That is rediculous. Do you think ATI (who paid $6M for exclusive marketing rights) would tolerate HL2 being delayed for NVIDIA's benefit?
And as for the generic DX9 favoring ATI cards inference. While it is true that the DX9 minimum spec is more aligned with ATI's hardware rather than the higher precision NVIDIA hardware, the source code has exposed a good deal of the custom, ATI-written shader code that is in HL2.
Your understanding was paid for by $6M and in cases like these it was money well spent.
Maybe this will give them a chance to take out all the ATI-specific custom shader code that has been found in the source so that it can actually be "generic DX9 code" that is running on the ATI cards as they and Valve claim.
Of course, maybe $6 million dollars (confirmed by ATI's earnings documents) will prevent that from happening (and would have prevented that from ever being known had the source not leaked).
http://www.spec.org/gpc/opc.data/vp71/summary.html
Note that there is only one system submitted with Wildcat 4 and it gets significantly beat by ATI and NVIDIA, especially by NVIDIA and especially in the Pro/E and 3ds max components of the benchmark suite.
Also note that there are ZERO systems submitted with Wildcat Pro.
What systems are submitted by the workstation vendors is almost as significant as the scores that are posted. Any time a system/card starts getting beat too bad on viewperf the results for that system are pulled. i.e. the posted solutions are the cream of the crop.
That board (and epoxy/heatsink) was NOT made and assembled by NVIDIA.
NVIDIA does not produce their consumer boards. All the boards are produced by other vendors such as PNY, Asus, MSI, Chaintech, Leadtek, Gainward, etc. using chips they buy from NVIDIA.
NVIDIA does produce and tightly control the manufactur%3of their Quadro cards.
The (true) Wildcat 4 cards are much slower than the Quadro FX 2000 and Quadro FX 3000. They are also slower than the ATI FireGL products.
The Wildcat VP products are slower than just about anything you put next to them.
3dlabs is no longer a significant player in the workstation graphics space.
The artical is flawed in many ways but the Quadro FX 3000 is shipping now and it thumps the FireGL X2 (as does the FX 2000).
It is pretty hard to argue that NVIDIA isn't dominating the workstation market at this point.
This article doesn't factor in a number of important things such as extreme model sizes (that kill the FireGL), driver quality (ATI Radeon is historically bad, New FireGL (ATI parts not IBM parts) is worse), and precision.
The FireGL X1/X2 have horribly low precision. The biggest area that shows this is in their sub pixel precision which results in many rendering errors per frame (holes, spots, tears, speckles).
Before Cg, HLSL, and OGL SLang PC game developers had to (and still do in many caases) ship shader machine language code in both ATI and NVIDIA flavors. This is why shader adoption has been so slow on the PCs. Writing assembly was tough but you also had to manage that per vender and per GPU (ATI vs. NVIDIA->NV2x vs. NV3x).
None of this is relavant to the XBox. The XBox is a closed system and every unit is the same so the software developers have been much more aggresive about supporting shaders. They only had to hand code for one GPU. More importantly there are millions and millions of DVDs out in livingrooms that ONLY have NVIDIA NV2x GPU machine language on them. There is no opportunity to rewrite this and there was no need to write for more than one profile. How they are going to support this is going to be very interesting.
This is much more critical and tough than supporting PS1 compatibility on a PS2. PS2 graphics were a complete superset of PS1 graphics and you can even use more advanced features (bi-linear filtering) when running a PS1 title on the PS2.
This is not the case with Xbox2. An ATI GPU is not going to be a superset of NVIDIA's chip level shader architecture or features. But more importantly, these shaders are VERY performance sensitive. The fragment shaders are running per fragment (future pixel) which is in a critical path for graphics performance. There is no time to emulate these in software or translate them. There is no time to move things onto or off of the GPU. These small, hand-tuned bits of GPU code have to run fast to maintain the target frame rate.
I will never say there isn't something they could do but at this point I don't see how they will be able to support the Xbox1 software that has been developed.
Moving forward they can start using HLSL and start compiling to two profiles to include the ATI specific GPU shader code (as soon as they know what this chip will look like). That won't happen for a while so most of the games of 2004 will be Xbox1 specific as well. HLSL does support dynamic compiling so that may be what they use until the Xbox2 chipset is ready.
