Jump to content

  • Log In with Google      Sign In   
  • Create Account


#ActualBrother Bob

Posted 22 July 2013 - 12:03 PM

I was just curious why my app still worked with immediate mode, when I thought immediate mode was dropped sometime after version 3.0.

 

It's always been my experience that if the code is in any way valid, nVidia, ATI, and Intel drivers will run it and even correct errors to the best of their abilities.  One of the design goals of OpenGL has always been backwards compatibility. 

 

Here is a page with people that are attempting to do what you are doing.  There are several extensions available to help you with this.

 

http://stackoverflow.com/questions/13062149/how-do-i-make-this-simple-opengl-code-works-in-a-lenient-3-3-and-4-2-profile

http://stackoverflow.com/questions/4693017/forcing-opengl-core-profile-only

 

You should consider what you are doing to be experimental with only limited truly real support.  The various video card manufacturers are not required to have fully implemented features in hardware before claiming that they meet the spec.  If the GPU's drivers can run a particular feature on the host CPU then they meet the spec so far as the "law" is concerned. 

 

It could be several years before all the various manufacturers really do have full hardware accelerated support that works consistently on all brands without huge portability issues popping up.  Unless you are interested in a particular 4.0+ feature such as 5 stage tessellation then you are asking for trouble by forcing a profile that has not been out in the wild for very long. 

Most of the example code that you find right now will be full of bugs, it will work on one GPU and not another.

This is likely why Apple appears to be behind the curve with support for these things.  It's often better to wait than to push something that will fail on someone's else's computer simply because that other person has a different driver or GPU installed.

 

If you do want to be a trail-blazer here, then just make sure you are constantly verifying everything you do on as many different GPU's as possible.  Check them all for stability as well as performance benefit. 

Try every possible method and extension to find the ones that work consistently across the board.  I'm not convinced that this is even possible at the current time with the 4.0+ specs.  We still have not seen this happen in the open source community with the 3.0-ish stuff.

 

It can take years for new technology to be decimated through the open source community in a form that is consistent and reliable,  but if you are one of the people who will help to pull this off then I say go for it. 

Just be forewarned and you will be mostly prepared is all I'm saying here.

 

Here is an example of what's going on here on a different thread for some 3.0-ish techniques.  You should expect much worse for what you are going after,4.0+ , but you can also learn from his method.  Just don't be so quick to blame one particular brand.  I have no doubt at all that some other person is having the reverse problem with both these companies.  It's the nature of forging a new path.

 

http://www.gamedev.net/topic/644327-custom-pseudo-instancing-faster-than-real-instancing-on-nvidia-cards/


#3marcClintDion

Posted 22 July 2013 - 03:19 AM

There is far too much arrogance and out right abuse by site moderators, they are teaching other people to behave this way.  The posts I've made will all be shorty removed and replaced with this notice.  Game development is not the only thing being taught here, bad behavior is being taught as well.


#2marcClintDion

Posted 05 July 2013 - 05:55 AM


I was just curious why my app still worked with immediate mode, when I thought immediate mode was dropped sometime after version 3.0.

 

It's always been my experience that if the code is in any way valid, nVidia, ATI, and Intel drivers will run it and even correct errors to the best of their abilities.  One of the design goals of OpenGL has always been backwards compatibility. 

 

Here is a page with people that are attempting to do what you are doing.  There are several extensions available to help you with this.

 

http://stackoverflow.com/questions/13062149/how-do-i-make-this-simple-opengl-code-works-in-a-lenient-3-3-and-4-2-profile

http://stackoverflow.com/questions/4693017/forcing-opengl-core-profile-only

 

You should consider what you are doing to be experimental with only limited truly real support.  The various video card manufacturers are not required to have fully implemented features in hardware before claiming that they meet the spec.  If the GPU's drivers can run a particular feature on the host CPU then they meet the spec so far as the "law" is concerned. 

 

It could be several years before all the various manufacturers really do have full hardware accelerated support that works consistently on all brands without huge portability issues popping up.  Unless you are interested in a particular 4.0+ feature such as 5 stage tessellation then you are asking for trouble by forcing a profile that has not been out in the wild for very long. 

Most of the example code that you find right now will be full of bugs, it will work on one GPU and not another.  

This is likely why Apple appears to be behind the curve with support for these things.  It's often better to wait than to push something that will fail on someone's else's computer simply because that other person has a different driver or GPU installed.

 

If you do want to be a trail-blazer here, then just make sure you are constantly verifying everything you do on as many different GPU's as possible.  Check them all for stability as well as performance benefit. 

Try every possible method and extension to find the ones that work consistently across the board.  I'm not convinced that this is even possible at the current time with the 4.0+ specs.  We still have not seen this happen in the open source community with the 3.0-ish stuff.

 

It can take years for new technology to be decimated through the open source community in a form that is consistent and reliable,  but if you are one of the people who will help to pull this off then I say go for it. 

Just be forewarned and you will be mostly prepared is all I'm saying here.

 

Here is an example of what's going on here on a different thread for some 3.0-ish techniques.  You should expect much worse for what you are going after,4.0+ , but you can also learn from his method.  Just don't be so quick to blame one particular brand.  I have no doubt at all that some other person is having the reverse problem with both these companies.  It's the nature of forging a new path.

 

http://www.gamedev.net/topic/644327-custom-pseudo-instancing-faster-than-real-instancing-on-nvidia-cards/


#1marcClintDion

Posted 05 July 2013 - 05:45 AM


I was just curious why my app still worked with immediate mode, when I thought immediate mode was dropped sometime after version 3.0.

 

It's always been my experience that if the code is in any way valid, nVidia, ATI, and Intel drivers will run it and even correct errors to the best of their abilities.  One of the design goals of OpenGL has always been backwards compatibility. 

 

Here is a page with people that are attempting to do what you are doing.  There are several extensions available to help you with this.

 

http://stackoverflow.com/questions/13062149/how-do-i-make-this-simple-opengl-code-works-in-a-lenient-3-3-and-4-2-profile

http://stackoverflow.com/questions/4693017/forcing-opengl-core-profile-only

 

You should consider what you are doing to be experimental with only limited truly real support.  The various video card manufacturers are not required to have fully implemented features in hardware before claiming that they meet the spec.  If the GPU's drivers can run a particular feature on the host CPU then they meet the spec so far as the "law" is concerned. 

 

It could be several years before all the various manufacturers really do have full hardware accelerated support that works consistently on all brands without huge portability issues popping up.  Unless you are interested in a particular 4.0+ feature such as 5 stage tessellation then you are asking for trouble by forcing a profile that has not been out in the wild for very long. 

Most of the example code that you find right now will be full of bugs, it will work on one GPU and not another.  

This is likely why Apple appears to be behind the curve with support for these things.  It's often better to wait than to push something that will fail on someone's else's computer simply because that other person has a different driver or GPU installed.

 

If you do want to be a trail-blazer here, then just make sure you are constantly verifying everything you do on as many different GPU's as possible.  Check them all for stability as well as performance benefit. 

Try every possible method and extension to find the ones that work consistently across the board.  I'm not convinced that this is even possible at the current time with the 4.0+ specs.  We still have not seen this happen in the open source community with the 3.0-ish stuff.

 

It can take years for new technology to be decimated through the open source community in a form that is consistent and reliable,  but if you are one of the people who will help to pull this off then I say go for it. 

Just be forewarned and you will be mostly prepared is all I'm saying here.


PARTNERS