What! Are you saying I must have only one of two cards in order to use OpenGL?
No, what I am saying is that if you want hardware accelerated OpenGL in Linux then you should buy one of these two cards because you will be supporting vendors who give out specs.
I'm sorry, but this is unacceptable. I have a Creative Graphics Blaster Exxtreme. I will have hardware 3D acceleration in Xfree86 4.0, but not having the support now, I am keenly aware of those who won't have it.
What type of graphics chip does it have? Nvidia? The Nvidia OpenGL drivers are horrible and it's all Nvidia's fault for not releasing specs.
To reiterate, a P100 and SVGA 1Meg video memory under Windows or OS/2 has better OpenGL performance than a K6-400 with 3Dnow, a GB Exxtreme and 8Megs of video memory on Linux.
That's fine but writing a super efficient software implementation is not something that 3D graphics hardware companies like SGI or Nvidia are going to care about. I didn't know that software Mesa was so slow but it is simply a problem that most people do not care about due to the cheap hardware accelerators available for $50.
You're right though, one of the hallmarks of Linux is that you can get decent performance out of old hardware but one of the other hallmarks of Linux is that if you don't like something fix it yourself so I suggest that if you really care about a fast software OpenGL implementation that you do it yourself.
While I respect your viewpoint I simply do not think that even the fastest software implementation will come anywhere close to the slowest 3D hardware with reasonable (i.e. not nvidia) drivers.
Under Windows, OS/2 or BeOS, OpenGL performs satisfactorily with a generic SVGA card with no acceleration.
Running what? An OpenGL screensaver?
Under Linux, you need a specific card to get anything that even comes close. Right now, I have a 3D OpenGL accelerated card, but OpenGL is useless under Linux.
What card do you have? OpenGL on Linux with a 3dfx or Matrox card is pretty good now.
While I think an efficient software implementation is a worthy goal, I don't think it is worth it. A year from now it is going to be virtually impossible to find a 2D only card. Even the integrated graphics on chipsets have decent low end 3D support and with good 3D cards like Matrox G400's available for under $100 I see the trend continuing.
Unfortunately most of the 3D card makers do not give out specs on the 3D part of their cards except under NDA. Matrox G[12]00 cards are supported for 2D under XFree86 but no Mesa support. Same for nVidia RivaTNTs.
I hope some of the companies release 3D specs soon, there *is* a market, I just bought a Voodoo2 solely for Linux Quake II.
1. Carmack was praising Apple's adoption of OpenGL more than the ATI Rage 128 which is also available for PC's in an AGP2X version rather than PCI66.
2. The G3's still have PCI slots and Darryl Straus (sp?) was commenting about porting Glide to Linux/PPC so you could put a Voodoo2 in there.
3. 3Dfx's Voodoo3 will be released exclusively by themselves. S3 is a competitor and makes the Savage3D chip. 3Dfx recently bought STB which makes finished boards utilizing 3Dfx and other chips from vendors such as nVidia.
Intel's "no more socket" talk was all marketing BS. If you look at a Celeron 266, 300, 300A, or 333 you'll see that the PCB is virtually empty. The point of the PCB was for the P-II cache which is now gone or on chip so get rid of it and put the CPU back in a socket. The packaging is much cheaper. If they made more slot instead of socket they'd just make less money.
Slashdot Mirror
No, what I am saying is that if you want hardware accelerated OpenGL in Linux then you should buy one of these two cards because you will be supporting vendors who give out specs.
I'm sorry, but this is unacceptable. I have a Creative Graphics Blaster Exxtreme. I will have hardware 3D acceleration in Xfree86 4.0, but not having the support now, I am keenly aware of those who won't have it.
What type of graphics chip does it have? Nvidia? The Nvidia OpenGL drivers are horrible and it's all Nvidia's fault for not releasing specs.
To reiterate, a P100 and SVGA 1Meg video memory under Windows or OS/2 has better OpenGL performance than a K6-400 with 3Dnow, a GB Exxtreme and 8Megs of video memory on Linux.
That's fine but writing a super efficient software implementation is not something that 3D graphics hardware companies like SGI or Nvidia are going to care about. I didn't know that software Mesa was so slow but it is simply a problem that most people do not care about due to the cheap hardware accelerators available for $50.
You're right though, one of the hallmarks of Linux is that you can get decent performance out of old hardware but one of the other hallmarks of Linux is that if you don't like something fix it yourself so I suggest that if you really care about a fast software OpenGL implementation that you do it yourself.
While I respect your viewpoint I simply do not think that even the fastest software implementation will come anywhere close to the slowest 3D hardware with reasonable (i.e. not nvidia) drivers.
Running what? An OpenGL screensaver?
Under Linux, you need a specific card to get anything that even comes close. Right now, I have a 3D OpenGL accelerated card, but OpenGL is useless under Linux.
What card do you have? OpenGL on Linux with a 3dfx or Matrox card is pretty good now.
While I think an efficient software implementation is a worthy goal, I don't think it is worth it. A year from now it is going to be virtually impossible to find a 2D only card. Even the integrated graphics on chipsets have decent low end 3D support and with good 3D cards like Matrox G400's available for under $100 I see the trend continuing.
Unfortunately most of the 3D card makers do not give out specs on the 3D part of their cards except under NDA. Matrox G[12]00 cards are supported for 2D under XFree86 but no Mesa support. Same for nVidia RivaTNTs.
I hope some of the companies release 3D specs soon, there *is* a market, I just bought a Voodoo2 solely for Linux Quake II.
1. Carmack was praising Apple's adoption of OpenGL more than the ATI Rage 128 which is also available for PC's in an AGP2X version rather than PCI66.
2. The G3's still have PCI slots and Darryl Straus (sp?) was commenting about porting Glide to Linux/PPC so you could put a Voodoo2 in there.
3. 3Dfx's Voodoo3 will be released exclusively by themselves. S3 is a competitor and makes the Savage3D chip. 3Dfx recently bought STB which makes finished boards utilizing 3Dfx and other chips from vendors such as nVidia.
Intel's "no more socket" talk was all marketing BS. If you look at a Celeron 266, 300, 300A, or 333 you'll see that the PCB is virtually empty. The point of the PCB was for the P-II cache which is now gone or on chip so get rid of it and put the CPU back in a socket. The packaging is much cheaper. If they made more slot instead of socket they'd just make less money.
Yeah, on 1-2 CPU machines. Linux doesn't scale too well on our 16 CPU boxes...