Nvidia is usually credited with inventing the GPU, because it marketed the GeForce 256 in 1999 as the world’s first one. But graphics chips existed earlier, and the term itself was used by 3Dlabs in 1997. What Nvidia did first was put transform and lighting onto a single consumer graphics chip.
- Nvidia popularized the term GPU with the GeForce 256 in October 1999.
- 3Dlabs used the term in 1997, and graphics chips came before that.
- The GeForce 256 was the first consumer chip to integrate transform and lighting.
- So the credit depends on whether you mean the term or the hardware.
Who Gets the Credit for the First GPU?
Nvidia does, in practice. In October 1999 it introduced the NV10, sold on the GeForce 256 card, and marketed it as the world’s first GPU. It combined transform, lighting, triangle setup, and rendering in a single chip. Jon Peddie’s history of the chip notes that Nvidia has been associated with the term ever since.
What Came Before the GeForce 256
Graphics hardware was already around. IBM’s Professional Graphics Controller from 1984 is often cited as an early accelerator card. It handled 2D and 3D work but sold poorly, partly because of its price and a lack of software support.
Then, in 1997, 3Dlabs in the UK developed its Glint Gamma processor, the first programmable transform and lighting engine, and introduced the term GPU, which it defined as a geometry processor unit. That makes the word older than Nvidia’s marketing. For more on the timeline, Tom’s Hardware’s computer history timeline covers the earliest graphics processors.
What Made the GeForce 256 Different
The important change was transform and lighting, known as T&L. Before this chip, 3D accelerators left that work to the CPU. Building it into the graphics chip cut costs for consumer cards and improved performance at the same time.
| Spec | GeForce 256 (NV10) |
|---|---|
| Introduced | October 1999 |
| Manufacturing process | TSMC 220 nm |
| Clock speed | 120 MHz |
| Transistors | 17 million |
| Die size | 139 mm² |
| API support | DirectX 7.0 |
| Interface | AGP 4X |
For comparison, a modern chip is built on far smaller processes and is covered in how GPUs are made today.
How the Term GPU Spread
ATI answered quickly with its Radeon chip, but called it a visual processing unit, or VPU. That name didn’t stick. Nvidia’s version did, and the industry has used GPU for every graphics processor since.
Bottom line: Nvidia didn’t create graphics acceleration, and it didn’t coin the term. It shipped the first single-chip consumer processor with transform and lighting, and it named it loudly enough that the name stuck.
Today the chip and the card are two different things, which is why a GPU isn’t quite the same as a graphics card. To see how the modern version processes graphics, read how a GPU works, or browse Nvidia’s current GeForce cards.
Frequently Asked Questions
What was the first GPU?
Nvidia’s GeForce 256 in 1999 is marketed as the first, but earlier graphics chips existed, including 3Dlabs’ 1997 Glint Gamma, which introduced the term GPU.
Did Nvidia invent the GPU?
Nvidia popularized the term and shipped the first consumer chip with integrated transform and lighting, but it did not invent graphics acceleration or coin the word.
What was the first Nvidia GPU?
The GeForce 256, built around the NV10 chip and introduced in October 1999.
What is transform and lighting?
It’s the math that positions 3D shapes and calculates how light hits them. Before the GeForce 256, most consumer cards left this work to the CPU.
