NVIDIA Wouldn’t Exist Without Xbox & DirectX Gaming — Microsoft CEO

[Marees]

Xbox CEO Asha Sharma hosted Microsoft CEO Satya Nadella in an internal Q&A, amplifying that Microsoft will "always" invest in gaming. "We're long on gaming. We'll continue to invest, and we'll always do so."​

News
By Jez Corden published 2 days ago
At a big internal Q&A session, Xbox CEO Asha Sharma hosted Microsoft CEO Satya Nadella, discussing the "long" vision for Xbox's future.

Nadella emphasized gaming's legacy and influence on Microsoft and the wider technology stack, talking up how gaming has served as an accelerator for cloud, Windows, the GPU-based server tech revolution, and beyond. But he also said that doesn't mean moving away from what people expect of gaming, and explained why Microsoft will "always" continue to invest in gaming.

"The trickle from that excellence to the rest of the company becomes straight forward. I joke with [NVIDIA CEO] Jensen Huang, if it wasn't for gaming [NVIDIA] wouldn't exist. Think about it, without DirectX, I don't think the entire GPU revolution, or the acceleration would've happened."

https://www.windowscentral.com/gami...satya-nadella-we-will-always-invest-in-gaming

https://wccftech.com/nvidia-wouldnt-exist-without-gaming-says-microsoft-ceo/

 

[DanNeely]

That quote says nothing about XBox, just DirectX. DirectX itself dates back to Windows 95, which launched 6 years before the XBox. And while the earliest versions of DirectX had a rocky start, it was well established as the dominant standard for Windows games before the XBox road its coattails.

 

[Marees]

That quote says nothing about XBox, just DirectX. DirectX itself dates back to Windows 95, which launched 6 years before the XBox. And while the earliest versions of DirectX had a rocky start, it was well established as the dominant standard for Windows games before the XBox road its coattails.

true. I put it as a reference to the fact that xbox has directX in its name

looking up Wikipedia it says directX was developed so that windows could be used as an alternative for Japanese game consoles

The project was codenamed the Manhattan Project, like the World War II project of the same name, and the idea was to displace the Japanese-developed video game consoles with personal computers running Microsoft's operating system.[8]​

​

The "Direct" part of the library was so named as these routines bypassed existing core Windows 95 routines and accessed the computer hardware only via a hardware abstraction layer (HAL).[19] Though the team had named it the "Game SDK" (software development kit), the name "DirectX" came from one journalist that had mocked the naming scheme of the various libraries. The team opted to continue to use that naming scheme and call the project DirectX.[8]​

​

In a console-specific version, DirectX was used as a basis for Microsoft's Xbox, Xbox 360 and Xbox One console API. The API was developed jointly between Microsoft and Nvidia, which developed the custom graphics hardware used by the original Xbox. The Xbox API was similar to DirectX version 8.1, but is non-updateable like other console technologies. The Xbox was code named DirectXbox, but this was shortened to Xbox for its commercial name.[29]​

https://en.wikipedia.org/wiki/DirectX

 

[Sycraft]

That quote says nothing about XBox, just DirectX. DirectX itself dates back to Windows 95, which launched 6 years before the XBox. And while the earliest versions of DirectX had a rocky start, it was well established as the dominant standard for Windows games before the XBox road its coattails.

Also I disagree, because one thing nVidia did very right was support for OpenGL as well. They didn't do their own "native" API like so many other vendors back in the day, they supported both DX and OGL and did so both with the same performance. So I feel like had DirectX not taken off, they would have been fine since OpenGL probably is what all game engines would have defaulted to.

That said, DirectX did help PC gaming overall since it wasn't just graphics and also kept advancing when OpenGL got real, real stagnant for a number of years.

 

[ERipley]

I remember when Nvidia came out and the shit they pulled on 3DFX, and one of their board partners..... And almost forgot the bs they pulled when they bought Ageia. Nvidia has been scummy for a long time

 

[DukenukemX]

Also I disagree, because one thing nVidia did very right was support for OpenGL as well. They didn't do their own "native" API like so many other vendors back in the day, they supported both DX and OGL and did so both with the same performance. So I feel like had DirectX not taken off, they would have been fine since OpenGL probably is what all game engines would have defaulted to.