Still, I don't understand how you will be able to say that NFL Fever 2005 will run on Xbox1 and Xbox2 but Halo/Halo2 is Xbox1 only. I think they will either find a way to ensure 100% compatibility or they will completely break it.
High Impact was the mid-range graphics in the I2 in 1996. SGI's Octane was released in 1997 with Impact graphics (which saw a increase in performance compared to your I2). Odyssey graphics (V8, V10) were launched for the Octane in 1999 which were a significant move forward over Impact graphics. Even the Odyssey launch was over a year before your PS2 launch so I am not sure why your I2 High Impact is at all relevant? Besides, the PS2's graphics couldn't even begin to compare to workstation graphics at the time it was launched so any direct comparisons (while pointless) will be in contrast to your claims.
As for SGI engineers at nvidia. Yep there are a good number, as well as Sun and LSI (founders), S3, 3dfx, ATI, Intel, Compaq, 3dlabs, DPI, HP, IBM, etc.
Since you are just a kid you get a second chance to take my advice... post about something you know. Maybe you could discuss Miller capacitance somewhere.
But the 30% that NVIDIA owns in the overall mobile market is a large majority of the high-end/performance segment which has higher revenue/margins.
Basically, your logic is flawed and over simplified but thanks for chiming in.
So as a Xbox developer, what are your views about a hard stop to the growth of your current target installed base and starting from scratch for the size of the new installed base?
All the while Sony will be adding to the PSX compatible family.
Goodness. The has a POS graphics processor. Just look at all the shitty PS2 ports to the Xbox compared to a native Xbox title.
It doesn't even have mipmapping in hardware (ack!) and the texture resolution of all PS2 games really blows.
BTW, the emotion engine isn't even the graphics processor.
Good lord... post about what you know.
There is a very long list of details and shaders that make up Dawn that are not included or properly rendered with the wrapper used to run Dawn on ATI.
Just looking at the screen shots should show anyone that knows what they are looking at that this is the case. Especially if you focus on her hair (which is where a majority of the data and shaded pixels and overdraw are).
This totally invalidates any performance comparisons using the wrapper version of Dawn and proves that running Xbox1 games on Xbox2 could be filled with issues and incompatibilities.
What it doesn't do is make good press which is why you will only every read about dawn with the wrapper running on ATI with no mention of the problems and missing features.
There was/is no DX8 shading language. DX8 allowed for GPU-specific machine code to be downloaded to the GPU but there was no shading language.
There weren't any high level shading languages for GPUs until NVIDIA's Cg and DirectX 9's HLSL (which are identical) came along. Neither of these were ready or around before the spring/summer of 2002.
I am not aware if the Xbox dev environment is allowing people to start using DX9 and HLSL at this point (probably) but the entire first batch of Xbox games with vertex and fragment shaders use NV2x specific machine code.
Wrong, there is very good reason. All of these vertex and fragment shaders are written in NV2x assembly code.
There is no "dx8" language. They write C/C++ to the dx8 spec for xbox1 but all the shader code had to all be written in machine language and downloaded onto the GPU via function calls in dx8. There was no HLSL or Cg at the time.
It will be very interesting to see what happens to all the NV2x specific machine language on DVDs in familyrooms all over the world when Xbox2 comes a long.
NVIDIA doesn't have a Pentium III license either but there is one in the Xbox chipset (nforce + geforce 3.5).
Your guessing and assumptions are not very valuable.
All of the vast amounts of vertex and fragment shader code for Xbox games are all written in NV2x GPU assembly code.
It will be interesting to see how they are able to get Xbox1 games run on Xbox2. This is not HLSL, OGL SLang, or Cg, this is GPU specific machine language.
Sony has already shown the enormous value of generation to generation compatibility (both in games and controller button layout). If Xbox2 is not compatibily with Xbox1 then...
It has 4 banks of 256 MB of texture RAM. The four banks are on separate pipelines and a copy of each texture needs to be duplicated in all four banks since you don't know what portion of the screen they will fall on.
As a result, this solution only has 256 MB of texture which isn't going to be much of a differentiation for very long. (Both NVIDIA and ATI have 256 MB boards coming--although that is shared with FB).
The low FSAA resolutions and lackluster performance is what will keep people from paying 10x-20x for this visualization solution.
One quick point to address all the 150 fps in Quake jokes:
Frame rate consistancy is what is most important (by far). A game that runs at 30 fps solid will also feel better than a game that runs at 60 fps some of the time but then bops back and forth between 30 fps and 60 fps.
