Radeon R420
Encyclopedia
The Radeon
R420 core from ATI Technologies
was the company's basis for its 3rd-generation DirectX
9.0/OpenGL
2.0-capable graphics cards
. Used first on the Radeon X800, R420 was produced on a 0.13 micrometer
(130 nm) low-K
process and used GDDR-3 memory. The chip was designed for AGP
.
Driver support of this core was discontinued as of Catalyst 9.4, and as a result there is no official Windows 7 support for any of the X700 - X850 products.
-based GPUs. R420 basically takes a "wider is better" approach to the previous architecture, with some small tweaks thrown in to enhance it in various ways. The chip came equipped with over double the pixel and vertex pushing resources compared to Radeon 9800 XT, with 16 DirectX 9.0b
pixel pipelines and 16 ROPs
. One would not be far off seeing the X800 XT basically as a pair of Radeon 9800 cores connected together and also running with a ~30% higher clock speed.
The R420 design was a 4 "quad" arrangement (4 pipelines per quad.) This organization internally allowed ATI to disable defective "quads" and sell chips with 12, 8 or even 4 pixel pipelines, an evolution of the technique used with Radeon 9500/9700 and 9800SE/9800. The separation into "quads" also allowed ATI to design a system to optimize the efficiency of the overall chip. Coined the "quad dispatch system", the screen is tiled and work is spread out evenly among the separate "quads" to optimize their throughput. This is how the R300-series chips performed their tasks as well, but R420 refined this by allowing programmable tile sizes in order to control work flow on a finer level of granularity. Apparently by reducing tile sizes, ATI was able to optimize for different triangle sizes.
When ATI doubled the number of pixel pipelines, they also raised the number of vertex shader engines from 4 to 6. This changed the ratio of pixel/vertex shaders from 2:1 (on R300) to 8:3, showing that ATI believed the workload in games as of 2004 and onward to be more pixel shader and texturing oriented than geometry based. Normal
and parallax mapping
were replacing sheer geometric complexity for model detail, so undoubtedly that was part of the reasoning. Strangely, the X700 mainstream card (RV410) had 6 vertex shaders while only being equipped with 2 quads. As such, this chip was obviously designed for a heavier geometry load than texturing, perhaps being tailored for a role as a FireGL chip. RV410 also significantly outgunned NVIDIA's GeForce 6600GT
(3 vertex shaders) on geometry throughput. With R420's and RV410's 6 vertex shaders combined with higher clock speeds than the previous generation, ATI was able to more than double the geometry processing capability of 9800XT.
Although the R420-based chips are fundamentally similar to R300-based cores, ATI did tweak and enhance the pixel shader units for more flexibility. A new pixel shader version (PS2.b) allowed slightly greater shader program flexibility than plain PS2.0, but was still shy of full PS3.0 capabilities. This new revision to PS2.0 increased the maximum number of instructions and registers available to pixel shader programs.
ATI revealed temporal anti-aliasing, a new anti-aliasing
technology their chips were capable of. By taking advantage of the frame-to-eye effects of a framerate higher than 60 frame/s, the GPU is able to better smooth aliased edges by rotating the anti-aliasing sampling pattern between frames. A 2X software setting became perceptively equivalent to 4X. Unfortunately, it required the system to be able to maintain at least 60 frame/s or temporal anti-aliasing would cause a noticeable flickering, because the user would be able to see the alternating AA patterns. If the framerate could not be maintained, the driver will disable temporal AA. However, in games which this performance level could be maintained, temporal AA was a nice addition to ATI's excellent anti-aliasing options.
Another notable addition to the core was a new kind of normal map
compression, dubbed "3Dc
". Similar to how texture compression
had been part of the Direct3D
specification for years and was used for compressing regular textures, normal map compression compacted this new type of surface detail layer. Because DirectX Texture Compression (DXTC) was block-based and not designed for a normal map's different data properties, a new compression method was needed to prevent loss of detail and other artifacting. 3Dc was based on a modified DXT5 mode, which in fact was a fallback option for hardware not supporting 3Dc. Software making heavy use of normal mapping could gain a significant speed boost from the savings in fillrate and bandwidth by using 3Dc. ATI showcased many of their chip's new features in the promotional real-time demo called, Ruby: The Doublecross.
Most of the rest of the GPU was extremely similar to R300. The memory controller and memory bandwidth optimization
techniques (HyperZ
) were identical.