That said, DirectX did help PC gaming overall since it wasn't just graphics and also kept advancing when OpenGL got real, real stagnant for a number of years.

I remember that most games supported both OpenGL and DirectX and I would flip between them often to see which gave me a better frame rate. If I was lucky, they supported Glide. But a lot of games didn't make use of DirectX like Pitfall the Mayan Adventure game. Gabe Newell would eventually port Doom over to Windows 95 because a lot of games during the 90's would still run in DOS like Duke Nukem 3D. Gabe Newell did this to prove that DirectX worked.

OpenGL worked fine for any game. Remember Doom 3 was using OpenGL and even Doom 2016 was still using OpenGL. The reason games used DirectX was because of Xbox. Development made sense to use DirectX over other API's. I wouldn't say that DirectX helped PC gaming either since Microsoft had screwed up a number of times. Some of them were directly meant to piss off Nvidia.

1. Remember when DX9 used 24-bit color for "True Color" but Nvidia was expecting Microsoft to use 16-bit? This is why GeforceFX cards performed so badly with certain DX9 games because they needed 24-bit and this forced FX cards to use 32-bit which was very slow.
2. Remember when DirectX10 sucked so hard that people avoided using the shadows feature because it was slow? Then AMD comes out with DX10.1 cards which magically fixed the performance? Sucked to be Nvidia.
3. Remember when DirectX10 was exclusive to Vista, an OS that nobody liked? We had a whole decade of DX9 because a lot of people refused to switch over to Vista and kept using XP.

Also, Nvidia graphic cards are really good at OpenGL. Yuzu the Switch emulator would originally only work on Nvidia cards because the Switch's hardware used an Nvidia chip. For a while, only Nvidia users could run this emulator through OpenGL. Even when Vulkan was eventually implemented, it was still better for Nvidia cards to use OpenGL. As a PC gamer, if DirectX went missing from history then I wouldn't have noticed. Developers would have moved onto OpenGL and life would have moved on.


View: https://youtu.be/KnWruWDTfSw?t=1168

 

[Sycraft]

I remember that most games supported both OpenGL and DirectX and I would flip between them often to see which gave me a better frame rate. If I was lucky, they supported Glide. But a lot of games didn't make use of DirectX like Pitfall the Mayan Adventure game. Gabe Newell would eventually port Doom over to Windows 95 because a lot of games during the 90's would still run in DOS like Duke Nukem 3D. Gabe Newell did this to prove that DirectX worked.

I don't ever know that it was most, but there were a fair few. Some had one pipeline that was dramatically better than others, but one I remember that was basically dead equal was Earth 2150. It's DX and OGL pipelines seemed to be dead on, and it had a simplistic benchmark utility. It was where I was able to definitively test my feeling that AMD cards sucked at OpenGL compared to DriectX. Like I could see them drag ass at Quake 3 comparatively... but maybe iD TEch just didn't work as well? Here they'd work well in DX but not in OGL.

Also, Nvidia graphic cards are really good at OpenGL. Yuzu the Switch emulator would originally only work on Nvidia cards because the Switch's hardware used an Nvidia chip. For a while, only Nvidia users could run this emulator through OpenGL. Even when Vulkan was eventually implemented, it was still better for Nvidia cards to use OpenGL. As a PC gamer, if DirectX went missing from history then I wouldn't have noticed. Developers would have moved onto OpenGL and life would have moved on.

I dunno man, there was a LOOONG time where OpenGL was stuck on version 1.4 and then later quite a while when it was stuck on 2.1. Companies added features using extensions, but those were always a mess in various ways. It helped having Direct3D getting regular updates that pushed forward what was expected of cards.

I'm not saying it wouldn't have worked out, it surely would have, but I do think DX helped.

