The Wall Street Journal was recently reporting that many soldiers are feeling over weighted with stuff to carry, so they are resorting to chucking all MRE items that aren't particularly high in carbs, and tossing the GI GPS units for civilian ones that use the same batteries as their headsets to reduce the different types of batteries that must be carried in the field. I'd imagine that such soldiers would be rather disappointed if they had to go back. According to the article, the Army is aware that this is happening. This article appeared about a month ago.
> Why is NVidia still supporting OpenGL? Well, for one, their chipsets have traditionally been built around OpenGL.
Ha! What NVidia was originally known for was as the company that could deliver as much DirectX performance as 3DFX could deliver Glide performance. It wasn't until some time later that they started being known for OpenGL.
Back in the good old days, the ideal gaming machine was a nice Pentium II 300 with a Riva 128 on AGP and a Voodoo SLI taking up 2 PCI slots. If only I'd had the money for that. Later the ideal became a TNT with VoodooII SLI, which I actually do have, but I aquired it long after it was considered cool.
Trailers are often made before all the effects work is finished, so it is quite possible that the pillar falling effect will be much better in the actual movie than it was in the trailer.
I don't know why one really has to have X11, except that other than mpg123, the options for console only seem limited. Even the program you link to, even though it is a console program, requires X11 be running.
I'll have to look into it. I generally refuse to use libraries that aren't nicely supported on linux, irix, and solaris (ideally windows and MacOS X would be supported, but really I do my primary work on linux, irix, and solaris, so if I can get my code written today, I'm willing to duplicate library code for Mac and Windows support later).
From an assembly point of view, I believe that they are both about the same for ease of use. I think that initially Altivec had a bit of a lead, but not for long. I'm just arguing that the C libraries for using Altivec are better. And the file you pointed out just moves mmx/sse/sse2 programming a step up from inlined assembly. With this, I am still limited to the very small steps that inlined assembly provides. I don't see anything in this for doing dot products or cross products, although it would be trivial to write such functions, so perhaps I shouldn't complain so much.
BTW, I don't have it handy, so I'll just ask. Does MS Visual C++ 6.0 provide this stuff? I seem to recall that to use MSVC++ at all with SSE/SSE2 I had to either by the Intel Compiler or upgrade to VisualStudio.NET.
No, not included. But Intel used to have a bunch of DSP code for free download from their web site. It used to be that you would find such things through their Code Central web site (http://cedar.intel.com/cgi-bin/ids.dll/topic.jsp? catCode=CAS). I'm not going to take the time to verify that such things are still there. And, it might not have been an exact replacement to vDSP. I was just trying to point out the ins and out of both sides.
You are, of course, correct. Intel does provide such a library. For a fee. Unless you can provide a link otherwise, I believe you are refering to Intel Performance Primitives ($199, not sure if royalty free, http://developer.intel.com/software/products/ipp/i pp30/) or Intel's Math Kernel Library, which is part of the Intel Performance Libraries (also $199, again not sure of royalty situation,http://developer.intel.com/software/prod ucts/perflib/).
I might consider purchasing these libraries if I really needed them, but in general, I find it unacceptable that Intel won't help developers get the most out of the platform for free. They already have some of the most expensive hardware in the business, and here they are charging us more.
While ISV certification is often very important, in this case of this article, where they were talking about a machine for a specific task, I don't believe certification is too important. MATLAB doesn't require certification (I don't think they even give certification to anyone), and I'm pretty sure that the Quadro viewer program mentioned also doesn't require certification other than that the machine have a proper quadro.
Apple provides a very nice, extremely easy to use Altivec library. It requires writing no assembly code, and I believe it even resorts back to non-Altivec means of execution if a program written using the library is executed on a G3. So, for instance, in Altivec, you can write things like:
result = vec_add( aVector, someOtherVector );
and it works properly regardless of what sort of vector you've chosen to use for aVector.
I've yet to see anything similar for 3D Now or SSE/SSE2. Everything I've seen for them is either a library that is too application specific (like a premade image recognition library), or requires using assembly and a compiler newer than VC++ 6.0 (maybe only SSE2 really requires that).
