GL4ES: another OpenGL over OpenGLES2 emulation - some tech. info and porting progress

@kas1e

Looking forward to your releases.
Greater combinations of Machines/Graphics cards running them may help isolate any problems .

Ok, let's start from the buggy ones (as find out roots and fix problems is more important than rereleasing another time quake3 :) ).

So, all the tests should be done on latest ogles2.library and warp3dnova.library. ogles2.library should be 1.22, and warp3dnova.library should be 1.58. It is very important to have those versions , as if not, tests may not help.

There is test archive of that LettersFall3 game:

http://kas1e.mikendezign.com/aos4/gl4 ... fall/find_the_bug/lf3.lha

There is how it looks like on my setup when i choice "options" or "howtoplay" or "highscores" or start a game (you can see distorted textures of text, they like doubled and shifted randombly):









My setup is x5k with HD VERDE (branded as some r250 if i remember right) and RadeonHD.chip 3.4 (some beta one, but bugs the same with all other previous versions too).

What is interesting, Daniel on his setup (if i remember right its some Sam), do not have this issues. Which mean it is kind of random at moment.

Will be nice if some of you can make tests on their setups. That bugs not always shows right away, you just need to go to the options of game , and try to navigate by "up/down" a bit, etc.

@kas1e

Quote:

It's the same with me
My config X1000 & Radeon R9 270 (RadeonHD v2.22).

I would add that in the case of Neverball compilations, which you threw on the server in March, I did not notice any changes in plus. The game starts without crash maybe once per 5 attempts, and then there are graphics corruption.

@mufa
Thanks for tests. First need to fix letterfall issues, then can check all the others.

So.. Its the same for you on another radeon with radeonhd2.22
What is your graphics.library version as well ? Did you on beta of os4, or on latest public one ?

Hi Kas1e,

Quick test on my A1X1000:
I've got the same distortions than on your screenshots.
I also remarked that when you select/unselect the game window, distortions change!
Resizing the game window also causes trouble in rendering with the game area overlapping window border!

Regards,

@Petrol
Thanks. Overlapping of window borders is probably SDL issue which if i remember well happens offten in other apps too, but good that you have same distortions. Can you point out what your graphics card, as well as version of graphics.library, as well as version of radeonhd.chip and if you on beta or on latest public version of os.

Sapphire Tri-X R9-280X (Tahiti XT - Radeon 7970) with 3GB
graphics.library 54.245
RadeonHD.chip 2.22
Beta system.

@kas1e

You're right, I tested game on beta AmigaOS 4.

I tested once again on the public version of AmigaOS 4.1FE u1, this problem does not occur, everything is OK.

@mufa
Do you mean that lettersfall works fine, and without distortions on public version of OS ? Are you sure ? Can you retest it again to be 100% sure ?

Why i ask, its because i trying to downgrade graphics.library before, and it didn't help.

Also be sure that on public version you have ogles2.library 1.22 , and warp3dnova.library 1.58.

@Petrol
Do you have ability to tests lettersfall on non beta version of OS ?

@kas1e

I tested on non beta system, and the problem seems to not occurs.
But, sometimes it crashed at exit, and when I ran it a second time, I was abble to see the same distortions.
So I tested it again on a non beta system without make it crashes. I ran it once, no distortion on the game screen, and I let the game screen opened. I then ran it again, and both game screens were corrupted!

@Petrol
And on non beta system you do tests with ogles2.library 1.22 and nova 1.58 ?

You also not need to run second copy, just try in first run navigate a bit over different options with up/down, goin to/from different menus and distortions probably will come.

All of this mean problem is on all setups, just seems shifts a bit in memory depending on how memory filled at moment of checking. And probably Mufa should see that distortions on his non-beta setup as well after playing more with it..

@Daniel
If you read this, can you try on your setup it as well by running few copies and navigate in them differently ?

@kas1e

Quote:

Yep!

@kas1e

I get the same garbled screenshots as you when clicking through the options menu.

X5000, Radeon 270
radeonHD.chip 2.22
ogles2.library 1.22
Warp3DNova 1.58

non-beta

@kas1e

I also get the same garbled screenshots as you when clicking through the options menu on both of my machines .
Both Non Beta AmigaOS 4.1FE Update1.

X5000/20 - Radeon 270X

radeonHD.chip 3.6
ogles2.library 1.22
Warp3DNova.library 1.58

Sam460ex - Radeon Rx560 (Polaris 11)

radeonRX.chip 1.11
ogles2.library 1.22
W3DN_GCN.library 1.58
Warp3DNova.library 1.58

@all
Thanks for tests. So its not random, its happens on all machines for everyone (x5k, x1k, sams), with beta and with non beta, with different versions of radeonhd and graphics.library, and with different gfx cards.

What mean its again one of 3 pieces cause issues: or gl4es, or warp3dnova or ogles2.library.

@Daniel
If you can, retest it on your setup again plz. It may looks good when just run it and go one time to options, but then, when you browse them a bit, go to/from few times, then issues appears. Just if you will be able to reproduce it as others, mabe you may have any ideas about.. In meantime i will try to dig in into with gl4es author, maybe he will help as usuall :)

@kas1e

Also all version of neverball in the archive linked to in bug report #0000318 work on both of my configurations above.

@Spectre660
For neverball later, first need to deal with letterfall issue

Some progress about lettersfall issue: gl4es author do capture of all calls which happens in game, and he see a lot of small drawing. Most drawing are composed of

glUniformMatrix4fv(....);
glDrawArrays(GL_TRIANGLE_FAN, 0, 4);

And some are a bit more complex when there is a texture binding, with a setting each time of GL_LINEAR for GL_MIN_FILTER and GL_MAG_FILTER

But then he found most interesting thing:



(press open in new tab, to see capture in full size).

There is the creation of texture (probably some SDL_ttf stuff, i.e. exactly where we have issues with that distored font), drawing of the texture, and immediatly after the draw the deletion of that texture! And he guess this is somehow an issue here with warp3d/ogles.

And to test that theory i just remove from the game all "glDelete*", calls, and violla, bug is gone.

So, for now he add workaround to gl4es for that, which i need to test today, and if all is ok, there bugreport for w3dnova or ogles2 need it. Through at moment do not know what the problem is : or it is immediately deleting after drawing, or, it is immediately deleting after drawing when there hundrend or thousands of small textures handled that way.

@kas1e

I think it's a Warp3D Nova bug. It looks like you can destroy a texture then create a new one that overwrites the old one before the GPU is done with it. Please file a bug report against W3D Nova, and include a download link with the test program.

Hans

@Hans
Quote:

Imho yes, as probably there is no any special traitement on glDeleteTextures in OGLES2

Quote:

Visually looks exactly like this, yes.

Quote:

Done, #0000342