 

[Lakados]

I don't ever know that it was most, but there were a fair few. Some had one pipeline that was dramatically better than others, but one I remember that was basically dead equal was Earth 2150. It's DX and OGL pipelines seemed to be dead on, and it had a simplistic benchmark utility. It was where I was able to definitively test my feeling that AMD cards sucked at OpenGL compared to DriectX. Like I could see them drag ass at Quake 3 comparatively... but maybe iD TEch just didn't work as well? Here they'd work well in DX but not in OGL.


I dunno man, there was a LOOONG time where OpenGL was stuck on version 1.4 and then later quite a while when it was stuck on 2.1. Companies added features using extensions, but those were always a mess in various ways. It helped having Direct3D getting regular updates that pushed forward what was expected of cards.

I'm not saying it wouldn't have worked out, it surely would have, but I do think DX helped.

Back in the early 2000’s I was working at a company maintaining their internal wrapper library, they had the common DX and OGL commands wrapped up and depending on passing the function a True or a False it would execute the function as one or the other.

Those wrappers got large and moving between GL versions was not an easy task.

 

[erek]

“NVIDIA Virtualizes Game Development With RTX PRO Server​

PRESS RELEASE by GFreeman Today, 03:56 Discuss (1 Comment)
Game development teams are working across larger worlds, more complex pipelines and more distributed teams than ever. At the same time, many studios still rely on fixed, desk-bound GPU hardware for critical production work. At the Game Developers Conference (GDC) this week in San Francisco, NVIDIA is showcasing a new approach to bring together disparate workflows using virtualized game development on NVIDIA RTX PRO Servers, powered by NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs and NVIDIA vGPU software.

With the RTX PRO Server, studios can centralize and virtualize core workflows across creative, engineering, AI research and quality assurance (QA) - all on shared GPU infrastructure in the data center. This enables teams to maintain the responsiveness and visual fidelity they expect from workstation-class systems while improving infrastructure utilization, scalability, data security and operational consistency across teams and locations.”

 

[M76]

nVidia wouldn't exist without gamers, and neither would directx or xbox these corporate types always seem to forget that.

With the RTX PRO Server, studios can centralize and virtualize core workflows across creative, engineering, AI research and quality assurance (QA) - all on shared GPU infrastructure in the data center. This enables teams to maintain the responsiveness and visual fidelity they expect from workstation-class systems while improving infrastructure utilization, scalability, data security and operational consistency across teams and locations.”

Translation from corporate speak: "We took away your workstations, but you can take turns using ours for a subscription fee"

 

[LukeTbk]

nVidia did very right was support for OpenGL as well.

that was my first thought, but DirectX is more than just Direct3D, Microsoft level of control made so an GPU company could spend a fortune in advance on the R&D side I guess and make the super fast yearly evolution with multi hardware vendor work, now that it slowed down and there is just 2 players it is a bit hard to go back at a time that an dictator API side was really useful.

Translation from corporate speak: "We took away your workstations, but you can take turns using ours for a subscription fee"

RTX pro server run is made to be run inside the (big) game studio own infrastructure if they want, to be used by their many world office. They can usually rent outside compute during peak time as they want, but they will tend to have their own, I think they are all hybrid like that. They have giant data size, tend to care about leaks and many workload run every day of the year in a very predictable way (ike all the compile/test run)

Take activision for an example:
https://www.nvidia.com/en-us/case-studies/activision/

They used RTX pro server rack to update their own infracstructure, a single 8xRTX 6000 96gb 4U rack cost ~120/180k with great margin and it does not change much of anything for Nvidia to sell them to ubisoft or AWS.

, a single RTX PRO 6000 Blackwell Server Edition GPU can support up to 48 concurrent users, that start to be impressive...

 

[Lakados]

nVidia wouldn't exist without gamers, and neither would directx or xbox these corporate types always seem to forget that.


Translation from corporate speak: "We took away your workstations, but you can take turns using ours for a subscription fee"

The worst part for me was the VMware licenses, it only worked with Enterprise…
Ran damned near flawlessly…. But VMWare’s price tag made us search out active alternatives.