Apple also provides a bunch of other libraries, like vDSP (I'm sure AMD and Intel provide an equivelent), and BLAS (this is a somewhat standardized library across platforms. My recall is that there is a SSE/SSE2 version, but Intel charges money for it, instead of giving it out for free), and in general, they make it easier for Apple developers to take advantage of Altivec than Intel does SSE2 or AMD does 3D Now. Unfortuntaly, a lot of developers want to maintain only one code base across all platforms, so they won't use the Apple provided tools (there are free unoptimized versions of BLAS for every platform though, so developers should at least use that so they can't get speed benefit on platforms that provide it), which sucks because GCC also sucks for speed, so people using vendor supplied compilers on other platforms (like Intel's on Windows or Linux, SGI's on Irix, Sun's on Solaris) get a nice speed boost that would require hand assembly optimization to get on a G4.
Unless this is I-frame only MPEG, I don't know why anyone would prefer it over MJPEG. And there is already at least one other camera that will record to a harddrive. The Cannon XL1s has built in minidv, but it will also record directly to a firewire harddrive. You then just disconnect the firewire HD and connect it to your Mac or PC. Presumably, you would also be able to use firewire networking to keep all 3 connected at once.
Don't remove the lame X. Just find a way to add mame after the X.
I so far have refused to buy an Xbox. However, I did see one game recently that was rather tempting. The game was Steel Battalion from Capcom. Thankfully though, this one game is $199, so I can't afford it anyway.
My point wasn't that Max wasn't a significant program. My point was that it isn't really one of the heavy weight biggies. I don't really know why, although I personally find it less intuitive than Softimage, and you have to spend a lot of money on plug ins (well, you used to, I don't know what the state of things is past version 3.0) to get it to do what Maya Complete will do (which is cheaper than the base Max anyway), and historically PCs weren't viewed as powerfull enough.
Unfortunately, powerfull Macs cost more than I can afford. I have an old 8100, and an old 840av, but so far I've mainly only used the 8100 for running photoshop and illustrator, and the 840av for capturing video to send to a linux machine.
I'm a developer, not really an artist. So, I figure in the real world, I would probably be working inside programs like Maya, XSI, Houdini, and Max writing scripts and plugins. But, until I get into such a job, I can work with a few stand alone programs (I really like Wings, Gimp has some cool features, though it isn't as polished as Photoshop, it is good enough for now, and I still hang on to my old copy of BMRT) and writing my own code.
I believe that Bryce should be able to import Wavefront files (*.obj) from Wings3D just fine. You just need to use a seperate program (like Photoshop) to generate texture maps.
OK, first, there is no point in discussing image editing programs since we've already established that she had good photoshop skills, and presumably a copy of photoshop.
As to modeling animation, there is no reason not to consider cheap programs and also no reason not to launch right into character animation. In particular, if character animation is really what she wants to do, Hash's Animation Master is an excellent choice. Unfortunately, it lacks certain things needed for a lot of professional work (like a renderer that is slow and not particularly good). Also Blender used to be fairly good until the corporate collapse. I expect that now that it is opensourced, it will get back to and exceed where it was. In the mean time, the old versions are still available. This program isn't a good for character animation though as some higher end offerings or A:M.
While I have little expertise here, there are a wide range of 3D program that appeal to still designers. Carrera Studio is one. Pixels3D seems to be another.
Also, a pair of fun programs are Poser and Bryce.
BTW, Lightwave and Max aren't really the biggies in the modelling and animation. The biggies are Maya and Softimage, followed not so closely by Houdini, Lightwave, and Max. Of those programs, I think Lightwave is the easiest to use, but Maya is the cheapest. Max is powerfull, but not as much as Maya, so I can't really see choosing it at this point, though lots of people do.
I would explore using cheaper programs before really committing to something like Maya or Lightwave. They are expensive and take dedication. And the important thing is artistic ability, not what software used, so use something fun and affordable for now.
Finally, compositing wasn't really brought up, but since you mentioned it, After Effects is not the heavy hitter. The heavy hitter would be Inferno or something from Quantel. After Effects is merely the cheapest program that can usually get the job done. Buy it if you need it and it's what you can afford. Otherwise skip it if you don't need it, or look else where if you can afford more. Like Combustion (from the the people who make Inferno).
Finally, using Macs are fine. Using linux is likely to be headache fraught, especially for video editing. And for free software, in this case linux is expensive because you generally have few choices for good cheap software, with notable exceptions (wings3d and gimp, mainly). I suppose you can use Windows, but I'd strongly encourage a nice Mac instead.
> A lot of power users probably use Final Cut Pro/DVD Studio Pro too but I bet there is a much much larger market that uses iMovie and iDVD because it's easy and it does what "the common man" needs done... this is the same place where GIMP will prevail.
