TDM 2.09: black menu with music playing

[HMart]

Is a 11 years old card with supposed support for D3D 11 (11.2) and OGL 4.4, so on the API front should be more than enough to play TDM but it apparently only has 512MB of VRAM and that can be a big limitation on the bigger missions.

Edited September 19, 2021 by HMart

[Alberto Salvia Novella]

Setting "r_glCoreProfile" to "0" prevents the game from launching.

[stgatilov]

6 hours ago, Alberto Salvia Novella said:

Setting "r_glCoreProfile" to "0" prevents the game from launching.

Mmm... probably some Mesa-based open-source Linux drivers?
Until recently, they did not support GL3+ with deprecated profile.

[Alberto Salvia Novella]

AMD CYPRESS (DRM 2.50.0 / 5.10.63-1-MANJARO, LLVM 12.0.1)

OpenGL 3.1

Mesa 21.2.1

[stgatilov]

23 hours ago, Alberto Salvia Novella said:

AMD CYPRESS (DRM 2.50.0 / 5.10.63-1-MANJARO, LLVM 12.0.1)
OpenGL 3.1
Mesa 21.2.1

As far as I remember, TDM 2.09 requires OpenGL 3.3 to start.

So either your OpenGL does not fit the requirement, or (as I guessed above) it has separate GL versions for core and deprecated profile. Perhaps version in core profile is newer than 3.3, but the version in deprecated profile is only 3.1

By the way, do you also have the problem with black screen in main menu and sound playing?

[Alberto Salvia Novella]

Yes.

[stgatilov]

3 hours ago, Alberto Salvia Novella said:

Yes.

We never found what was the reason behind this issue.
It broke after we started asking for GL 3.3 instead of GL 3.1, and it does not break in compatibility mode.

I don't think it is easy to revert back to the old version, and not sure how to debug it either.

[cabalistic]

18 minutes ago, stgatilov said:

We never found what was the reason behind this issue.
It broke after we started asking for GL 3.3 instead of GL 3.1, and it does not break in compatibility mode.

I wonder if it's related to index buffers. Menu elements are (afaik) the last elements that are drawn from a CPU-provided index list instead of a GPU index buffer. Far as I know, the spec doesn't require one, or at least I couldn't find anything to the contrary, but it's possible some drivers expect them, anyway?

[stgatilov]

26 minutes ago, cabalistic said:

I wonder if it's related to index buffers. Menu elements are (afaik) the last elements that are drawn from a CPU-provided index list instead of a GPU index buffer. Far as I know, the spec doesn't require one, or at least I couldn't find anything to the contrary, but it's possible some drivers expect them, anyway?

Maybe this?

And it seems that vertex attributes can only be specified with glVertexAttribPointer, which also seems to require offset into bound buffer object.

[stgatilov]

Citation from "E.2.2 Removed Features" in Core Specification:

[cabalistic]

Well, that'd be that, then. Unfortunately, fixing the menu elements to use an index buffer isn't an entirely trivial fix, or I'd have done it sooner, but I guess we'll have to put it on the agenda somewhere

[stgatilov]

4 minutes ago, cabalistic said:

Well, that'd be that, then. Unfortunately, fixing the menu elements to use an index buffer isn't an entirely trivial fix, or I'd have done it sooner, but I guess we'll have to put it on the agenda somewhere

I think this should fix it:

But I still get draw surfaces with invalid indexCache durign gameplay as soon as "objective complete" text pops up.

[stgatilov]

It seems that the easiest approach is going through all R_AddDrawSurf calls and adding the same thing everywhere.

I wonder why R_AddDrawSurf does not create caches automatically inside... but I guess not the best time to refactor it.

[stgatilov]

I think I fixed all calls in svn rev 9595.

Here are executables from the latest SVN, most likely compatible with the latest dev build.
It would be great if someone can verify that everything works properly now even with "r_glCoreProfile 2" (default).

[duzenko]

1 hour ago, stgatilov said:

I think I fixed all calls in svn rev 9595.