The VisSim industry has done a better job of saying "we only need 60 Hz (fps) but we better never ever see you dip below that or you are out!" This forces hardware and software to be optimized for locking at 60 fps. I can tell you that a 60 fps Air Force flight simulator will always feel higher performance than a soupped up PC running Quake at 100 fps but dipping down to 50 fps or worse when things get hairy.
The biggest evidence of this issue being unimportant in PC gaming is the number of people or games that run with vertical blank (vblank) synchronization turned off. This is wrong wrong wrong in my opinion but most gamers are willing to live with enormous visual artifacts from partially completed frames to get that max fps and lowest input latency when things tough on the system.
So, to all those that mock high fps benchmarks, I challege you to post information on a recent 3D game, gfx card, system, and config that allows you to play with all the gfx features on (or those that are important to you) with vsync on using a 60 Hz display and only double buffering which locks at 60 fps solid without ever dipping below that.
That is when things have become fast enough for _that particular_ game.
Products like the NV30 and R300 help push the bar but are still not overkill. Take the above challenge and now turn on 16x multi-sampled FSAA (same as an SGI Onxy/IR), 8x anisotropic filtering (often more important than FSAA), 1600x1024 (the native resolution of my DFP), 128bit pixel depth (which NV30 can do before scan out), and include very complex vertex and fragment (pixel) programs. With all of that, turn Vsync on (as it should be) and have this entire combination run at 60 fps per second regardless of what is going on in the game at any and every given moment.
When we can do all of that, we are finished.:-)
The problem then becomes the content creators who continue to push the envolope. GeForce FX is launching with a demo that has Jurasic Park/Toy Story quality rendering tied to real-time dynamics and feature film quality animation. However, it is not quite to the level of Gollum in LOTR: Two Towers. Imagine Doom4 with 50 characters on the screen that all look like Yoda or Gollum. My point is that there is always room and applications for higher performance.
100 of these running around locked at 60 fps is the new goal:
http://notendur.centrum.is/~czar/misc/gollum.jpg
The problem are the states that have laws that say all their electoral votes must go to the winner of the popular election in the state. Even if he won at 51% to 49%.
Some states will devide the electoral votes and some follow the above method.
Slashdot Mirror
And as for the generic DX9 favoring ATI cards inference. While it is true that the DX9 minimum spec is more aligned with ATI's hardware rather than the higher precision NVIDIA hardware, the source code has exposed a good deal of the custom, ATI-written shader code that is in HL2.
Your understanding was paid for by $6M and in cases like these it was money well spent.
Maybe this will give them a chance to take out all the ATI-specific custom shader code that has been found in the source so that it can actually be "generic DX9 code" that is running on the ATI cards as they and Valve claim.
Of course, maybe $6 million dollars (confirmed by ATI's earnings documents) will prevent that from happening (and would have prevented that from ever being known had the source not leaked).
Also note that there are ZERO systems submitted with Wildcat Pro.
What systems are submitted by the workstation vendors is almost as significant as the scores that are posted. Any time a system/card starts getting beat too bad on viewperf the results for that system are pulled. i.e. the posted solutions are the cream of the crop.
http://www.spec.org/gpc/opc.data/vp71/summary.html
That board (and epoxy/heatsink) was NOT made and assembled by NVIDIA. NVIDIA does not produce their consumer boards. All the boards are produced by other vendors such as PNY, Asus, MSI, Chaintech, Leadtek, Gainward, etc. using chips they buy from NVIDIA. NVIDIA does produce and tightly control the manufactur%3of their Quadro cards.
The (true) Wildcat 4 cards are much slower than the Quadro FX 2000 and Quadro FX 3000. They are also slower than the ATI FireGL products. The Wildcat VP products are slower than just about anything you put next to them. 3dlabs is no longer a significant player in the workstation graphics space.
The artical is flawed in many ways but the Quadro FX 3000 is shipping now and it thumps the FireGL X2 (as does the FX 2000). It is pretty hard to argue that NVIDIA isn't dominating the workstation market at this point.
With Quadro4 it actually became a different chip so you can't turn the workstation specific parts of the chip on with softquadro any more.
