The race to accelerate 3D in the mainstream consumer PC market began in the 1990s. It was the defunct 3DFX that managed to pave the way with its Voodoo and with Voodoo 2 it became the 3D queen of today. NVIDIA didn’t let itself wait long and was confronted by the Riva TNT and TNT 2, but without a doubt it was the NVIDIA GeForce 256 that was able to deliver a blow of authority so great that This marked the beginning of the end for 3DFX.

We can say that the GeForce 256 was the first true GPU because it was the first of its kind to fully address jobs that were processed on the CPU until thenor, alternatively, they needed additional hardware that improved performance but forced them to spend more money. These tasks are known as transform and lighting or T&L.

The GeForce 256 was the first GPU to handle transform and lighting tasks, because it did dedicated engine for that. These operations were not performed by software, which was possible on other graphics cards that lacked the mentioned engine. at the cost of a significant reduction in performance, but rather they sped up in that dedicated and exclusive engine that said the graphics card was supplied.

What was T&L and why was it so important?

We can consider them as the first two of the four most important steps that are taken. in the GPU pipeline. They are very computationally intensive and require the execution of very precise mathematical instructions. They are a very heavy load and running them through the CPU was not efficient at all.

They group both under a simple and easily understandable definition the process of describing 3D data of elements and objects present in different scenarios and frames of reference and create what is known as “gaps”divided into world spaces, visual spaces and screens. Each of these spaces is ideal for performing one or more of the operations required to create a 3D image.

For example, the world space is the one that contains all the 3D objects that are part of that 3D world, while the visual world is the one used for lighting and removal. The screen space is what is used to store the scene in the frame buffer. They all use different coordinate systems, i.e. 3D data it must be converted or transformed from one space to another as they move through the graphics pipeline.

Well, that’s where it comes into play. transformation enginewhich takes care of performing all the mathematical operations necessary to complete all the necessary transformation cycles. Like I said, it used to be done through the CPU.

The transform module converts 3D data from one reference frame to another reference frame and adapts all data to the current view before continuing The next step is lighting. This is necessary during the redraw process of each scene. After completion, lighting is applied, which is an essential step to achieve a 3D image of highly realistic quality.

The lighting module takes care of a set of math operations that must always be performed in order to work with this effect. A specializes in this task can calculate distance vectors of light sources from objects in a 3D sceneas well as the distance vectors from these objects to the viewer’s camera. A vector contains direction and distance information, hence its meaning.

This is also possible with this specialized engine separate length and distance information from direction information, as this separation simplifies the work to be done in future steps within the 3D GPU pipeline. All lighting calculations at the time were used for peak lighting as well as creating other important effects such as more realistic fog that relied on the distance between the camera and the subject rather than just the Z values ​​of the subject.

GeForce 256 marked a new era

I made it possible a big leap in graphics quality and performance thanks to the integration of these specialized transformation and lighting engines. It was one of the most important evolutions of the late 1990s in the world of 3D acceleration, although the migration of T&L functions from CPU to GPU took some time.

Its impact was so great that to keep their graphics cards from looking outdated, those who didn’t have GPUs with hardware T&L supported the feature through software. It was a bad idea, because while it did, games that required said technology worked the performance was terribly bad. It would be like trying to run software-accelerated ray tracing today, which is a bad idea.

Implementing this advancement at the hardware level For NVIDIA, it was crucial to beat all of its rivals in the 3D acceleration sector, including the then almighty 3DFX, which didn’t know its cards and not only lost the war, but was bought by NVIDIA. Without the GeForce 256, Jensen Huang’s company would not have become the green giant we know today.

Time does not pass in vain and today GeForce 256 specifications are those outdated graphics cards, but at the time It was a real object of desire.and a revolutionary graphics card. To close the article, I leave you with a summary with its complete specifications, including the selling price:

• NV10 core made in a 220 nm node.
• 17 million transistors.
• 4 pixel shaders.
• 4 texturing units.
• 4 raster units.
• 128-bit or 64-bit bus, depending on version.
• 32 MB of SDR or DDR memory, depending on the version.
• Bandwidth 4,800 GB/s for the version with DDR and 128-bit bus, 1,144 GB/s for the version with SDR and 64-bit bus.
• 120 MHz GPU.
• AGP 4x interface.
• TDP 50 watts.
• The DDR memory version had an introductory price of $279.