 

[HockeyJon]

Also I disagree, because one thing nVidia did very right was support for OpenGL as well. They didn't do their own "native" API like so many other vendors back in the day, they supported both DX and OGL and did so both with the same performance. So I feel like had DirectX not taken off, they would have been fine since OpenGL probably is what all game engines would have defaulted to.

That said, DirectX did help PC gaming overall since it wasn't just graphics and also kept advancing when OpenGL got real, real stagnant for a number of years.

Nvidia also started developing CUDA in 2004 and was investing heavily in GPUs as general compute infrastructure while Wall Street sat there for years saying "what does is this gaming company think they're doing lmao".

 

[HockeyJon]

That quote says nothing about XBox, just DirectX. DirectX itself dates back to Windows 95, which launched 6 years before the XBox. And while the earliest versions of DirectX had a rocky start, it was well established as the dominant standard for Windows games before the XBox road its coattails.

Asha Sharma might understand this if she were an actual gamer, but why would you want an actual gamer running Xbox?

 

[LukeTbk]

nVidia wouldn't exist without gamers, and neither would directx or xbox these corporate types always seem to forget that.

I am not sure they ever do, it seems way more the other way around the modern developped world that seem to think the economy come from consumer and not producer, at every turn trying to flatter them and them getting extremely entitled (while they will without thinking for a second buy a cheaper option if available with infinite greed)

 

[DukenukemX]

I don't ever know that it was most, but there were a fair few. Some had one pipeline that was dramatically better than others, but one I remember that was basically dead equal was Earth 2150. It's DX and OGL pipelines seemed to be dead on, and it had a simplistic benchmark utility. It was where I was able to definitively test my feeling that AMD cards sucked at OpenGL compared to DriectX. Like I could see them drag ass at Quake 3 comparatively... but maybe iD TEch just didn't work as well? Here they'd work well in DX but not in OGL.

Original Half Life made use of OpenGL, Direct3D, and Glide. I would often switch between them because when I died in Counter Strike, I would blame it on my frame rate. I think Descent 3 was another game that made use of all three API's.

I dunno man, there was a LOOONG time where OpenGL was stuck on version 1.4 and then later quite a while when it was stuck on 2.1.

OpenGL 1.4 was released in July 2002 and was succeeded by OpenGL 1.5 in July 2003. From the looks of it, they never stopped updating OpenGL. The biggest gap was 2 years? More than likely ATI/AMD/Nvidia/Intel were probably not keeping up with OpenGL versions.

1.0	June 30, 1992	Initial release.
1.1	March 4, 1997[49][50]	Texture objects, Vertex Arrays
1.2	March 16, 1998	3D textures, BGRA and packed pixel formats,[51] introduction of the imaging subset useful to image-processing applications
1.2.1	October 14, 1998	A concept of ARB extensions
1.3	August 14, 2001	Multitexturing, multisampling, texture compression
1.4	July 24, 2002	Depth textures, GLSlang[52]
1.5	July 29, 2003	Vertex Buffer Object (VBO), Occlusion Queries[53]
2.0	September 7, 2004	GLSL 1.1, MRT, Non Power of Two textures, Point Sprites,[54] Two-sided stencil[53]
2.1	July 2, 2006	GLSL 1.2, Pixel Buffer Object (PBO), sRGB Textures[53]
3.0	August 11, 2008	GLSL 1.3, Texture Arrays, Conditional rendering, Frame Buffer Object (FBO)[55]
3.1	March 24, 2009	GLSL 1.4, Instancing, Texture Buffer Object, Uniform Buffer Object, Primitive restart[56]
3.2	August 3, 2009	GLSL 1.5, Geometry Shader, Multi-sampled textures[57]
3.3	March 11, 2010	GLSL 3.30, Retain as much OpenGL 4.0 functionality as possible, New blending functions, Sampler Objects, new texture and vertex formats
4.0	March 11, 2010	GLSL 4.00, Tessellation on GPU, shaders with 64-bit precision[58]
4.1	July 26, 2010	GLSL 4.10, Developer-friendly debug outputs,https://en.wikipedia.org/wiki/OpenGL#cite_note-GL4.1-60 compatibility with OpenGL ES 2.0[59]

I'm not saying it wouldn't have worked out, it surely would have, but I do think DX helped.