This article doesn't factor in a number of important things such as extreme model sizes (that kill the FireGL), driver quality (ATI Radeon is historically bad, New FireGL (ATI parts not IBM parts) is worse), and precision. The FireGL X1/X2 have horribly low precision. The biggest area that shows this is in their sub pixel precision which results in many rendering errors per frame (holes, spots, tears, speckles).
Before Cg, HLSL, and OGL SLang PC game developers had to (and still do in many caases) ship shader machine language code in both ATI and NVIDIA flavors. This is why shader adoption has been so slow on the PCs. Writing assembly was tough but you also had to manage that per vender and per GPU (ATI vs. NVIDIA->NV2x vs. NV3x).
None of this is relavant to the XBox. The XBox is a closed system and every unit is the same so the software developers have been much more aggresive about supporting shaders. They only had to hand code for one GPU. More importantly there are millions and millions of DVDs out in livingrooms that ONLY have NVIDIA NV2x GPU machine language on them. There is no opportunity to rewrite this and there was no need to write for more than one profile. How they are going to support this is going to be very interesting.
This is much more critical and tough than supporting PS1 compatibility on a PS2. PS2 graphics were a complete superset of PS1 graphics and you can even use more advanced features (bi-linear filtering) when running a PS1 title on the PS2.
This is not the case with Xbox2. An ATI GPU is not going to be a superset of NVIDIA's chip level shader architecture or features. But more importantly, these shaders are VERY performance sensitive. The fragment shaders are running per fragment (future pixel) which is in a critical path for graphics performance. There is no time to emulate these in software or translate them. There is no time to move things onto or off of the GPU. These small, hand-tuned bits of GPU code have to run fast to maintain the target frame rate.
I will never say there isn't something they could do but at this point I don't see how they will be able to support the Xbox1 software that has been developed.
Moving forward they can start using HLSL and start compiling to two profiles to include the ATI specific GPU shader code (as soon as they know what this chip will look like). That won't happen for a while so most of the games of 2004 will be Xbox1 specific as well. HLSL does support dynamic compiling so that may be what they use until the Xbox2 chipset is ready.
Still, I don't understand how you will be able to say that NFL Fever 2005 will run on Xbox1 and Xbox2 but Halo/Halo2 is Xbox1 only. I think they will either find a way to ensure 100% compatibility or they will completely break it.
High Impact was the mid-range graphics in the I2 in 1996. SGI's Octane was released in 1997 with Impact graphics (which saw a increase in performance compared to your I2). Odyssey graphics (V8, V10) were launched for the Octane in 1999 which were a significant move forward over Impact graphics. Even the Odyssey launch was over a year before your PS2 launch so I am not sure why your I2 High Impact is at all relevant? Besides, the PS2's graphics couldn't even begin to compare to workstation graphics at the time it was launched so any direct comparisons (while pointless) will be in contrast to your claims.
As for SGI engineers at nvidia. Yep there are a good number, as well as Sun and LSI (founders), S3, 3dfx, ATI, Intel, Compaq, 3dlabs, DPI, HP, IBM, etc.
Since you are just a kid you get a second chance to take my advice... post about something you know. Maybe you could discuss Miller capacitance somewhere.
Hmmm... That has got to be around $50,000 to $100,000 towards ATI's bottom line. I better rush out an4ey some ATYT stock. :-P
But the 30% that NVIDIA owns in the overall mobile market is a large majority of the high-end/performance segment which has higher revenue/margins. Basically, your logic is flawed and over simplified but thanks for chiming in.
So as a Xbox developer, what are your views about a hard stop to the growth of your current target installed base and starting from scratch for the size of the new installed base? All the while Sony will be adding to the PSX compatible family.
Goodness. The has a POS graphics processor. Just look at all the shitty PS2 ports to the Xbox compared to a native Xbox title. It doesn't even have mipmapping in hardware (ack!) and the texture resolution of all PS2 games really blows. BTW, the emotion engine isn't even the graphics processor. Good lord... post about what you know.
Just looking at the screen shots should show anyone that knows what they are looking at that this is the case. Especially if you focus on her hair (which is where a majority of the data and shaded pixels and overdraw are).
This totally invalidates any performance comparisons using the wrapper version of Dawn and proves that running Xbox1 games on Xbox2 could be filled with issues and incompatibilities.
What it doesn't do is make good press which is why you will only every read about dawn with the wrapper running on ATI with no mention of the problems and missing features.
There weren't any high level shading languages for GPUs until NVIDIA's Cg and DirectX 9's HLSL (which are identical) came along. Neither of these were ready or around before the spring/summer of 2002.