I don't think that many people are going to find the GIMP much easier to learn than Photoshop. Nor should that be the GIMP team's goal. Their goal so far has been to make the best, most powerfull, image editor possible, not the easiest to use. And that is the only way I'd have it. Though, there are still areas I wish they would work of ease of use.
Adobe PhotoDelux would be a better equivelent of iMovie and iDVD, but I don't know of any free equivelent.
However, new people might learn the gimp that would have otherwise just pirated Photoshop.
Also, FilmGimp does things well that photoshop does badly, like handling video and 16bit images. Sure, photoshop does 16bit, except that, to my understanding, a lot of features disappear, and not just from among the third party plugins.
Most dedicated Photoshop users probably won't switch soon, but maybe they find reasons to use Gimp anyway (the gimp seems like it is well on its way to being better for scripting, but I could be wrong), and there are many people who find photoshop to be the wrong tool.
I bought some import PSX games to test my mod chip out. I bought the mod chip for two purposes. First, to really play imported games, particularly from france and japan, and second to try my hand at writing my own games, and this mod chip was supposed to include the ability to easily download the game to ram through a serial port.
Anyway, the "import" PSX games I bought at the store, I realized when I got home that there was something fishy. Where in the world would people be selling video games in english for NTSC systems other than the US and Canada (and isn't Canada the same region as far as video games go?)? And upon close inspection, the jewel case booklet and that sheet in the back looked suspect. Darn it. It is people like that who give those of us who really want imports a bad name.
Of course, the solution would be to work on pushing for a unified free driver between ATI, Matrox, 3DLabs (as possible, meaning perhaps only the Permedia line), and NVidia. However, I suspect it might be harder to do here than across card from one manufactor.
Further, while it seems that for some cards, we have the info needed to program the rasterizer, I don't think anybody has specs on geometry engines for any card. And even if we did, optimization would have to be a trial and error process since we don't know what is ideal for the card. It would be something interesting to work on though, to try and see if one could write TNT and pre geometry engine ATI drivers that are faster than the vendor supplied ones, then try to attack the geometry engine problem. I think that the DRI guys are doing this somewhat already, except for they seem more focused on getting things to work rather than optimizing them (which is very sensible really).
> I decided long ago that I did not want to become like those sad people that cling to long-dead platforms, constantly predicting their return and vocally defending their virtues. Let BeOS rest in peace.
I like those people. Like the ones who do insane amounts of hacking to add upgrades and get more speed out of their Amigas. Or Acorns. Or Ataris. I think they do more than most groups to make the computing world interesting, and I applaud their efforts. And I wish I had the time and money to join them. I think the efforts of Apple II and C64 people are cool too, but have less desire to join them.
Programable vertex and pixel shaders and massive fill rates are potentially very usefull for paint programs.
For instance, instead of the CPU rendering each tile of the image, the paint program could send all the tiles of all the layers to the graphics card and let the graphics card render the screen. The tiles would reside on the graphics card until they are changed, or they need to be removed for new tiles from scrolling or zooming the image. Well, so far this just would require AGP8x and massive file rates. But there are other tasks that could be implemented as shaders here. For instance, pixel shaders could handle the alpha blending rather than redrawing the scene for each layer. There are many tasks that could be pushed to the video card, but some of them would be hampered by the fact that while AGP 8x is fast to write to, reading back from it is about the same speed as PCI, which could be detrimental for many tasks.
None of this is supported by the Gimp, or FilmGimp, or any other paint program that I know of for linux. I'm led to believe that some Irix programs do things like this though.
BTW, I hear that the new NVidia cards did away with hardware T&L, and instead they do it using vertex shaders, with the GL driver making it work seemlessly. I can't really confirm this from anyplace officially, but it would make sense to reduce redundancy this way.
I have a Geforce3 in a P2 machine, and it only takes me about 45 seconds to startx.
I bought the Geforce3 because I previously had a TNT2 and they both use the same driver. I would have rather had a FireGL card, but most of all I needed something that would work in under an hour because I needed faster GL performance and had no spare time for extensive tinkering.
I guess it is time to upgrade my nvidia drivers again. Unfortunately, it will be sometime before I can afford a FireGL card now.