I think DirectX did more harm than good, and I say this for various reasons.

1. Very often new versions of DirectX were tied to a specific version of Windows. We saw this with DX10 with Vista and DX12 with Windows 10. People did get DirectX10 unofficially working on Windows 2000 and Microsoft even got DX12 worked on Windows 7 for World of Warcraft. You don't have this problem with OpenGL and Vulkan.
2. DirectX is only available on Windows while OpenGL and Vulkan is available to every platform with the exception of Apple devices. This helped Windows gaming but not gamers in general.
3. Microsoft was so lazy when it came to updating DirectX that AMD went ahead and created their own API known as Mantle. Eventually, Microsoft woke up from their cocaine trip and made DX12 work very similar to that of Mantle. I'm not saying DX12 is Mantle, but Mantle definitely influenced DX12.

 

[M76]

RTX pro server run is made to be run inside the (big) game studio own infrastructure if they want, to be used by their many world office. They can usually rent outside compute during peak time as they want, but they will tend to have their own, I think they are all hybrid like that. They have giant data size, tend to care about leaks and many workload run every day of the year in a very predictable way (ike all the compile/test run)

Well, excuse me for taking them at their word:

all on shared GPU infrastructure in the data center.

Take activision for an example:
https://www.nvidia.com/en-us/case-studies/activision/

They used RTX pro server rack to update their own infracstructure, a single 8xRTX 6000 96gb 4U rack cost ~120/180k with great margin and it does not change much of anything for Nvidia to sell them to ubisoft or AWS.

And how many workstations could you buy for 180k? More than 8, so it is not that good of a deal.

, a single RTX PRO 6000 Blackwell Server Edition GPU can support up to 48 concurrent users, that start to be impressive...

This only sounds impressive to the corporate suits, not for the user who knows that if more than 1 person does something io or computationally intensive it all bogs down. There is no replacement for having your own dedicated workstation.

 

[Sycraft]

Original Half Life made use of OpenGL, Direct3D, and Glide. I would often switch between them because when I died in Counter Strike, I would blame it on my frame rate. I think Descent 3 was another game that made use of all three API's.

Unreal Tournament did too, but it had pretty drastically different performance internally between them.

OpenGL 1.4 was released in July 2002 and was succeeded by OpenGL 1.5 in July 2003. From the looks of it, they never stopped updating OpenGL. The biggest gap was 2 years? More than likely ATI/AMD/Nvidia/Intel were probably not keeping up with OpenGL versions.

I don't know what the issue with 1.5 was, but I never saw any adoption, among games or graphics drivers. Perhaps the updates weren't worth it, or it required things that hardware didn't support. Either way, there was this stagnation on 1.4, in pro software and games, where DX kept moving and cards had to keep piling on extensions which meant if a game DID use it what it could support differed between vendors because of different vendor extensions.

Very often new versions of DirectX were tied to a specific version of Windows. We saw this with DX10 with Vista and DX12 with Windows 10. People did get DirectX10 unofficially working on Windows 2000 and Microsoft even got DX12 worked on Windows 7 for World of Warcraft. You don't have this problem with OpenGL and Vulkan.

I mean this is true but it seems to be much less of a big deal than people make it out to be. The big screaming was over Vista and, well, people needed to scream less . Windows 7, the OS people seemed to love so much, was just Vista with a new coat of paint and some minor tweaks. The performance issues with DX10 were something that was unavoidable because it was doing things so different and initial DX10 supporting titles really weren't, they were doing things in DX9 style inside DX10 which was slow. It was a radical change in how GPUs were programmed. Once programmers learned it and new engines got out that were really written to use it (which took a couple years) things got much better. This is why people remember DX11 more fondly, because it was out and being used by the time a lot of the transition happened. There were plenty of games that "supported" DX10 that really just threw their DX9 shaders in to it and bolted on something showy and didn't do a good job, so it was slow whereas a DX11 game tended to be actually written for DX11 and worked well.

