Jump to content

  • Log In with Google      Sign In   
  • Create Account

Geri

Member Since 06 Jul 2006
Offline Last Active Yesterday, 12:13 PM

#5139215 Software rendering tutorials/examples

Posted by Geri on 15 March 2014 - 06:24 AM

JohnnyCode: i disaggree. maybe in *theory*, you can get the biggest performance, but using gpgpu would so much limit the number of compatible systems and would make so many extra work and compatibility issue, it would basically double the required work, and would make the whole thing almost unsellable, becouse there is a limited number of users with proper computers to run it. maybe such minor works like antialiasing, and other various filters effectively can done with gpu in this case, but that would not significantly boost the rendering ,,pipeline'', and should be maked with the possibility to disable it and use the software fallback.




#5138771 Avoiding dark areas in ray tracing

Posted by Geri on 13 March 2014 - 04:04 PM

have some fake color multiplier even on the dark locations, otherwise you will get this. i use 0.6 or 0.8 ambient light multiplier for shadowed places, those are ideal for my taste, but the practicular number basically varies on every scene.




#5138770 Software rendering tutorials/examples

Posted by Geri on 13 March 2014 - 03:54 PM

i dont suggest to study these outdated technologies like the q2. those optimization tricks will not work any more, and keep in mind that we have superscalar cpu-s now. also remember that you will need an agressively multithreaded code to achieve fast speed with nice quality. just write your renderer, and if you find a part that is too slow, then meditate/study/doodle that practicular problem.

 

http://geri.comeze.com/darktower/

 

my last (software rendered) free game, done for a game competition. it uses ray tracing, runs playable in 320x160.




#5128506 Writing a software renderer

Posted by Geri on 03 February 2014 - 01:25 PM

i dont really suggest to copy codes to do your renderer, you should fully understand and have control over the whole process instead.




#5118891 TitaniumGL, opengl multiwrapper for your game (opengl,d3d,multicore soft-render)

Posted by Geri on 23 December 2013 - 01:49 PM

no, they dropped that plan

 

vista and win7 have the same opengl fallback that windows xp had. both xp, vista and win7 have a software rendered opengl fallback, and a special d3d9 based fallback which is deactivated. this last does not work with generic softwares at all, becouse its deactivated and not oficially exist, it cant be accessed, only can be turned on by installing some special tools to allow it. the opengl software renderer of microsoft, which is the generic fallback, can render a few thousand polygon around half or 1 fps even on the newest core i7 cpus, so it not really works in practice, and does not works with generic games or softwares at all. microsoft wrappers supports opengl 1.1 without any extension except to control vsync and abgr, and they crash from the most texture formats or even from the most basic functions. microsoft have this fallback unchanged since ~2001. in practice, both its useless, no software is compatible with them. microsoft was unable to deliver they opengl 1.5 wrapper/implementation at all, they discontinued the development after a few month of struggling. they opengl 1.5 plans are not available any more. http://msdn.microsoft.com/en-us/library/windows/desktop/ee417756%28v=vs.85%29.aspx

 

(the case is the same with directx, they originally planned to have a fast and good multithreaded software renderer fallback for directx, but they cancelled that too, software renderer only available after installing directx sdk only after special linking and initialization of the software, its cant be used in practice. also thats 0.5-1 fps so its useless in practice again)




#5110997 Design help on software rasterizer / renderer

Posted by Geri on 21 November 2013 - 09:05 AM

2. if you write your own renderer, why you even want to write generic vertex and pixel shader support? you can just code your effects inside the renderer pipeline for your pleasure.




#5110988 AMD's Mantle API

Posted by Geri on 21 November 2013 - 07:53 AM

MJP:

 

Do you really think that EA would be investing development resources towards developing a Mantle version of their flagship engine if the whole thing was just a smoke screen put up by AMD?
 

why not? especially if amd pays enough for porting the graphics engine for they new api? i dont see why a desision of EA would make any relevance in the current situation, unless if some developer is EA-fan. its like drinking coffe is good, becouse hitler have drinken a lot of cafe too biggrin.png

 

hurray to amd for reinventing wheel and making pointless hype around it

 

and yes, we still talking about gpu-s - and only about gpu-s. if we would not have gpu, we would not need to use such apis at all. based on that previous measure, opengl almost can do the same performance as they promise in mantle. so i still dont see why mantle would be good for me in reality.

 

becouse opengl can do 100k+ draw call playable too on decent hw. i should measure directx too, but i am under linux at the moment.




#5110840 AMD's Mantle API

Posted by Geri on 20 November 2013 - 03:24 PM

i measured how many draw call i can do with opengl

i used one triangle in every draw call

after ~half million of draw calls i do swapbuffer to avoid crash of driver

 

tradicionalis pipeline: 13 331 229 call/sec
vertex array pipeline: 11 674 026 call/sec
vertex buffer object pipeline: 5 391 546 call/sec

OpenGL vendor string: Advanced Micro Devices, Inc.
OpenGL renderer string: ATI Radeon HD 4200
OpenGL version string: 3.3.11672 Compatibility Profile Context
OpenGL shading language version string: 3.30

 

uname -r -a
Linux a1 3.4.6-2.10-desktop #1 SMP PREEMPT Thu Jul 26 09:36:26 UTC 2012 (641c197) x86_64 x86_64 x86_64 GNU/Linux
 

http://legend.uw.hu/drawcalltest.bin.zip




#5098351 AMD's Mantle API

Posted by Geri on 02 October 2013 - 03:04 PM

i got a message from an insider. says that for now, amd gives the sdk only for its special partners. amd says that sdk will be public only in the next year, or later.