I am not aware if the Xbox dev environment is allowing people to start using DX9 and HLSL at this point (probably) but the entire first batch of Xbox games with vertex and fragment shaders use NV2x specific machine code.
Wrong, there is very good reason. All of these vertex and fragment shaders are written in NV2x assembly code. There is no "dx8" language. They write C/C++ to the dx8 spec for xbox1 but all the shader code had to all be written in machine language and downloaded onto the GPU via function calls in dx8. There was no HLSL or Cg at the time. It will be very interesting to see what happens to all the NV2x specific machine language on DVDs in familyrooms all over the world when Xbox2 comes a long.
NVIDIA doesn't have a Pentium III license either but there is one in the Xbox chipset (nforce + geforce 3.5). Your guessing and assumptions are not very valuable.
It will be interesting to see how they are able to get Xbox1 games run on Xbox2. This is not HLSL, OGL SLang, or Cg, this is GPU specific machine language.
Sony has already shown the enormous value of generation to generation compatibility (both in games and controller button layout). If Xbox2 is not compatibily with Xbox1 then...
Go Nebraska!!! 1997 National Champions!
It has 4 banks of 256 MB of texture RAM. The four banks are on separate pipelines and a copy of each texture needs to be duplicated in all four banks since you don't know what portion of the screen they will fall on.
As a result, this solution only has 256 MB of texture which isn't going to be much of a differentiation for very long. (Both NVIDIA and ATI have 256 MB boards coming--although that is shared with FB). The low FSAA resolutions and lackluster performance is what will keep people from paying 10x-20x for this visualization solution.
The content is on film at 24 fps and the projectors double-shutter the film to have it flash at 48 fps, reducing the flashing of yester-year.
BTW, It is only when the camera pans to you REALLY notice the 24 fps content.
I hate the fact that the new digital projection standards (and HDTV for that matter with 1080-24p) are designed around this ancient frame rate.
One quick point to address all the 150 fps in Quake jokes:
Frame rate consistancy is what is most important (by far). A game that runs at 30 fps solid will also feel better than a game that runs at 60 fps some of the time but then bops back and forth between 30 fps and 60 fps.
The VisSim industry has done a better job of saying "we only need 60 Hz (fps) but we better never ever see you dip below that or you are out!" This forces hardware and software to be optimized for locking at 60 fps. I can tell you that a 60 fps Air Force flight simulator will always feel higher performance than a soupped up PC running Quake at 100 fps but dipping down to 50 fps or worse when things get hairy.
The biggest evidence of this issue being unimportant in PC gaming is the number of people or games that run with vertical blank (vblank) synchronization turned off. This is wrong wrong wrong in my opinion but most gamers are willing to live with enormous visual artifacts from partially completed frames to get that max fps and lowest input latency when things tough on the system.
So, to all those that mock high fps benchmarks, I challege you to post information on a recent 3D game, gfx card, system, and config that allows you to play with all the gfx features on (or those that are important to you) with vsync on using a 60 Hz display and only double buffering which locks at 60 fps solid without ever dipping below that.
That is when things have become fast enough for _that particular_ game.
Products like the NV30 and R300 help push the bar but are still not overkill. Take the above challenge and now turn on 16x multi-sampled FSAA (same as an SGI Onxy/IR), 8x anisotropic filtering (often more important than FSAA), 1600x1024 (the native resolution of my DFP), 128bit pixel depth (which NV30 can do before scan out), and include very complex vertex and fragment (pixel) programs. With all of that, turn Vsync on (as it should be) and have this entire combination run at 60 fps per second regardless of what is going on in the game at any and every given moment.
When we can do all of that, we are finished. :-)
The problem then becomes the content creators who continue to push the envolope. GeForce FX is launching with a demo that has Jurasic Park/Toy Story quality rendering tied to real-time dynamics and feature film quality animation. However, it is not quite to the level of Gollum in LOTR: Two Towers. Imagine Doom4 with 50 characters on the screen that all look like Yoda or Gollum. My point is that there is always room and applications for higher performance.
100 of these running around locked at 60 fps is the new goal: http://notendur.centrum.is/~czar/misc/gollum.jpg
Some states will devide the electoral votes and some follow the above method.
That is the flaw in the system, if you ask me.
(Guess which method Flordia uses. :-)