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

	Bottom Previous Topic Next Topic
Register To Post

« 1 ... 40 41 42 (43)

kas1e

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

Posted on: 2024/5/5 8:25 #841

Home away from home

@Raziel
IMHO every developer who need fresh version can build it (i wrote instructions in gl4es sdk).

But, there lately were some regressions in gl4es cause massive slowdowns on os4, dunno if they were dealt with or not. To be on safe side just use version from my gl4es sdk

Join us to improve dopus5!
AmigaOS4 on youtube

davec555

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

Posted on: 2024/5/6 3:04 #842

Just popping in

@Raziel sent you an email.

Dave

Anonymous

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

#843

@davec555

Thank you

jarokuczi

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

Posted on: 2024/5/12 16:06 #844

Just popping in

Hello, fellow developers as an exercise I'm trying to port quakespasm to amigaos4. I'm kind of new to game development and graphic libraries. So far I managed to boot the game with some visual glitches.

Resized Image

The build crashes a lot and the performance is bad but it starts.

Two problems related to gl4es seem to be causing some of the above.

The first is strange as I can't find any function call with the below values. 0x8036 is defined in gl.h under the name GL_UNSIGNED_INT_10_10_10_2.


LIBGL: Unsupported source data type: 0x8036

LIBGL: swizzle error: (GL_RGBA, 0x8036 -> GL_RGBA, GL_UNSIGNED_BYTE)

LIBGL: Unsupported source data type: 0x8036

LIBGL: swizzle error: (GL_RGBA, 0x8036 -> GL_RGBA, GL_UNSIGNED_BYTE)

LIBGL: Unsupported source data type: 0x8036

LIBGL: swizzle error: (GL_RGBA, 0x8036 -> GL_RGBA, GL_UNSIGNED_BYTE)

It is used as type value for lightmaps if r_lightmapwide is set to 1, which is not the case.


const qboolean wide10bits = r_lightmapwide.value != 0;

    const GLenum type = wide10bits ? GL_UNSIGNED_INT_10_10_10_2 : GL_UNSIGNED_BYTE;

The other problem is about shader compilation (which explains lighting problems):


Warning: GLSL program failed to compile: ERROR: Code generation failed.

Error log:

Boolean uniforms aren't supported yet:

        offset: 0, size: 1, Bool UseAlphaTest

INTERNAL ERROR: OpBranchConditional instruction's source parameters are missing

Is missing one or more source parameters:

                34: OpPhi: Bool: tmp34 << [tmp24, lb5], [tmp33, lb25]

INTERNAL ERROR: OpBranchConditional instruction's source parameters are missing

Boolean uniforms aren't supported yet:

        offset: 1, size: 1, Bool UseOverbright

INTERNAL ERROR: OpBranchConditional instruction's source parameters are missing

Boolean uniforms aren't supported yet:

        offset: 2, size: 1, Bool UseFullbrightTex

INTERNAL ERROR: OpBranchConditional instruction's source parameters are missing

Reading uniforms with an offset that isn't a multiple of 32-bits in size isn't supported:

        offset: 3, size: 16, FloatVector4 color

INTERNAL ERROR: ExtInstruction is missing parameters

INTERNAL ERROR: Store instruction's source parameter is incomplete:

                    OpStore: : tmp99 >> result



Warning: GLSL program failed to compile: ERROR: Code generation failed.

Error log:

Boolean uniforms aren't supported yet:

        offset: 0, size: 1, Bool UseLightmapOnly

INTERNAL ERROR: OpBranchConditional instruction's source parameters are missing

Boolean uniforms aren't supported yet:

        offset: 1, size: 1, Bool UseAlphaTest

INTERNAL ERROR: OpBranchConditional instruction's source parameters are missing

Is missing one or more source parameters:

                41: OpPhi: Bool: tmp41 << [tmp31, lb26], [tmp40, lb32]

INTERNAL ERROR: OpBranchConditional instruction's source parameters are missing

Boolean uniforms aren't supported yet:

        offset: 2, size: 1, Bool UseLightmapWide

INTERNAL ERROR: OpBranchConditional instruction's source parameters are missing

Boolean uniforms aren't supported yet:

        offset: 3, size: 1, Bool UseOverbright

INTERNAL ERROR: OpBranchConditional instruction's source parameters are missing

Boolean uniforms aren't supported yet:

        offset: 4, size: 1, Bool UseFullbrightTex

INTERNAL ERROR: OpBranchConditional instruction's source parameters are mi

Does anyone have similar issues?

Thanks
jarokuczi

Anonymous

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

#845

@jarokuczi

regarding boolean uniforms, yeah, they are not supported (still)
did a bug report/feature requests years back, but none came of it

heres what they did to work around it on scummvm
https://github.com/search?q=repo%3Ascu ... lean%20uniforms&type=code

kas1e

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

Posted on: 2024/5/13 3:05 #846

Home away from home

@jarokuczi
Can't remember about swizzle errors (i were sure Daniel fix it all before in ogles2, but maybe you need to try to contact with him about), but as Raziel said boolean uniforms is not supported, but as they _very_ easy to be workarounded, Hans (in case of shaders errors it's mostly warp3dnova area) just didn't find motivation to implement them : and implementation of them will take lots of time as he says, and better to just fix shader's code manually (that matter of changing few bits only, as on Raziel's example)

Join us to improve dopus5!
AmigaOS4 on youtube

Anonymous

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

#847

@kas1e

not sure where else to post, but i guess i might have found a bug in gl4es?

please take a look: https://github.com/scummvm/scummvm/pul ... 6#issuecomment-2453049173

thank you

Maijestro

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

Posted on: 7/25 16:58 #848

Just can't stay away

Hey everyone,

I'm using gl4es to run OpenGL over GLES2 on AmigaOS4 (Warp3D Nova via ogles2.library). Rendering works nicely, but as soon as the render window opens, a second "newlib stdout" console pops up too. It's filled with gl4es' own startup log — things like:


LIBGL: Initialising gl4es

LIBGL: Using OGLES2.library v3 revision 3

LIBGL: OGLES2 Library and Interface open successfuly

LIBGL: Targeting OpenGL 2.1

LIBGL: NPOT texture handled in hardware

LIBGL: try to use VBO

LIBGL: Force texture for Attachment color0 on FBO

LIBGL: Current folder is:...

I tried building with -DGL4ES_SILENCE_MESSAGES (double-checked it's actually on the compile line, did a clean rebuild too), but the window keeps showing up with the same output.

Is there some other spot printing to stdout/stderr that ignores that define — maybe in src/agl/ or loader.c/hardext.c? Would love to get rid of it completely. Happy to test a patch if anyone has a pointer.

Thanks!

MacStudio ARM M1 Max Qemu//Pegasos2 AmigaOs4.1 FE / AmigaOne x5000/40 AmigaOs4.1 FE

davec555

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

Posted on: 7/25 17:45 #849

Just popping in

@Maijestro look in the gl4es docs. I think there is an env you can set to mute that output

Register To Post	« 1 ... 40 41 42 (43)
	Top Previous Topic Next Topic

Currently Active Users Viewing This Thread: 1 ( 0 members and 1 Anonymous Users )