Curie
Nvidia GPU microarchitecture from 2004 for GeForce 6 and 7 series
Curie is a graphics processing unit microarchitecture developed by Nvidia and released in 2004. It powered the GeForce 6 and 7 series cards, introducing hardware support for advanced anti-aliasing and video decoding features before being replaced by the Tesla design.
- Released in 2004 as successor to Rankine
- Powers GeForce 6 and 7 series graphics cards
- Supports DirectX 9.0c and OpenGL 2.1
- Introduced first-generation Nvidia PureVideo
- Manufactured on 130 nm to 80 nm processes
What it is
Curie is the internal codename for a graphics processor design created by Nvidia. It launched in 2004 to succeed the earlier Rankine architecture. This design formed the foundation for the GeForce 6 and GeForce 7 series of graphics cards. It is named after the physicist Marie Curie. The architecture introduced several key multimedia and rendering improvements, including the first generation of Nvidia PureVideo technology for video playback. It also brought back support for Z compression and added hardware acceleration for multi-sample anti-aliasing. Due to its lack of unified shaders, it supports DirectX up to version 9.0c and OpenGL 2.1.
How it came about
Nvidia developed the Curie microarchitecture to replace the Rankine design. The company released it in 2004 as part of its ongoing strategy to improve graphics performance and multimedia capabilities. The naming convention follows Nvidia’s tradition of using famous scientists, with this iteration honouring Marie Salomea Skłodowska–Curie. It served as the bridge between the older fixed-function pipeline designs and the later unified shader models found in subsequent generations. After its run with the GeForce 6 and 7 series, Nvidia moved on to the Tesla microarchitecture.
Inside the chip
The Curie design does not use unified shaders, which limits its compatibility with newer graphics APIs. It supports DirectX 9.0c and Shader Model 3.0, along with OpenGL 2.1. A notable feature is the inclusion of hardware support for the MSAA anti-aliasing algorithm, allowing up to 4x smoothing of jagged edges. It also reintroduced Z compression to help manage memory bandwidth more efficiently. The first generation of Nvidia PureVideo was integrated into this architecture, providing dedicated hardware for video decoding tasks. These features were aimed at improving both gaming visual quality and media playback performance on personal computers.
How it was made
Chips based on the Curie microarchitecture were manufactured using several different process nodes. The production technologies included 130 nm, 110 nm, 90 nm, and 80 nm processes. This range of manufacturing scales allowed Nvidia to produce various tiers of graphics cards, from entry-level to high-end models, under the GeForce 6 and 7 branding. The specific transistor counts and clock speeds varied across these different implementations and product lines.
Where it went
This microarchitecture was used exclusively in Nvidia’s GeForce 6 and GeForce 7 series graphics processing units. These cards were widely deployed in desktop personal computers for gaming and general multimedia use. The support for DirectX 9.0c made them suitable for many games of that era, while the PureVideo technology enhanced their appeal for home theatre PC builds and video playback applications. The architecture also supported Scalable Link Interface technology for linking multiple GPUs.
Versions and family
The Curie architecture underpins two main product families: the GeForce 6 series and the GeForce 7 series. It succeeded the Rankine microarchitecture and was eventually replaced by the Tesla microarchitecture. Within these series, various models were produced using different manufacturing processes, ranging from 130 nm down to 80 nm. The architecture itself remained consistent in its core features, such as the lack of unified shaders and support for DirectX 9.0c, across all its variants.
Why it matters
Curie marked a significant step in Nvidia’s graphics lineage by introducing hardware-accelerated video decoding through PureVideo. This feature helped establish GPUs as essential components for media consumption, not just gaming. The re-introduction of Z compression and hardware MSAA support improved rendering efficiency and image quality for users. As the last architecture to support DirectX 9.0c without unified shaders, it represents the end of an era in graphics programming before the industry shifted towards more flexible shader models in the subsequent Tesla generation.
Questions people ask about Curie
What graphics cards use the Curie microarchitecture?
The Curie microarchitecture is used in the Nvidia GeForce 6 series and the GeForce 7 series of graphics processing units. These cards were popular for gaming and multimedia tasks during the mid-2000s.
Which version of DirectX does Curie support?
Curie supports DirectX 9.0c and Shader Model 3.0. It does not support newer versions like DirectX 10 or 11 because it lacks unified shaders, which were introduced in later architectures.
Who is the Curie architecture named after?
The architecture is named after the Polish physicist Marie Salomea Skłodowska–Curie. This follows Nvidia’s practice of naming its GPU microarchitectures after prominent scientists and historical figures.
Good reading and sources
The facts on this page come from the Wikipedia article "Curie (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.