Slashdot Mirror
The Wall Street Journal was recently reporting that many soldiers are feeling over weighted with stuff to carry, so they are resorting to chucking all MRE items that aren't particularly high in carbs, and tossing the GI GPS units for civilian ones that use the same batteries as their headsets to reduce the different types of batteries that must be carried in the field. I'd imagine that such soldiers would be rather disappointed if they had to go back. According to the article, the Army is aware that this is happening. This article appeared about a month ago.
I believe it is actually 1.1 with some tweaks to make it look more like 1.2 for compatibility.
However, graphic card vendors can supply other versions.
> Why is NVidia still supporting OpenGL? Well, for one, their chipsets have traditionally been built around OpenGL.
Ha! What NVidia was originally known for was as the company that could deliver as much DirectX performance as 3DFX could deliver Glide performance. It wasn't until some time later that they started being known for OpenGL.
Back in the good old days, the ideal gaming machine was a nice Pentium II 300 with a Riva 128 on AGP and a Voodoo SLI taking up 2 PCI slots. If only I'd had the money for that. Later the ideal became a TNT with VoodooII SLI, which I actually do have, but I aquired it long after it was considered cool.
Trailers are often made before all the effects work is finished, so it is quite possible that the pillar falling effect will be much better in the actual movie than it was in the trailer.
Actually, they pioneered it with the original Voodoos. Ahh those were the days.
I keep meaning to take my pair of Voodoo2 cards and try SLIing them in a machine. I've never actually done that with them.
I don't know why one really has to have X11, except that other than mpg123, the options for console only seem limited. Even the program you link to, even though it is a console program, requires X11 be running.
I'll have to look into it. I generally refuse to use libraries that aren't nicely supported on linux, irix, and solaris (ideally windows and MacOS X would be supported, but really I do my primary work on linux, irix, and solaris, so if I can get my code written today, I'm willing to duplicate library code for Mac and Windows support later).
Thank you for the link.
From an assembly point of view, I believe that they are both about the same for ease of use. I think that initially Altivec had a bit of a lead, but not for long. I'm just arguing that the C libraries for using Altivec are better. And the file you pointed out just moves mmx/sse/sse2 programming a step up from inlined assembly. With this, I am still limited to the very small steps that inlined assembly provides. I don't see anything in this for doing dot products or cross products, although it would be trivial to write such functions, so perhaps I shouldn't complain so much.
BTW, I don't have it handy, so I'll just ask. Does MS Visual C++ 6.0 provide this stuff? I seem to recall that to use MSVC++ at all with SSE/SSE2 I had to either by the Intel Compiler or upgrade to VisualStudio.NET.
No, not included. But Intel used to have a bunch of DSP code for free download from their web site. It used to be that you would find such things through their Code Central web site (http://cedar.intel.com/cgi-bin/ids.dll/topic.jsp? catCode=CAS). I'm not going to take the time to verify that such things are still there. And, it might not have been an exact replacement to vDSP. I was just trying to point out the ins and out of both sides.
You are, of course, correct. Intel does provide such a library. For a fee. Unless you can provide a link otherwise, I believe you are refering to Intel Performance Primitives ($199, not sure if royalty free, http://developer.intel.com/software/products/ipp/i pp30/) or Intel's Math Kernel Library, which is part of the Intel Performance Libraries (also $199, again not sure of royalty situation,http://developer.intel.com/software/prod ucts/perflib/).
I might consider purchasing these libraries if I really needed them, but in general, I find it unacceptable that Intel won't help developers get the most out of the platform for free. They already have some of the most expensive hardware in the business, and here they are charging us more.
While ISV certification is often very important, in this case of this article, where they were talking about a machine for a specific task, I don't believe certification is too important. MATLAB doesn't require certification (I don't think they even give certification to anyone), and I'm pretty sure that the Quadro viewer program mentioned also doesn't require certification other than that the machine have a proper quadro.
Apple provides a very nice, extremely easy to use Altivec library. It requires writing no assembly code, and I believe it even resorts back to non-Altivec means of execution if a program written using the library is executed on a G3. So, for instance, in Altivec, you can write things like:
result = vec_add( aVector, someOtherVector );
and it works properly regardless of what sort of vector you've chosen to use for aVector.
I've yet to see anything similar for 3D Now or SSE/SSE2. Everything I've seen for them is either a library that is too application specific (like a premade image recognition library), or requires using assembly and a compiler newer than VC++ 6.0 (maybe only SSE2 really requires that).
