Radeon R400 series

The R420 GPU, developed by ATI Technologies, was the company's basis for its third-generation DirectX 9.0/OpenGL 2.0 compatible graphics cards. First used in the Radeon X800, the R420 was produced in a 0.13 micrometer (30 nm) low-K photolithography process and used GDDR3 memory. The chip was designed for AGP graphics cards.

Driver support for this kernel was discontinued starting with Catalyst 9.4 and as a result, there is no official Windows 7 support for any of the X700 - X850 products.

Development

In terms of DirectX features supported, R420 (codenamed Loki) was very similar to R300. R420 basically takes a "wider, better" of the previous architecture, with some small adjustments to improve it in various ways. The chip came equipped with more than double the pixel and vertex pushing resources compared to the Radeon 9800 One wouldn't be far from seeing the X800 XT as basically a pair of Radeon 9800 cores connected together and also running at a ~30% higher clock speed.

The design of the R420 was a 4 "quad" (4 pipes per quad) This organization internally allowed ATI to disable "quads" defective and sell chips with pipelines of 12, 8 or even 4 pixels, an evolution of the technique used with the Radeon 9500/9700 and 9800SE/9800. The separation into "quads" It also allowed ATI to design a system to optimize overall chip efficiency. Coined the "quad dispatch system", the screen is divided into tiles and work is distributed evenly among the "quads" separated to optimize their performance. This is how the R300 series chips also performed their tasks, but the R420 refined this by allowing programmable tile sizes to control the workflow at a finer level of granularity. Apparently, by reducing the size of the tiles, ATI was able to optimize for different triangle sizes.

When ATI doubled the number of pixel pipelines, it also raised the number of vertex shader engines from 4 to 6. This changed the pixel/vertex shader ratio from 2:1 (on R300) to 8:3, showing that ATI believed that the workload in games as of 2004 was more geared toward pixel shading and textures than geometry. Normal and parallax mapping were replacing pure geometric complexity with model detail, so that was certainly part of the reasoning. Interestingly, the main X700 board (RV410) had 6 vertex shaders and was only equipped with 2 quads. As such, this chip was obviously designed for a heavier geometry load than texturing, perhaps designed for a role like a FireGL chip. The RV410 also significantly outperformed NVIDIA's GeForce 6600GT (3-vertex shaders) in geometry performance. With the R420 and RV410's 6-vertex shaders combined with higher clock speeds than the previous generation, ATI was able to more than double the 9800XT's geometry processing power.

Although R420-based chips are fundamentally similar to R300-based cores, ATI modified and improved the pixel shading units for greater flexibility. A new version of pixel shading (PS2.b) allowed slightly more flexibility of the shading program than the simple PS2.0, but still did not reach the full capabilities of PS3.0. This new revision of PS2.0 increased the maximum number of instructions and registers available for pixel shader programs.

ATI revealed Temporal Anti-Aliasing, a new anti-aliasing technology that its chips were capable of. By taking advantage of the frame-by-eye effects of a frame rate above 60 frames/s, the GPU can better smooth aliased edges by rotating the anti-aliasing sampling pattern between frames. A 2X software configuration became perceptually equivalent to 4X. Unfortunately, it required the system to be able to maintain at least 60 frames per second, or temporal anti-aliasing would cause noticeable flickering, because the user would be able to see the alternating AA patterns. If the frame rate could not be maintained, the driver will disable temporary AA. However, in games where this level of performance could be maintained, Temporal AA was a nice addition to ATI's excellent anti-aliasing options. Please note that "Temporary AA" of ATI was actually a temporal dithering filter for spatial AA, not a de facto temporal smoothing (which must involve a controlled combination of the temporal subsamples of consecutive displays).

Another notable addition to the core was a new type of normal map compression, called "3Dc". Similar to how texture compression had been part of the Direct3D specification for years and was used to compress regular textures, regular map compression compacted this new type of surface detail layer. Because DirectX Texture Compression (DXTC) was block-based and not designed for the different data properties of a normal map, a new compression method was needed to avoid loss of detail and other artifacts. 3Dc was based on a modified DXT5 mode, which was in fact an alternative option for hardware that did not support 3Dc. Software that makes heavy use of normal mapping could gain a significant speed boost from fill rate and bandwidth savings by using 3Dc. ATI showed off many of its chip's new features in the real-time promotional demo called Ruby: The Doublecross.