R420 was actually a secondary 4th generation project for ATI, with the original R400 plan being scrapped. R400 would have been more feature-complete, likely with full PS3.0 support among other enhancements, but it is believed that ATI deemed R400 unnecessarily complex for the applications that would be available, and potentially risky to develop on the available semiconductor manufacturing processes of the time. R400 technology was thus moved to the subsequent generation, renamed to "R500" (became "R520
"), while the 4th generation was served with the R300-derived R420.
The earliest Radeon X800 series cards were based on the R420 core. The line included the Radeon X800 XT Platinum Edition and the Radeon X800 Pro. The X800 XT PE came clocked at 520 MHz core and 560 MHz RAM, with 16 pipelines enabled. The X800 Pro came clocked at 475/450 MHz with one quad disabled, leaving 12 pixel pipelines functional. Essentially, the X800 Pro is built 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 X300 and X600, little more than PCI Express
versions of the Radeon 9600
series, were intended to be the new mainstream products. X300 was based on RV370, a 110 nm chip, while X600 used RV380 which was built with the high-performance 130 nm Low-K
process. Both had identical feature-sets, however. Later the X550 was created, a quietly launched addition to the Radeon X series set to replace the X300 series, using the same chip as X300 (RV370).
The X700 (RV410) series replaced the X600 in September 2004. X700 Pro is clocked at 425 MHz core, and produced on a 0.11 micrometre process. RV410 used an interesting layout, consisting of 8 pixel pipelines connected to 4 ROPs
(similar to GeForce 6600) while maintaining the 6 vertex shaders of X800. The 110 nm process was a cost-cutting process, designed not for high clock speeds but for reducing die size while maintaining high yields. An X700 XT was planned for production, and reviewed by various hardware web sites, but was never released. It was believed that X700 XT set too high of a clock ceiling for ATI to profitably produce. X700 XT was also not adequately competitive with nVidia's impressive GeForce 6600GT. ATI would go on produce a card in the X800 series to compete instead.
The Radeon X800 "R430"-based 110 nanometer series was introduced at the end of 2004 along with ATI's new X850 cards. The X800 was designed to replace the position X700 XT failed to secure, with 12 pipelines and a 256-bit RAM bus. The card more than surpassed the 6600GT with performance similar to that of the GeForce 6800. A close relative, the new X800 XL, was positioned to dethrone NVIDIA's GeForce 6800 GT with higher memory speeds and a full 16 pipelines to boost performance. R430 was unable to reach high clock speeds, being mainly designed to reduce the cost per GPU, and so a new top-of-the-line core was still needed. The new high-end R4x0-generation arrived with the X850 series, equipped with various core tweaks for slightly higher performance than the "R420"-based X800 series. The "R480"-based X850 line was available in 3 forms: the X850 Pro, the X850 XT, and the X850 XT Platinum Edition, and was built on the reliable high-performance 130 nm Low-K
process.
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. It used any "R480" X850 die or "R430" X800 XL die that had 2 functional quads and could run at 475 MHz. They were meant to compete with the GeForce 6600GT beside the previous "R430"-based X800. ATI also released the X800 GTO, which was a 12 pipeline card (3 quads) using either "R480" or "R430" dies clocked at 400 MHz. This card performed between the X800 GT and the X800 XL. It was faster than the plain GeForce 6800, but slower than GeForce 6800 GT. High sales for this card were due to its relatively high performance coupled with a cost only slightly higher than the X800 GT. The overclocking community discovered that the R480-based GTO could frequently reach clock speeds near the X850 XT.
Finally, another SKU was the X800 GTO², again based on R480. It was again manufactured by Sapphire Technology
, like the X800 GTO. This card usually came with a 3 quad configuration, like X800 GTO. The GTO² was unique in the GTx series because, with a BIOS change, they could almost always be turned into a full 4 quad card. Some X800 GTO² cards shipped with the full 4 quads already enabled, but of these some were R430 instead of R480 and weren't able to reach X850-like clock speeds. The final variations of the GTO series were the special GTO boards with 16 pipelines officially enabled, such as Powercolor's "R430"-based X800 GTO-16.
Radeon
Radeon is a brand of graphics processing units and random access memory produced by Advanced Micro Devices , first launched in 2000 by ATI Technologies, which was acquired by AMD in 2006. Radeon is the successor to the Rage line. There are four different groups, which can be differentiated by...
R420 core from ATI Technologies
ATI Technologies
ATI Technologies Inc. was a semiconductor technology corporation based in Markham, Ontario, Canada, that specialized in the development of graphics processing units and chipsets. Founded in 1985 as Array Technologies Inc., the company was listed publicly in 1993 and was acquired by Advanced Micro...
was the company's basis for its 3rd-generation DirectX
DirectX
Microsoft DirectX is a collection of application programming interfaces for handling tasks related to multimedia, especially game programming and video, on Microsoft platforms. Originally, the names of these APIs all began with Direct, such as Direct3D, DirectDraw, DirectMusic, DirectPlay,...
