Were you referring to me or to someone else?...Sleipnir wrote: ↑The Intel graphics drivers that ship with windows doesn't include working OpenGL support. Grab the latest drivers from the Intel website and try again.
...it doesn't work on Windows XP SP3?!!
Problem is, that Intel claims that for my driver OpenGL works.Sleipnir wrote: ↑It was aimed at anyone with an Intel card. We have also recently fixed a bug which we believe to be the cause of the "GL Error: 1282" errors - this will appear in the next playtest, which should be released sometime in the next week.
Or at least this is what it claims. Is there any test to see if OpenGL works?... I've got no settings to see if it actually works. Or even NET framework. I've tried as well what rami proposed, and it still doesn't work.
Try this (unofficial) test build: http://sleipnirstuff.com/openra/devtest ... 110625.exe
The game works, yet the editor doesn't.Sleipnir wrote: ↑Try this (unofficial) test build: http://sleipnirstuff.com/openra/devtest ... 110625.exe
Tho...what could be the problem with .NET framework then?...
With this build OpenRA gets as far as opening the window and showing the loading screen ('articulating splines, etc), before crashing again, showing the "this application needs to close'' dialog.Sleipnir wrote: ↑Try this (unofficial) test build: http://sleipnirstuff.com/openra/devtest ... 110625.exe
Contents of exception.log:
Code: Select all
System.InvalidOperationException: GL Error: GL_INVALID_ENUM
at OpenRA.Renderer.Cg.GraphicsDevice.CheckGlError()
at OpenRA.Renderer.Cg.GraphicsDevice.CreateTexture()
at OpenRA.Graphics.Sheet.get_Texture()
...
Code: Select all
Graphics:
Renderer: Cg
Mode: Windowed
WindowedSize: 640,480
Code: Select all
System.Reflection.TargetInvocationException: Exception has been thrown by the target of an invocation. ---> System.InvalidProgramException: Unsupported GPU. See graphics.log for details.
at OpenRA.Renderer.Glsl.GraphicsDevice..ctor(Int32 width, Int32 height, WindowMode window, Boolean vsync)
...
Code: Select all
Unsupported GPU: Missing extensions: GL_ARB_vertex_shader,GL_ARB_fragment_shader
...
Driver: 6.14.10.4926 (apparently the latest)
OS: WinXP Pro SP3 UK
Some random GPU caps viewer (kludx) suggests the card does indeed lack vertex shaders (but does have pixel shader 2.0 support). Anything else we can check / do?
- Attachments
-
- GL_graphics.log.txt
- (1.32 KiB) Downloaded 814 times
-
- GL_exception.log.txt
- (1.03 KiB) Downloaded 722 times
-
- CG_exception.log.txt
- (876 Bytes) Downloaded 980 times
@twarpie:
The 950 will be slow, but it should be capable of running OpenRA without crashing using the Cg renderer.
That `GL Error: GL_INVALID_ENUM' exception is definitely caused a bug - either in OpenRA, or in your drivers (which is still our problem - we should work around these issues, or at least test for them early and exit gracefully with a decent error).
We will add some extra debug logging into the next build, which should hopefully give a bit more information on what is causing this.
The 950 will be slow, but it should be capable of running OpenRA without crashing using the Cg renderer.
That `GL Error: GL_INVALID_ENUM' exception is definitely caused a bug - either in OpenRA, or in your drivers (which is still our problem - we should work around these issues, or at least test for them early and exit gracefully with a decent error).
We will add some extra debug logging into the next build, which should hopefully give a bit more information on what is causing this.
Yes, I assumed as much with regards to the performance of the GMA.Sleipnir wrote: ↑@twarpie:
The 950 will be slow, but it should be capable of running OpenRA without crashing using the Cg renderer.
Is this going to be a 'release' build, or should I do a check out from git?Sleipnir wrote: ↑That `GL Error: GL_INVALID_ENUM' exception is definitely caused a bug - either in OpenRA, or in your drivers (which is still our problem - we should work around these issues, or at least test for them early and exit gracefully with a decent error).
We will add some extra debug logging into the next build, which should hopefully give a bit more information on what is causing this.
Seems the 950 does support
Code: Select all
GL_ARB_fragment_program
GL_ARB_vertex_program
Code: Select all
GL_ARB_fragment_shader
GL_ARB_vertex_shader
This is what the Cg toolkit, used by our Cg renderer, does.twarpie wrote: ↑Perhaps the shaders could be converted / compiled / rewritten (not a shader expert here)?
The fixes that I mentioned above will go into the playtest build that we intend on releasing in the near future. I may also release another unofficial test build with the fixes for people to try if the playtest is delayed much longer.
Ah right.
Ok. Will await the playtest then.Sleipnir wrote: ↑The fixes that I mentioned above will go into the playtest build that we intend on releasing in the near future. I may also release another unofficial test build with the fixes for people to try if the playtest is delayed much longer.
Thanks for all your effort btw, very nice project.
Just some further experimentation: placing the OpenGL32.dll from GLIntercept (source available) in the OpenRA directory allows OpenRA to run. Still using the Cg renderer.
Log file from GLIntercept logs a lot of INVALID_ENUM errors, but the game never crashes.
Obviously not a solution, or even a workaround.
Log file from GLIntercept logs a lot of INVALID_ENUM errors, but the game never crashes.
Obviously not a solution, or even a workaround.
See attachments. Archive contains all logs (OpenRA & GLIntercept). GLIntercept was configured with the 'gliConfig_FullDebug.ini'.Sleipnir wrote: ↑Interesting... Could you please post the GLIntercept logs? They'll hopefully point us at the calls that are failing.
The GL renderer crashes, Cg runs.
- Attachments
-
- GL_Logs.zip
- (11.99 KiB) Downloaded 708 times
-
- CG_Logs.zip
- (32.92 KiB) Downloaded 832 times
Any progress on this? Hope this doesn't come across as whiny: just anxious to play the game on my lacking gma950Sleipnir wrote: ↑Interesting... Could you please post the GLIntercept logs? They'll hopefully point us at the calls that are failing.