DirectX is only available on Windows while OpenGL and Vulkan is available to every platform with the exception of Apple devices. This helped Windows gaming but not gamers in general.

True.

Microsoft was so lazy when it came to updating DirectX that AMD went ahead and created their own API known as Mantle. Eventually, Microsoft woke up from their cocaine trip and made DX12 work very similar to that of Mantle. I'm not saying DX12 is Mantle, but Mantle definitely influenced DX12.

MS got lazy recently, but that is different than back in the early days of OpenGL vs DirectX when there was a lot of really fast development. Also, OpenGL was just as lazy is recent times. The last few OpenGL updates were very much "Bolt on the shit that cards are doing to support the new DX, not necessarily in a great fashion."

With Mantle and DX12 and Vulkan... man I have mixed feelings. Going with a more complex, lower level, API has helped with the ability to make things faster for sure, and that's good. However it does seem to have come at the cost of complexity that has been killing off game engines. Part of the reason for the massive consolidation to things like Unreal Engine, even from studios that have the teams to do their own, is that it is apparently getting very time consuming to implement one. The graphics APIs aren't the only reason for it, but that's a part of it.

 

[LukeTbk]

Well, excuse me for taking them at their word:

yes you can use third party data center or your own that you make their own compute center (+smaller local around the world office)

And how many workstations could you buy for 180k? More than 8, so it is not that good of a deal.

they would be idling a lot of the time, when you have office all around the world with different time zone, a single rack working near 24/24 can quickly be worth it.

There is no replacement for having your own dedicated workstation.

of course but 96vram rtx 6000 to everyone would cost a lot more versus time share, life of compromise (of course nvidia would much prefer selling 4 to 16 times more rtx 6000, but they have competition from AWS and others that studios massively use, so they must adapt).

True.

Could have helped gamers during the high speed era, building consensus over many os and hardware vendors could have been slower, versus microsoft ability to simply dictate it.

From the looks of it, they never stopped updating OpenGL. The biggest gap was 2 years? More than likely ATI/AMD/Nvidia/Intel were probably not keeping up with OpenGL versions

By the late 90s and early 00s when you add things already added by dx or by vendor own extensions which became way too common as the committee could not reach consensus fast, not so much that they were not keeping up than making their own Nv_, ATI_ extensions all the time, we had quite the bit of per vendor shaders in my limited days of openGL 3d engines in the 00s. Lot of stuff appeared first in openGL, but were standardized much faster in DX because of Microsoft power (which could have helped gamers, having all hardwares vendors supporting the same standard, helping both gpu competition and gamedevs).

Vulkan idea to make the drivers/api part so "dump" that it get out of the way pretty much get rid of the committee slow speed and other issues, but bring new one

 

[M76]

yes you can use third party data center or your own that you make their own compute center (+smaller local around the world office)

I was pointing out that they said explicitly that it runs in a data center. Game companies don't have their own data centers.

they would be idling a lot of the time, when you have office all around the world with different time zone, a single rack working near 24/24 can quickly be worth it.

Again this only sounds good to the suits who only see buzzwords like idling. Getting work done on a computer that you remote into from the other side of the world is anything but efficient.

of course but 96vram rtx 6000 to everyone would cost a lot more versus time share,

99% of staff doesn't need a 96GB vram RTX 6000 for game development. And if you share that with just 3 people then you are already better off with a local 5090. unless you specifically need 96GB Vram for something the local hw will always win. And if you do need the 96GB Vram because you run a 200 billion parameter ai, then you have to time share the thing anyway.

life of compromise

The only one getting compromised is the poor sod who is supposed to do actual work, who now has to remote into a shared server instead of having a dedicated workstation.