Apple also provides a bunch of other libraries, like vDSP (I'm sure AMD and Intel provide an equivelent), and BLAS (this is a somewhat standardized library across platforms. My recall is that there is a SSE/SSE2 version, but Intel charges money for it, instead of giving it out for free), and in general, they make it easier for Apple developers to take advantage of Altivec than Intel does SSE2 or AMD does 3D Now. Unfortuntaly, a lot of developers want to maintain only one code base across all platforms, so they won't use the Apple provided tools (there are free unoptimized versions of BLAS for every platform though, so developers should at least use that so they can't get speed benefit on platforms that provide it), which sucks because GCC also sucks for speed, so people using vendor supplied compilers on other platforms (like Intel's on Windows or Linux, SGI's on Irix, Sun's on Solaris) get a nice speed boost that would require hand assembly optimization to get on a G4.
Unless this is I-frame only MPEG, I don't know why anyone would prefer it over MJPEG. And there is already at least one other camera that will record to a harddrive. The Cannon XL1s has built in minidv, but it will also record directly to a firewire harddrive. You then just disconnect the firewire HD and connect it to your Mac or PC. Presumably, you would also be able to use firewire networking to keep all 3 connected at once.
Don't remove the lame X. Just find a way to add mame after the X.
I so far have refused to buy an Xbox. However, I did see one game recently that was rather tempting. The game was Steel Battalion from Capcom. Thankfully though, this one game is $199, so I can't afford it anyway.
My point wasn't that Max wasn't a significant program. My point was that it isn't really one of the heavy weight biggies. I don't really know why, although I personally find it less intuitive than Softimage, and you have to spend a lot of money on plug ins (well, you used to, I don't know what the state of things is past version 3.0) to get it to do what Maya Complete will do (which is cheaper than the base Max anyway), and historically PCs weren't viewed as powerfull enough.
Unfortunately, powerfull Macs cost more than I can afford. I have an old 8100, and an old 840av, but so far I've mainly only used the 8100 for running photoshop and illustrator, and the 840av for capturing video to send to a linux machine.
I'm a developer, not really an artist. So, I figure in the real world, I would probably be working inside programs like Maya, XSI, Houdini, and Max writing scripts and plugins. But, until I get into such a job, I can work with a few stand alone programs (I really like Wings, Gimp has some cool features, though it isn't as polished as Photoshop, it is good enough for now, and I still hang on to my old copy of BMRT) and writing my own code.
I believe that Bryce should be able to import Wavefront files (*.obj) from Wings3D just fine. You just need to use a seperate program (like Photoshop) to generate texture maps.
That is some really lame advice.
OK, first, there is no point in discussing image editing programs since we've already established that she had good photoshop skills, and presumably a copy of photoshop.
As to modeling animation, there is no reason not to consider cheap programs and also no reason not to launch right into character animation. In particular, if character animation is really what she wants to do, Hash's Animation Master is an excellent choice. Unfortunately, it lacks certain things needed for a lot of professional work (like a renderer that is slow and not particularly good). Also Blender used to be fairly good until the corporate collapse. I expect that now that it is opensourced, it will get back to and exceed where it was. In the mean time, the old versions are still available. This program isn't a good for character animation though as some higher end offerings or A:M.
While I have little expertise here, there are a wide range of 3D program that appeal to still designers. Carrera Studio is one. Pixels3D seems to be another.
Also, a pair of fun programs are Poser and Bryce.
BTW, Lightwave and Max aren't really the biggies in the modelling and animation. The biggies are Maya and Softimage, followed not so closely by Houdini, Lightwave, and Max. Of those programs, I think Lightwave is the easiest to use, but Maya is the cheapest. Max is powerfull, but not as much as Maya, so I can't really see choosing it at this point, though lots of people do.
I would explore using cheaper programs before really committing to something like Maya or Lightwave. They are expensive and take dedication. And the important thing is artistic ability, not what software used, so use something fun and affordable for now.
Finally, compositing wasn't really brought up, but since you mentioned it, After Effects is not the heavy hitter. The heavy hitter would be Inferno or something from Quantel. After Effects is merely the cheapest program that can usually get the job done. Buy it if you need it and it's what you can afford. Otherwise skip it if you don't need it, or look else where if you can afford more. Like Combustion (from the the people who make Inferno).
Finally, using Macs are fine. Using linux is likely to be headache fraught, especially for video editing. And for free software, in this case linux is expensive because you generally have few choices for good cheap software, with notable exceptions (wings3d and gimp, mainly). I suppose you can use Windows, but I'd strongly encourage a nice Mac instead.