9.0/OpenGL
OpenGL
OpenGL is a standard specification defining a cross-language, cross-platform API for writing applications that produce 2D and 3D computer graphics. The interface consists of over 250 different function calls which can be used to draw complex three-dimensional scenes from simple primitives. OpenGL...
2.0-capable graphics cards
Graphics processing unit
A graphics processing unit or GPU is a specialized circuit designed to rapidly manipulate and alter memory in such a way so as to accelerate the building of images in a frame buffer intended for output to a display...
. Used first on the Radeon X800, R420 was produced on a 0.13 micrometer
Micrometre
A micrometer , is by definition 1×10-6 of a meter .In plain English, it means one-millionth of a meter . Its unit symbol in the International System of Units is μm...
(130 nm) low-K
Low-K
In semiconductor manufacturing, a low-κ dielectric is a material with a small dielectric constant relative to silicon dioxide. Although the proper symbol for the dielectric constant is the Greek letter κ , in conversation such materials are referred to as being "low-k" rather than "low-κ"...
process and used GDDR-3 memory. The chip was designed for AGP
Accelerated Graphics Port
The Accelerated Graphics Port is a high-speed point-to-point channel for attaching a video card to a computer's motherboard, primarily to assist in the acceleration of 3D computer graphics. Since 2004 AGP has been progressively phased out in favor of PCI Express...
.
Driver support of this core was discontinued as of 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 supported DirectX features, R420 (codenamed Loki) was very similar to the R300Radeon R300
The Radeon R300 is the third generation of Radeon graphics chips from ATI Technologies. The line features 3D acceleration based upon Direct3D 9.0 and OpenGL 2.0, a major improvement in features and performance compared to the preceding Radeon R200 design. R300 was the first fully Direct3D...
-based GPUs. R420 basically takes a "wider is better" approach to the previous architecture, with some small tweaks thrown in to enhance it in various ways. The chip came equipped with over double the pixel and vertex pushing resources compared to Radeon 9800 XT, with 16 DirectX 9.0b
DirectX
Microsoft DirectX is a collection of application programming interfaces for handling tasks related to multimedia, especially game programming and video, on Microsoft platforms. Originally, the names of these APIs all began with Direct, such as Direct3D, DirectDraw, DirectMusic, DirectPlay,...
pixel pipelines and 16 ROPs
Render Output unit
The Render Output Unit, often abbreviated as "ROP", and sometimes called Raster Operations Pipeline, is one of the final steps in the rendering process of modern 3D accelerator boards. The pixel pipelines take pixel and texel information and process it, via specific matrix and vector operations,...
. One would not be far off seeing the X800 XT basically as a pair of Radeon 9800 cores connected together and also running with a ~30% higher clock speed.
The R420 design was a 4 "quad" arrangement (4 pipelines per quad.) This organization internally allowed ATI to disable defective "quads" and sell chips with 12, 8 or even 4 pixel pipelines, an evolution of the technique used with Radeon 9500/9700 and 9800SE/9800. The separation into "quads" also allowed ATI to design a system to optimize the efficiency of the overall chip. Coined the "quad dispatch system", the screen is tiled and work is spread out evenly among the separate "quads" to optimize their throughput. This is how the R300-series chips performed their tasks as well, but R420 refined this by allowing programmable tile sizes in order to control work flow on a finer level of granularity. Apparently by reducing tile sizes, ATI was able to optimize for different triangle sizes.
When ATI doubled the number of pixel pipelines, they also raised the number of vertex shader engines from 4 to 6. This changed the ratio of pixel/vertex shaders from 2:1 (on R300) to 8:3, showing that ATI believed the workload in games as of 2004 and onward to be more pixel shader and texturing oriented than geometry based. Normal
Normal mapping
In 3D computer graphics, normal mapping, or "Dot3 bump mapping", is a technique used for faking the lighting of bumps and dents. It is used to add details without using more polygons. A common use of this technique is to greatly enhance the appearance and details of a low polygon model by...
and parallax mapping
Parallax mapping
Parallax mapping is an enhancement of the bump mapping or normal mapping techniques applied to textures in 3D rendering applications such as video games...
were replacing sheer geometric complexity for model detail, so undoubtedly that was part of the reasoning. Strangely, the X700 mainstream card (RV410) had 6 vertex shaders while only being equipped with 2 quads. As such, this chip was obviously designed for a heavier geometry load than texturing, perhaps being tailored for a role as a FireGL chip. RV410 also significantly outgunned NVIDIA's GeForce 6600GT
GeForce 6 Series
The GeForce 6 Series is Nvidia's sixth generation of GeForce graphic processing units. Launched on April 14, 2004, the GeForce 6 family introduced PureVideo post-processing for video, SLI technology, and Shader Model 3.0 support .-GeForce 6 Series features:-SLI:The Scalable Link...
