Nvidia GeForce 256 is 25 years old
NVIDIA celebrated the 25th anniversary of the GeForce 256, the first GPU, which revolutionized graphics processing by integrating T&L features, leading to significant advancements in modern GPUs and performance.
Read original article
On August 31, 2024, NVIDIA celebrated the 25th anniversary of the GeForce 256, which was launched on October 11, 1999, and is recognized as the world's first GPU. This innovation marked a significant shift in the graphics processing landscape by integrating transform and lighting (T&L) features directly into the GPU, which previously relied on the CPU or separate hardware. This advancement allowed for reduced costs and improved performance, enabling NVIDIA to surpass its competitors. At the time of its release, NVIDIA faced competition from 35 companies, with only AMD remaining in the market today. Over the years, GPUs have evolved dramatically, now featuring advanced capabilities such as ray tracing and tensor cores, and are essential in various fields including gaming, professional visualization, and AI. The GeForce 256 had a transistor count of approximately 23 million and supported DirectX 7.0, while its modern counterpart, the GeForce RTX 4090, boasts 76.3 billion transistors and supports DirectX 12 Ultimate. The evolution of GPUs has seen a staggering increase in performance and capabilities, with the GeForce 256 serving as a foundational milestone in this journey.
- NVIDIA's GeForce 256 is recognized as the first GPU, launched in 1999.
- The integration of T&L features into the GPU marked a significant technological advancement.
- NVIDIA faced competition from 35 companies at the time, with only AMD remaining today.
- Modern GPUs have evolved to include advanced features like ray tracing and tensor cores.
- The transistor count in GPUs has increased dramatically from 23 million in the GeForce 256 to 76.3 billion in the RTX 4090.
Related
Nvidia NVLink Switch Chips Change to the HGX B200
NVIDIA introduced the HGX B200 board at Computex 2024, featuring two NVLink Switch chips instead of four, aiming to enhance performance and efficiency in high-performance computing applications by optimizing GPU configurations.
Nvidia: An Overnight Success Story 30 Years in the Making
Nvidia, founded in 1993 for 3D graphics cards, shifted to AI computing, led by founders Jensen Huang, Chris Malachowsky, and Curtis Priem. Strategic decisions propelled Nvidia to AI innovation leadership.
Nvidia's Blackwell Reworked – Shipment Delays and GB200A Reworked Platforms
Nvidia's Blackwell family faces production challenges causing shipment delays, impacting targets for 2024-2025. The company is extending Hopper product lifespans and shifting focus to new systems and simpler packaging solutions.
Nvidia launches a new RTX 4070
Nvidia has introduced an updated RTX 4070 graphics card with GDDR6 memory to enhance supply and availability, maintaining similar performance to the previous model, available globally from September 2024.
Nvidia Announces Financial Results for Second Quarter Fiscal 2025
NVIDIA reported Q2 FY25 revenues of $30 billion, a 122% year-over-year increase. Data Center revenue reached $26.3 billion. The company anticipates Q3 revenue of $32.5 billion, focusing on AI innovations.
3 commentsAccording to Wiki [1], the Geforce 256 has 17M transistor and not the ~23M they suggest. ~13W Power TDP instead of 50W.
The Geforce only has ~140mm2 Die Size so should take that into account when making comparison to AD102 which has a die size of 609mm2. Or roughly 4.4 times the Die Size with 4400 times the transistor. 1000 times increase in the span of 23 years.
Unfortunately we wont see a 3nm GPU until 2026. Which is sad because with the margin and hype and demand of AI I was hoping we could see 2nm GPU by 2026.
Edit: A sort of stupid idea came into my mind, I wonder if Intel or in this case Nvidia could open source their Geforce 256 hardware design, given this is 25 years old already and bare little significance of the modern GPU.
[1] https://en.wikipedia.org/wiki/List_of_Nvidia_graphics_proces...
It was the first decent gfx card I ever bought, my little brother and I pooled all our money in order to do it! We picked it up in some little hole in the wall in Fremont I'd found through Pricewatch.com.
Prior to that we'd gotten ripped by Fry's buying garbage PCI Diamond Crystal something or other cards, they didn't even stock recent NVidia cards back then. Then I found HardOCP and learned what to get. Still miss Kyle Bennett's sass.
The late nineties were something else, loved those crazy Athlon and Duron builds that could OC like crazy. It was so much fun! I really don't know of anything comparable these days, because for the past 15 years nearly zero games really required the latest and greatest hardware to be playable at reasonable settings. Homemade QinLED and other home-flashed iOT devices are fun, but not in the same galaxy.
I miss the A-Bit motherboards, they were a cut above. Never had one die on me ever (while under active use*).
* Storing old PC components seems sensitive - tossing them in a box in anti-static bags in a dry garage doesn't seem to do the trick.
P.s. - TFA claims the GeForce 256 was "the world's first GPU", but why not consider the NVidia Riva TNT a real GPU? And what about the They were capable at the time. And c'mon, the 3DFX Voodoo was the real first GPU.
If you want to feel bad, click this SEO stuffed and gamed link for "most beloved GPU of all time" and notice the lack of mention of 3DFX. Gaslighting trash.
https://www.google.com/search?q=most+beloved+GPU+of+all+time
Related
Nvidia NVLink Switch Chips Change to the HGX B200
NVIDIA introduced the HGX B200 board at Computex 2024, featuring two NVLink Switch chips instead of four, aiming to enhance performance and efficiency in high-performance computing applications by optimizing GPU configurations.
Nvidia: An Overnight Success Story 30 Years in the Making
Nvidia, founded in 1993 for 3D graphics cards, shifted to AI computing, led by founders Jensen Huang, Chris Malachowsky, and Curtis Priem. Strategic decisions propelled Nvidia to AI innovation leadership.
Nvidia's Blackwell Reworked – Shipment Delays and GB200A Reworked Platforms
Nvidia's Blackwell family faces production challenges causing shipment delays, impacting targets for 2024-2025. The company is extending Hopper product lifespans and shifting focus to new systems and simpler packaging solutions.
Nvidia launches a new RTX 4070
Nvidia has introduced an updated RTX 4070 graphics card with GDDR6 memory to enhance supply and availability, maintaining similar performance to the previous model, available globally from September 2024.
Nvidia Announces Financial Results for Second Quarter Fiscal 2025
NVIDIA reported Q2 FY25 revenues of $30 billion, a 122% year-over-year increase. Data Center revenue reached $26.3 billion. The company anticipates Q3 revenue of $32.5 billion, focusing on AI innovations.