Then why did you just do it?
> A lot of power users probably use Final Cut Pro/DVD Studio Pro too but I bet there is a much much larger market that uses iMovie and iDVD because it's easy and it does what "the common man" needs done... this is the same place where GIMP will prevail.
I don't think that many people are going to find the GIMP much easier to learn than Photoshop. Nor should that be the GIMP team's goal. Their goal so far has been to make the best, most powerfull, image editor possible, not the easiest to use. And that is the only way I'd have it. Though, there are still areas I wish they would work of ease of use.
Adobe PhotoDelux would be a better equivelent of iMovie and iDVD, but I don't know of any free equivelent.
However, new people might learn the gimp that would have otherwise just pirated Photoshop.
Also, FilmGimp does things well that photoshop does badly, like handling video and 16bit images. Sure, photoshop does 16bit, except that, to my understanding, a lot of features disappear, and not just from among the third party plugins.
Most dedicated Photoshop users probably won't switch soon, but maybe they find reasons to use Gimp anyway (the gimp seems like it is well on its way to being better for scripting, but I could be wrong), and there are many people who find photoshop to be the wrong tool.
I bought some import PSX games to test my mod chip out. I bought the mod chip for two purposes. First, to really play imported games, particularly from france and japan, and second to try my hand at writing my own games, and this mod chip was supposed to include the ability to easily download the game to ram through a serial port.
Anyway, the "import" PSX games I bought at the store, I realized when I got home that there was something fishy. Where in the world would people be selling video games in english for NTSC systems other than the US and Canada (and isn't Canada the same region as far as video games go?)? And upon close inspection, the jewel case booklet and that sheet in the back looked suspect. Darn it. It is people like that who give those of us who really want imports a bad name.
Of course, the solution would be to work on pushing for a unified free driver between ATI, Matrox, 3DLabs (as possible, meaning perhaps only the Permedia line), and NVidia. However, I suspect it might be harder to do here than across card from one manufactor.
Further, while it seems that for some cards, we have the info needed to program the rasterizer, I don't think anybody has specs on geometry engines for any card. And even if we did, optimization would have to be a trial and error process since we don't know what is ideal for the card. It would be something interesting to work on though, to try and see if one could write TNT and pre geometry engine ATI drivers that are faster than the vendor supplied ones, then try to attack the geometry engine problem. I think that the DRI guys are doing this somewhat already, except for they seem more focused on getting things to work rather than optimizing them (which is very sensible really).
> I decided long ago that I did not want to become like those sad people that cling to long-dead platforms, constantly predicting their return and vocally defending their virtues. Let BeOS rest in peace.
I like those people. Like the ones who do insane amounts of hacking to add upgrades and get more speed out of their Amigas. Or Acorns. Or Ataris. I think they do more than most groups to make the computing world interesting, and I applaud their efforts. And I wish I had the time and money to join them. I think the efforts of Apple II and C64 people are cool too, but have less desire to join them.
Programable vertex and pixel shaders and massive fill rates are potentially very usefull for paint programs.
For instance, instead of the CPU rendering each tile of the image, the paint program could send all the tiles of all the layers to the graphics card and let the graphics card render the screen. The tiles would reside on the graphics card until they are changed, or they need to be removed for new tiles from scrolling or zooming the image. Well, so far this just would require AGP8x and massive file rates. But there are other tasks that could be implemented as shaders here. For instance, pixel shaders could handle the alpha blending rather than redrawing the scene for each layer. There are many tasks that could be pushed to the video card, but some of them would be hampered by the fact that while AGP 8x is fast to write to, reading back from it is about the same speed as PCI, which could be detrimental for many tasks.
None of this is supported by the Gimp, or FilmGimp, or any other paint program that I know of for linux. I'm led to believe that some Irix programs do things like this though.
BTW, I hear that the new NVidia cards did away with hardware T&L, and instead they do it using vertex shaders, with the GL driver making it work seemlessly. I can't really confirm this from anyplace officially, but it would make sense to reduce redundancy this way.
I have a Geforce3 in a P2 machine, and it only takes me about 45 seconds to startx.
I bought the Geforce3 because I previously had a TNT2 and they both use the same driver. I would have rather had a FireGL card, but most of all I needed something that would work in under an hour because I needed faster GL performance and had no spare time for extensive tinkering.
I guess it is time to upgrade my nvidia drivers again. Unfortunately, it will be sometime before I can afford a FireGL card now.