#5097710 AMD's Mantle API

Posted by Geri on 29 September 2013 - 07:35 PM

Managed code can be as fast as, or even faster than, native code sometimes.

 

managed code never in real life code was faster than native code ever, and probably never will be. native code outperforms the managed code in 2-10 times according to the REAL benchmarks by programmers who was at least able to turn on compiler optimisation flags at least (most c# evangelists even fail at this basic point). but who want speed-ciritic code, must use true programming languages (oh god, i will get a lot of votedowns from c# warriors, but at least, i am honest :D). however this is an offtopic here, i dont know why somebody put c# up in this thread aniway, its not a good way to use it as a synonym in this question. (and also i am sure c# programmers will have library to access this api too, so its irrelevant, lets forget about it alreday.)

 

 

 

You're also wrong about OpenGL

 

OpenGL ES, a different api? yeah, like directx9 is a different api from directx10....

this is just riding on the words after loosing real arguments, however, we isnt even started any debate, so this discussion starts extremely wrong alreday :P

 

 

first of all, you don't optimize OpenGL because OpenGL is not software, it's a specification

 

oh great, you should be some kind of brain surgeon i think

i would not think that you in reality in clear how opengl works alreday (no, i will not teach it even if you would pay for it), but congratulation for pointing out that opengl is a specification - nobody sayd otherwise - you maybe should apply with this fact to be a professor on an university, or something, with writing to a doctor dissertation that opengl is a specification :D  (no offense!)

i have talked about opengl driver vendors of course, you may have heard about drivers at least.

 

 

including the very first ARB-approved extension - multitexture

 

thats the most unsignificant expansion to opengl according to the api speed. becouse the first heavy thing for the api was the vertex array and list capability in opengl 1.1, then the element arrays and later, vertex buffer objects. these made the api to be like a sphagetty inside. however the es versions and the later core profiles over version 3 does not have the older pipes at least, only the most significant for the actual generation they meant for.

 

 

One is that Mantle only tackles one bottleneck, whereas many other bottlenecks exist in a graphics pipeline.

 

yes. there is many kind of bottlenecks in opengl, however, mantle will mostly loose the speed due to architectural reasons later, this is unavoidable. if we start to compare opengl and mantle on future hardwares, we see that mantle will not have any positives according to opengl that baiscally scales from anything from the last two decade to anything in the next x0 years at least.

 

 

Likewise, if it turns out that you need significantly more complex code to use the API correctly

 

yes, they will need to code manually a lot of thing that opengl probably offered for them, however, if they coming from opengl es2, they probably alreday needed to have these specific code for example to tesselate they quads to triangles, and manage matrices. i suggest careful skepticism in the direction of mantle, i am afraid that we will see the reinvention of opengl es3. at least i hope they dont want to give us an another useless and unsupported opencl fork, and they will give a true graphics api, becouse if i would want to have fully generic programing, i would code it on cpu (and yeah, i alreday doing that).




#5097664 AMD's Mantle API

Posted by Geri on 29 September 2013 - 03:17 PM

samoth: c# game developers are mostly beginners, or not programmers at all. the games created in c# is very limited due to the slowness of the language, but they do speed-critic parts in c++ and call it as dlls, such as game engines. in my oppinion, the problem with mantle is that it will become outdated after 1 or 2 generation of newer gpu-s, since the internal working of the gpu will probably change so much that the api will not be fast as they originally meant to, becouse it will not fit so well to the hardware. after a few generations, the api will produce the same speed issues like opengl (or directx) does. however, opengl is widely used in every platforms and hardly optimized in the past 20 years by the gpu manufacturers, so mantle will be not so largely faster in its long-term life cycle as they originally mean it, but it will be very limited compared to opengl that has many many features, compatibility even with the age of 3dinosaurs to run older, newer, whatever games on a common code base. as they dont give out any specifications, i assume, its just purely crap, until i dont get a freely downloadable specification that proofs otherwise.




#5097513 AMD's Mantle API

Posted by Geri on 28 September 2013 - 06:03 PM

is the function list or technical documentation available somewhere?




#4839981 [discussion]DirectX 9 or 10 or 11?

Posted by Geri on 25 July 2011 - 07:47 AM

9


#4819853 DirectX 9.0 vs OpenGl 2.1

Posted by Geri on 05 June 2011 - 02:44 PM

Posted Image


Posted Image


#4814666 Hints for my new engine (v3)

Posted by Geri on 23 May 2011 - 11:44 AM

is part of any game engine
Yes it is, but i currently want ideas mostly for the graphics engine, and technologies depthly inside the code. Maker3D relies on this also, but Maker's editor or script language / etc does not affects this. I am not needing any suggestions related on this at the moment. So i am currently only curious about the things i have mentioned.

there are a lot of already great opensource
I not writing everything myself ofc. I using some existing format loaders when necessary. But i dont think i would let any third party code in the rendering core, even if it is opensource or public domain, even if its mutch faster. If a bug appears in it, fixing would be very painfull, maybee also would strict me away from some later platforms, or would decrase the scalability or the flexibility of my code.

Is there any game in development with Maker3D?
I have no such informations.




PARTNERS