 

[Marees]

The Nvidia GeForce3 launched 25 years ago — underappreciated at launch, its impact shaped the industry​

News
By Zak Killian published March 1, 2026

When it launched in February of 2001, the GeForce 3 was a critical turning point in the history of graphics processors, as this was the first GPU to include any real sort of programmability by way of being the first GPU with DirectX 8.0 pixel and vertex shader support. What that means is that graphics programmers could now write programs that run on the GPU

before the GeForce 3, virtually all graphics processors were "dumb" fixed-function accelerators. You would hand them data in a specific format from the CPU, they would do their magic, and then output the modified framebuffer. Any code you wrote ran on the CPU, which meant that any special graphical effects you wanted had to be performed on the CPU.

DirectX 8 enabled radically more lifelike water effects, as seen in The Elder Scrolls III: Morrowind, and it also famously enabled per-pixel lighting and true Dot3 bump-mapping, as seen in Doom 3.

Another launch late in 2001 solidified the GeForce 3 as the foundation of things to come: Microsoft's Xbox. The original Xbox famously used NVIDIA graphics, but fewer realize that NVIDIA also provided the sound hardware and memory controller for the machine. NVIDIA's work on the Xbox would become the foundation for its beloved but relatively short-lived "nForce" series of motherboard chipsets.

https://www.tomshardware.com/pc-com...ated-at-launch-its-impact-shaped-the-industry

 

[Marees]

The Nvidia GeForce3 launched 25 years ago — underappreciated at launch, its impact shaped the industry​

News
By Zak Killian published March 1, 2026

When it launched in February of 2001, the GeForce 3 was a critical turning point in the history of graphics processors, as this was the first GPU to include any real sort of programmability by way of being the first GPU with DirectX 8.0 pixel and vertex shader support. What that means is that graphics programmers could now write programs that run on the GPU

before the GeForce 3, virtually all graphics processors were "dumb" fixed-function accelerators. You would hand them data in a specific format from the CPU, they would do their magic, and then output the modified framebuffer. Any code you wrote ran on the CPU, which meant that any special graphical effects you wanted had to be performed on the CPU.

DirectX 8 enabled radically more lifelike water effects, as seen in The Elder Scrolls III: Morrowind, and it also famously enabled per-pixel lighting and true Dot3 bump-mapping, as seen in Doom 3.

Another launch late in 2001 solidified the GeForce 3 as the foundation of things to come: Microsoft's Xbox. The original Xbox famously used NVIDIA graphics, but fewer realize that NVIDIA also provided the sound hardware and memory controller for the machine. NVIDIA's work on the Xbox would become the foundation for its beloved but relatively short-lived "nForce" series of motherboard chipsets.

https://www.tomshardware.com/pc-com...ated-at-launch-its-impact-shaped-the-industry

“No GeForce, No AI,” Declares NVIDIA’s Jensen Huang, as He Celebrates 25 Years of GeForce 3, the GPU That Started Everything​

Muhammad Zuhair
Mar 12, 2026 at 10:58am EDT

While talking with GeForce members, Jensen recalled that with GeForce 3, NVIDIA made the transition from fixed-function accelerators to programmable shaders, with the core idea being to infuse the "artistic" touch of developers into each game.

Jensen says that the shift towards this newer programming approach eventually paved the way for CUDA, which added parallelism to GPU computation.

https://wccftech.com/no-geforce-no-...ang-as-he-celebrates-the-launch-of-geforce-3/

 

[Marees]

“No GeForce, No AI,” Declares NVIDIA’s Jensen Huang, as He Celebrates 25 Years of GeForce 3, the GPU That Started Everything​

Muhammad Zuhair
Mar 12, 2026 at 10:58am EDT

While talking with GeForce members, Jensen recalled that with GeForce 3, NVIDIA made the transition from fixed-function accelerators to programmable shaders, with the core idea being to infuse the "artistic" touch of developers into each game.

Jensen says that the shift towards this newer programming approach eventually paved the way for CUDA, which added parallelism to GPU computation.

https://wccftech.com/no-geforce-no-...ang-as-he-celebrates-the-launch-of-geforce-3/

It’s been 25 years since Nvidia GeForce 3 — and I think gamers accidentally built the AI era​

Opinion
By Jason England published February 27, 2026
We’re still living in the world it created

‘Your world, programmed’​

So let’s look back. In 2001, Nvidia CEO Jensen Huang talked about GeForce 3 as a way to “unleash cinematic realism,” all by moving away from “fixed-function” chips and giving creators a blank canvas to make with.