(3 vertex shaders) on geometry throughput. With R420's and RV410's 6 vertex shaders combined with higher clock speeds than the previous generation, ATI was able to more than double the geometry processing capability of 9800XT.
Although the R420-based chips are fundamentally similar to R300-based cores, ATI did tweak and enhance the pixel shader units for more flexibility. A new pixel shader version (PS2.b) allowed slightly greater shader program flexibility than plain PS2.0, but was still shy of full PS3.0 capabilities. This new revision to PS2.0 increased the maximum number of instructions and registers available to pixel shader programs.
ATI revealed temporal anti-aliasing, a new anti-aliasing
Anti-aliasing
In digital signal processing, spatial anti-aliasing is the technique of minimizing the distortion artifacts known as aliasing when representing a high-resolution image at a lower resolution...
technology their chips were capable of. By taking advantage of the frame-to-eye effects of a framerate higher than 60 frame/s, the GPU is able to better smooth aliased edges by rotating the anti-aliasing sampling pattern between frames. A 2X software setting became perceptively equivalent to 4X. Unfortunately, it required the system to be able to maintain at least 60 frame/s or temporal anti-aliasing would cause a noticeable flickering, because the user would be able to see the alternating AA patterns. If the framerate could not be maintained, the driver will disable temporal AA. However, in games which this performance level could be maintained, temporal AA was a nice addition to ATI's excellent anti-aliasing options.
Normal mapping
In 3D computer graphics, normal mapping, or "Dot3 bump mapping", is a technique used for faking the lighting of bumps and dents. It is used to add details without using more polygons. A common use of this technique is to greatly enhance the appearance and details of a low polygon model by...
compression, dubbed "3Dc
3Dc
3Dc , also known as DXN, BC5, or Block Compression 5 is a lossy data compression algorithm for normal maps invented and first implemented by ATI. It builds upon the earlier DXT5 algorithm and is an open standard...
". Similar to how texture compression
Texture compression
Texture compression is a specialized form of image compression designed for storing texture maps in 3D computer graphics rendering systems. Unlike conventional image compression algorithms, texture compression algorithms are optimized for random access....
had been part of the Direct3D
Direct3D
Direct3D is part of Microsoft's DirectX application programming interface . Direct3D is available for Microsoft Windows operating systems , and for other platforms through the open source software Wine. It is the base for the graphics API on the Xbox and Xbox 360 console systems...
specification for years and was used for compressing regular textures, normal map compression compacted this new type of surface detail layer. Because DirectX Texture Compression (DXTC) was block-based and not designed for a normal map's different data properties, a new compression method was needed to prevent loss of detail and other artifacting. 3Dc was based on a modified DXT5 mode, which in fact was a fallback option for hardware not supporting 3Dc. Software making heavy use of normal mapping could gain a significant speed boost from the savings in fillrate and bandwidth by using 3Dc. ATI showcased many of their chip's new features in the promotional real-time demo called, Ruby: The Doublecross.
Most of the rest of the GPU was extremely similar to R300. The memory controller and memory bandwidth optimization
Bandwidth optimization
Bandwidth optimization is one of many concerns webmasters deal with when hosting web content. Since most web hosts charge by bandwidth used or have an account limit on bandwidth, a prudent webmaster will squeeze and compress their files as much as possible without affecting the content's integrity....
techniques (HyperZ
HyperZ
HyperZ is the name of a set of computer graphics processing techniques used by ATI Technologies in their Radeon video cards.On the Radeon R100-based cores, Radeon DDR through 7500, where HyperZ debuted, ATI claimed a 20% improvement in overall rendering efficiency...
) were identical.
R420 was actually a secondary 4th generation project for ATI, with the original R400 plan being scrapped. R400 would have been more feature-complete, likely with full PS3.0 support among other enhancements, but it is believed that ATI deemed R400 unnecessarily complex for the applications that would be available, and potentially risky to develop on the available semiconductor manufacturing processes of the time. R400 technology was thus moved to the subsequent generation, renamed to "R500" (became "R520
Radeon R520
ATI's "R520" core is the foundation for a line of DirectX 9.0c and OpenGL 2.0 3D accelerator X1000 video cards. It is ATI's first major architectural overhaul since the "R300" core and is highly optimized for Shader Model 3.0. The Radeon X1000 series using the core was introduced on October 5,...