Here are executables from the latest SVN, most likely compatible with the latest dev build.
It would be great if someone can verify that everything works properly now even with "r_glCoreProfile 2" (default).

What is it supposed to fix? Core profile on pre-GCN Radeons? Why did all other drivers work if it's about the removed legacy stuff? I can test tomorrow, but I'm skeptical.

[Alberto Salvia Novella]

I will test too when I have a moment.

[stgatilov]

7 hours ago, duzenko said:

Why did all other drivers work if it's about the removed legacy stuff?

Because they are more permissive.
Maybe some popular game used CPU pointers in core profile, so drivers decided to support it.

Anyway, it is a big error for core profile, and it must be fixed.

[duzenko]

4 hours ago, stgatilov said:

Because they are more permissive.
Maybe some popular game used CPU pointers in core profile, so drivers decided to support it.

Anyway, it is a big error for core profile, and it must be fixed.

Agree about fixing the error

I just can’t imagine a game requiring a core profile and then doing client arrays. Oh, wait…

[Alberto Salvia Novella]

Probably since OpenGL <=3.1 allowed to use Client-Side Vertex Arrays, instead of Vertex Buffer Objects, drivers just followed the same capability. Even if it didn't make sense programmatically.

For instance they removed that possibility in OpenGL 3.2, likely just to make it explicit that you should not program that shitty, 80s graphics style.

[stgatilov]

17 minutes ago, Alberto Salvia Novella said:

Probably since OpenGL <=3.1 allowed to use Client-Side Vertex Arrays, instead of Vertex Buffer Objects, drivers just followed the same capability. Even if it didn't make sense programmatically.
For instance they removed that possibility in OpenGL 3.2, likely just to make it explicit that you should not program that shitty, 80s graphics style.

Yes, there was some bureaucratic mess related to deprecation during GL 3.0 and 3.1, which was finally resolved in GL 3.2. Like "deprecated but not removed yet" which later turned into "removed in core". It can easily explain it worked before version bump, but stopped working after it.

[Alberto Salvia Novella]

Is it rare that I find these topics entertainingly funny?

Visualizing the faces of those programmers coding graphics in RAM, while those of OpenGL talking about "well, how are we going to prevent this?"

I guess I'm a sophisticated guy.

[duzenko]

On 9/21/2021 at 9:02 PM, stgatilov said:

I think I fixed all calls in svn rev 9595.

Here are executables from the latest SVN, most likely compatible with the latest dev build.
It would be great if someone can verify that everything works properly now even with "r_glCoreProfile 2" (default).

You did

Spoiler

  

On 5/15/2021 at 5:04 PM, nbohr1more said:

@cabalistic it sounds like the GUI is using legacy GL somewhere?

You were right all along

It's still frustrating as heck that it didn't raise any GL errors

And I'm not sure what we're left with in the end - half-core profiles that do something specifically forbidden by specification? Crazy

[stgatilov]

Your screenshot shows revision 9540?

[duzenko]

2 hours ago, stgatilov said:

Your screenshot shows revision 9540?

Notools project does not update svn counter in build. It’s current svn.

[StudentManiac]

On 9/19/2021 at 10:39 AM, stgatilov said:

Then the same workaround should work: seta r_glCoreProfile "0" in darkmod.cfg

It is a rather old graphics card, right?

Yes it is and its second hand. My PC needs a serious overhaul/rebuild.

On 9/19/2021 at 12:44 PM, HMart said:

Is a 11 years old card with supposed support for D3D 11 (11.2) and OGL 4.4, so on the API front should be more than enough to play TDM but it apparently only has 512MB of VRAM and that can be a big limitation on the bigger missions.

It can be, my computer does struggle with the larger missions. I do have it in mind to upgrade at some point but for the time being its still alive and working. I do have other games I can't run at all on my computer so likely in the near future I will have to get on and upgrade things in it. Its just the issue of time, money and getting the right parts.

Ps. If you think that is an old computer part my screen is an old LG from I think 2004.