Tesla
Nvidia GPU microarchitecture with unified shaders
Tesla is a graphics processing unit microarchitecture developed by Nvidia and released in November 2006. It introduced unified shaders to replace fixed pipelines, powering the GeForce 8 through 300 series and Quadro cards.
- First Nvidia microarchitecture with unified shaders
- Released in November 2006
- Used 90 nm to 40 nm manufacturing processes
- Powered GeForce 8 through 300 series
- Replaced fixed-pipeline design with scalar processors
What it is
Tesla is a graphics processing unit microarchitecture created by Nvidia. It launched in November 2006 as the successor to the Curie design. This architecture marked a major shift by implementing unified shaders, replacing the older fixed-pipeline approach. It powered several generations of consumer and professional graphics cards, including the GeForce 8, 9, 100, 200, and 300 series. The design also supported Quadro professional visualisation modules and specific computing hardware. Manufactured using process nodes ranging from 90 nanometres down to 40 nanometres, Tesla enabled support for Direct3D 10 and OpenGL standards.
How it came about
Nvidia developed Tesla to move away from the separate functional units found in previous GPUs. Before this release, pixel and vertex shaders operated independently. The new design grouped these into a homogeneous collection of universal floating point processors. This change allowed the hardware to handle a wider variety of tasks more flexibly. The architecture competed directly with AMD TeraScale, which shared similar unified design principles derived from Xbox 360 work. Tesla served as the foundation for Nvidia graphics until the Fermi microarchitecture arrived to take its place.
Inside the chip
The core innovation was the use of scalar stream processors instead of vector units. Each processor handles one component at a time, simplifying construction while maintaining flexibility. These units run at high clock speeds to compensate for lower individual throughput. For example, the GeForce 8800 GTX stream processors operate at 1.35 GHz while the rest of the chip runs at 575 MHz. Each Streaming Multiprocessor contains eight shader processors and two special function units. The design supports 128-bit high dynamic range rendering and improved texture filtering algorithms. It also introduced coverage sampling anti-aliasing to enhance image quality without severe performance penalties.
How it was made
Chips based on this microarchitecture were produced using multiple manufacturing processes. Initial designs used a 90 nanometre node, with later iterations moving to 80, 65, 55, and 40 nanometres. This progression allowed for smaller dies and improved efficiency over the product lifespan. The family includes various chip identifiers such as G80, G92, and GT200. These silicon variants powered different tiers of graphics cards, from entry-level models to high-end enthusiast boards. The manufacturing diversity helped Nvidia cover a broad market segment with tailored performance levels.
Where it went
Tesla powered the GeForce 8, 9, 100, 200, and 300 series of consumer graphics cards. It also appeared in the GeForce 405 model. Professional users relied on Quadro FX, Quadro x000, and Quadro NVS series cards built on this architecture. Additionally, Nvidia Tesla computing modules utilized this design for general purpose processing. The architecture supported gaming, visualisation, and early parallel computing workloads through the CUDA platform. Its ability to handle unified shaders made it suitable for both graphics rendering and computational tasks.
Versions and family
The microarchitecture spawned numerous chip variants including G80, G84, G86, and G92. Later revisions like G92B and GT200B offered refinements. The GT200 series brought improvements to dual-issue instruction handling compared to the earlier G80 and G90 chips. While G8x and G9x chips lacked double precision support, the GT200 provided limited double precision capability. These variants allowed Nvidia to segment the market effectively, offering different performance tiers across consumer and professional lines.
Why it matters
Tesla established the unified shader model as the standard for modern graphics processing. By removing the distinction between vertex and pixel shaders, it simplified programming and improved hardware utilisation. This design choice influenced subsequent generations of GPU development across the industry. The introduction of scalar processors changed how graphics engines managed workloads, favouring flexibility over rigid vector operations. It laid the groundwork for Nvidia dominance in both gaming and professional visualisation markets during the late 2000s.

Questions people ask about Tesla
What was the main innovation of the Tesla microarchitecture?
Tesla introduced unified shaders, replacing separate pixel and vertex shader units with a single type of universal processor. This allowed the hardware to handle various tasks more flexibly and efficiently than previous fixed-pipeline designs.
Which graphics card series used the Tesla architecture?
The Tesla microarchitecture powered the GeForce 8, 9, 100, 200, and 300 series. It was also used in the GeForce 405, various Quadro professional cards, and Nvidia Tesla computing modules.
How did Tesla processors differ from previous GPU designs?
Previous GPUs used vector processing units. Tesla switched to scalar stream processors that handle one component at a time. This simplification allowed for higher clock speeds and greater flexibility in task assignment.
Good reading and sources
- Tesla (microarchitecture) on WikipediaEncyclopedia
- More photos on Wikimedia CommonsPhotos
- Wikidata recordOpen data
The facts on this page come from the Wikipedia article "Tesla (microarchitecture)", written by its contributors and used under the CC BY-SA 4.0 licence, and from its Wikidata record (CC0). EveryChip rewrote the text in its own words, with AI assistance, and compares every page with the article to make sure no passage is copied. It may be reused under the same CC BY-SA 4.0 licence with a link back.
It may contain mistakes, and figures change over time. Nothing here is investment or buying advice. If something is wrong, tell us. Last refreshed 2026-11-14.
Like Tesla
All graphics processorsMore from Nvidia
All of them
🇺🇸 United StatesRSX Reality Synthesizer
Graphics processor for the PlayStation 3 consoleGPU · GPU architecture · 2006