Most of the rest of the GPU was extremely similar to the R300. The memory controller and memory bandwidth optimization techniques (HyperZ) were identical.

R420 was actually a fourth-generation side project for ATI, with the original R400 plan, internally codenamed 'Crayola', being scrapped. R400 would have been more feature-rich, with Unified Shader Model 3 shading support among other improvements, but it is believed that ATI considered R400 to be unnecessarily complex for the applications that would be available and potentially risky to develop in the available semiconductor manufacturing processes of the time. The R400 architecture was implemented as well as the Xenos chip used in the Xbox 360 video game console, and became the basis for the Qualcomm Adreno 200 mobile GPU, initially called AMD Z430. In the Radeon line, support for Direct3D 9.0c features It moved to the next generation based on the R500 architecture, while the fourth generation was served with the R420 derived from R300.

R420 and next ATI GPU launches without major architectural changes

The first Radeon X800 series cards were based on the R420 core. The lineup included the Radeon X800 XT Platinum Edition and the Radeon X800 Pro. The X800 XT PE came with a 520 MHz core and 560 MHz RAM, with 16 pipelines enabled. The X800 Pro came with a frequency of 475/450 MHz with quad disabled, leaving the 12 pixel pipelines functional. Essentially, the X800 Pro is based on semi-defective R420 cores. An X800 Pro VIVO (Video-in-Video-out) was also released and was popular with overclockers because the disabled quad could usually be enabled, resulting in a fully functional X800 XT PE at a lower cost.

The Radeon X700 series (RV410) replaced the X600 in September 2004. RV410 used a design consisting of 8 pixel pipelines connected to 4 ROPs (similar to GeForce 6 600) while maintaining the 6 vertex shaders of X800. 110nm was a cost-cutting process, designed not for high clock speeds, but to reduce die size while maintaining high yields. An X700 XT was planned for production and reviewed by several hardware websites, but it was never released. The X700 XT was believed to set a clock ceiling too high for ATI to produce profitably. The X700 XT was also not competitive enough with nVidia's impressive GeForce 6 600GT. ATI would go on to produce a card in the X800 series to compete instead.

The 110 nanometer series based on Radeon X800 "R430" It was introduced in late 2004 along with ATI's new X850 cards. The X800 was designed to replace the position that the X700 XT could not secure, with 12 pipelines and a 256-bit RAM bus. The card far outperformed the 6600GT with performance similar to the GeForce 6 800. A close relative, the new X800 XL, was positioned to dethrone NVIDIA's GeForce 6800 GT with memory speeds higher frequencies and 16 full channels for increased performance. R430 was unable to achieve high clock speeds as it was primarily designed to reduce the cost per GPU, so a new top-of-the-line core was still needed. The new high-end R4x0 generation arrived with the X850 series, equipped with several basic settings for slightly better performance than the 'R420'-based X800 series. The X850 line based on "R480" It was available in 3 forms: the X850 Pro, the X850 XT and the X850 XT Platinum Edition, and was based on the reliable Low- High performance 130nm K.

In 2005, ATI had a large number of dies that "worked" but not well enough to be used on the X800 or X850 series cards. So a new SKU was created, the X800 GT. Used any "R480" X850 or die "R430" X800 XL that had 2 functional quads and could run at 475 MHz. They were intended to compete with the GeForce 6600GT along with the previous "R430"-based X800. ATI also released the X800 GTO, which was a 12-pipe (3 quads) card that used "R480" or "R430" with a frequency of 400 MHz. This card ran between the X800 GT and the X800 XL. It was faster than the regular GeForce 6 800, but slower than the GeForce 6800 GT. The high sales of this card were due to its relatively high performance along with a slightly higher cost than the X800 GT. The overclocking community found that the R480-based GTO was frequently able to reach clock speeds close to the X850 XT.

Finally, another SKU was the X800 GTO², again based on the R480. It was again manufactured by Sapphire Technology, as was the X800 GTO. This card usually came with a 3 quad configuration, like the X800 GTO. The GTO² was unique in the GTx series because, with a BIOS change, it could almost always be converted to a full quad 4 card. Some X800 GTO² cards shipped with the full 4 quads already enabled, but some of them were R430 instead of R480 and were unable to achieve similar clock speeds as the X850. The final variations of the GTO series were special GTO boards with 16 officially enabled channels, such as the X800 GTO-16 based on "R430" by Powercolor.

Models