some sort of C compiler these days
<7> Maloeran: no idea. Probably. I honestly know SFA about them.
<7> i've got to find out the exact model of them that i'll be getting so i know which of the dozen or so sdk's to get
<5> some day, I'll do something with those things, heh
<6> Eheh. Let's play with FPGAs, ``Erik, that seems much more interesting
<7> friend of mine has a handful of atmel's and all the riggings locally so that's what i'm gonna use ;)
<5> unfortunately, the sdk's and stuff are all winderz only
<7> ``Erik: yeah, BLECH.
<5> I should call 'em up and say "please to be givink me macosX tools, da?"
<7> Maloeran: an FPGA would be a bit overkill for this input device :) though they are interesting devices for sure
<7> ``Erik: do eet
<6> When do we begin with the raytracing hardware, ``Erik? :) I wouldn't mind doing that at the same time as the software, there's no way I can take up to 31st March to complete that task
<5> heh, I guess we'll talk about it, uh, next week?
<6> *nods* Excellent
<7> Maloeran: have you tried rayforce on the latest CPUs?
<5> and after all the beaurocratic bull**** red tape, it MIGHT take till the end of march...
<7> intel core somesuch
<5> core duo
<6> I haven't Kutulu, I'm very much biased for AMD still.. :)
<6> ``Erik, and what about a non-paid initiative?
<7> Maloeran: same here, heh ;) but, i've heard good things about the performance
<5> for hw? I can't use the fpga I have for 'personal' projects... misappropriate of gov't resources
<6> Ew. FPGAs are cheap anyhow
<7> Maloeran: the dev kits aren't always cheap though?
<5> the experimentors board is a few hundred bucks, it came with a 'demo' of the sdk, and the licensed sdk/ide is a couple grand
<5> for the one I got, which boasts being an fpga at dirt cheap prices, heh
<5> cyclone 2 or something
<7> xilinx?
<5> no, uh
<7> altera
<6> For a moment, I thought buying the sdk thing software wouldn't be necessary
<5> altera nios 2 eval kit, with a cyclone ep1c12 fpga
<7> neat
<5> couple years old, but virtually unused
<6> How much does it cost to get an ASIC design printed, Erik?
<5> uhhhh, that depends on process, size, etc... masks are usually in the hundreds of thousands, I think?
<5> <-- never done it
<7> Maloeran: after raycasting in an FPGA, ***uming for graphics, shading and the rest would be done externally?
<5> some asian fabs do short runs 'cheap' (like, mask, chips, packages, all under a million)
<6> Visualisation would be done by fpga entirely, way too much waste in sending raw results back to the processor
<5> a lot of limited run circuits use fpga's for 'custom' circuits instead of asics... due to price
<7> scene setup -> fpga -> framebuffer ?
<6> Pretty much
<5> 'visualization' is not the primary purpose for the raytracer you're supposed to be writing, mal :)
<7> ``Erik: gigabyte's iram ramdisk thinger uses an FPGA on the card
<6> Aw :), returning raw results from the fpga is always an option, of course
<7> Maloeran: what about shaders?
<5> I think most hw that you can 'flash upgrade' has an fpga somewhere in it... heh
<5> other than cmos thingies
<7> ``Erik: wouldn't surprise me. Makes sense for some apps, like the iram being able to be upgraded via software to take larger or faster RAM sticks, or something
<6> Kutulu, it should all be on hardware, ideally
<7> Maloeran: of course, but for shading there'd be a lot more stuff to implement in the FPGA
<7> hmmm..
<5> the problem is bandwidth to the fpga... not many fpga's can hold several gigabytes
<7> fortunately raytracing is an embarrasingly parallel problem, so a scene can be cut up to survive ram bounds
<7> (i think.)
<5> then you don't get effective load balancing
<5> and a single ray will probably traverse several devices...
Return to
#opengl
or
Go to some related
logs:
#politics
teflon trackpad
#mirc
#politics
#narcotics
FARTSAUCE
#stocks
xorg 3007fp
#beginner
#heroin
|
|