"), while the 4th generation was served with the R300-derived R420.
Chronology of releases
The X300 and X600, little more than PCI Express
PCI Express
PCI Express , officially abbreviated as PCIe, is a computer expansion card standard designed to replace the older PCI, PCI-X, and AGP bus standards...
versions of the Radeon 9600
Radeon R300
The Radeon R300 is the third generation of Radeon graphics chips from ATI Technologies. The line features 3D acceleration based upon Direct3D 9.0 and OpenGL 2.0, a major improvement in features and performance compared to the preceding Radeon R200 design. R300 was the first fully Direct3D...
series, were intended to be the new mainstream products. X300 was based on RV370, a 110 nm chip, while X600 used RV380 which was built with the high-performance 130 nm Low-K
Low-K
In semiconductor manufacturing, a low-κ dielectric is a material with a small dielectric constant relative to silicon dioxide. Although the proper symbol for the dielectric constant is the Greek letter κ , in conversation such materials are referred to as being "low-k" rather than "low-κ"...
process. Both had identical feature-sets, however. Later the X550 was created, a quietly launched addition to the Radeon X series set to replace the X300 series, using the same chip as X300 (RV370).
The X700 (RV410) series replaced the X600 in September 2004. X700 Pro is clocked at 425 MHz core, and produced on a 0.11 micrometre process. RV410 used an interesting layout, consisting of 8 pixel pipelines connected to 4 ROPs
Render Output unit
The Render Output Unit, often abbreviated as "ROP", and sometimes called Raster Operations Pipeline, is one of the final steps in the rendering process of modern 3D accelerator boards. The pixel pipelines take pixel and texel information and process it, via specific matrix and vector operations,...
(similar to GeForce 6600) while maintaining the 6 vertex shaders of X800. The 110 nm process was a cost-cutting process, designed not for high clock speeds but for reducing die size while maintaining high yields. An X700 XT was planned for production, and reviewed by various hardware web sites, but was never released. It was believed that X700 XT set too high of a clock ceiling for ATI to profitably produce. X700 XT was also not adequately competitive with nVidia's impressive GeForce 6600GT. ATI would go on produce a card in the X800 series to compete instead.
Low-K
In semiconductor manufacturing, a low-κ dielectric is a material with a small dielectric constant relative to silicon dioxide. Although the proper symbol for the dielectric constant is the Greek letter κ , in conversation such materials are referred to as being "low-k" rather than "low-κ"...
process.
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
Stock Keeping Unit
→A stock-keeping unit or SKU is a number or code used to identify each unique product or item for sale in a store or other business.It is a unique identifier for each distinct product and service that can be purchased...
was created, the X800 GT. It used any "R480" X850 die or "R430" X800 XL die that had 2 functional quads and could run at 475 MHz. They were meant to compete with the GeForce 6600GT beside the previous "R430"-based X800. ATI also released the X800 GTO, which was a 12 pipeline card (3 quads) using either "R480" or "R430" dies clocked at 400 MHz. This card performed between the X800 GT and the X800 XL. It was faster than the plain GeForce 6800, but slower than GeForce 6800 GT. High sales for this card were due to its relatively high performance coupled with a cost only slightly higher than the X800 GT. The overclocking community discovered that the R480-based GTO could frequently reach clock speeds near the X850 XT.
Finally, another SKU was the X800 GTO², again based on R480. It was again manufactured by Sapphire Technology
Sapphire Technology
Sapphire Technology is a Hong Kong-based technology company, which produces graphics cards for personal computers and workstations, motherboards, TV tuner cards, digital audio players and LCDTVs...
, like the X800 GTO. This card usually came with a 3 quad configuration, like X800 GTO. The GTO² was unique in the GTx series because, with a BIOS change, they could almost always be turned into a full 4 quad card. Some X800 GTO² cards shipped with the full 4 quads already enabled, but of these some were R430 instead of R480 and weren't able to reach X850-like clock speeds. The final variations of the GTO series were the special GTO boards with 16 pipelines officially enabled, such as Powercolor's "R430"-based X800 GTO-16.
External links
- "Inside ATI's R420 (architectural analysis)" by aths, 3DCenter.Org, May 4, 2004
- ATi's X800 Pulls Off Another Coup in the Graphics Performance War — features, architecture and benchmarks of the Radeon X800
- ATI's Radeon X700 XT graphics card — An in-depth look at the unreleased X700 XT.