This came down to the introduction of Programmable Vertex and Pixel Shaders. Think of a 3D video game like a movie set. To get a character on screen, the computer has to figure out where things are and what they look like.

Before GeForce 3, these jobs were hardwired into the chip and developers were stuck with them. If you wanted to draw water in a harbor scene, you were stuck with the water Nvidia’s engineers gave you.

Now, with this increased programmability, the Vertex shaders (the “where”) are able to be controlled by the coder — like a construction worker moving the scenery around. And on top of that, the programmable pixel shaders (the “what color”) was also able to be controlled.

The legacy​

“Without GeForce, there would be no AI today. Without AI, there would be no DLSS today. It's harmonious," Huang said back in January.

https://www.tomsguide.com/computing...-i-think-gamers-accidentally-built-the-ai-era

 

[DukenukemX]

The Nvidia GeForce3 launched 25 years ago — underappreciated at launch, its impact shaped the industry​

News
By Zak Killian published March 1, 2026

When it launched in February of 2001, the GeForce 3 was a critical turning point in the history of graphics processors, as this was the first GPU to include any real sort of programmability by way of being the first GPU with DirectX 8.0 pixel and vertex shader support. What that means is that graphics programmers could now write programs that run on the GPU

before the GeForce 3, virtually all graphics processors were "dumb" fixed-function accelerators. You would hand them data in a specific format from the CPU, they would do their magic, and then output the modified framebuffer. Any code you wrote ran on the CPU, which meant that any special graphical effects you wanted had to be performed on the CPU.

DirectX 8 enabled radically more lifelike water effects, as seen in The Elder Scrolls III: Morrowind, and it also famously enabled per-pixel lighting and true Dot3 bump-mapping, as seen in Doom 3.

Another launch late in 2001 solidified the GeForce 3 as the foundation of things to come: Microsoft's Xbox. The original Xbox famously used NVIDIA graphics, but fewer realize that NVIDIA also provided the sound hardware and memory controller for the machine. NVIDIA's work on the Xbox would become the foundation for its beloved but relatively short-lived "nForce" series of motherboard chipsets.

https://www.tomshardware.com/pc-com...ated-at-launch-its-impact-shaped-the-industry

That era of DirectX was probably the worst era. Sucked for whatever owned a graphics card, as I don't think there were any ATI or Nvidia winners here. DirectX 8 was pretty good, but then AMD comes out with their Radeon 8500 which used DX8.1, which was better. Except that the Radeon 8500 was still slow and buggy, with even John Carmack finding bugs in the drivers. But then DX9 was released, and graphics got much better. Except that Nvidia wasn't happy with how DX9 turned out and fought back by pushing developers to use DX8.1 in place of DX9. This had the unfortunate effect to screw over Geforce 3 owners, since they can't do DX8.1 Eventually games like Half Life 2 did make use of proper DX9 and that's when GeforceFX cards were showing their problems.

The problems get even worse when moving on since there wasn't just one version of DX9. There were three versions. The one we had been using was known as DX9a, but when ATI released their new graphic cards, they were DX9b. Nvidia moved onto DX9c which used Shader Model 3. This was such a problem because graphic cards like ATI Radeon 9800's and GeforceFX's could not play DX9c games despite being a DX9 graphics card. Valve had a lot to say about this and even made the Lost Coast to show that things like HDR could be done with DX9a.

View: https://youtu.be/VJyCYA8JFoo?t=347

Despite what Valve has said and done, most new games going forward used DX9c. BioShock for example requires DX9c and people have even made hacks that allowed these games to run on older graphic cards using DX9a. DirectX didn't settle down as a graphics API until DX11 where it didn't make any huge changes that requires people to go out and buy another graphics card for an API change.

View: https://youtu.be/ZPn7Z6XINwg?si=r-NIw0JAEQ3VszuF