R600 (ASIC)

The Radeon HD 2000 series is a series of desktop graphics chip company AMD and the successor to the Radeon X1000 series. This makes it the sixth-generation GPUs with the name ATI Radeon. All graphics processors of this series support pixel, geometry and vertex shader 4.0 to DirectX 10 The mobile graphics solutions are marketed as ATI Mobility Radeon HD 2000 series. It was replaced by the ATI Radeon HD 3000 series.

• 4.1 Performance

History

The architecture of the R600 GPUs and their derivatives, which are used in the Radeon HD 2000 series is used, although it has never been used in graphics cards for PCs, not entirely new. It is based in principle on the Xenos GPU architecture of the Xbox 360, which bore the code name R400. At the time when the architecture was created for the Xbox 360, it was planned, even the desktop version to create this architecture. This idea, however, was then discarded, and both the Radeon X series and the Radeon X1000 series were made at the basis of the second Radeon architecture, which was used on the Radeon 8000 series.

After the initial planning of the AMD Radeon HD 2000 series should be published before Christmas 2006. However, this date could not be maintained, in fact it was ( which was introduced with DirectX 10 ), allowing both Windows Vista as introduced by Nvidia's GeForce 8 series competition. Nvidia even managed not only to present the high-end models ahead of AMD, but also the mainstream cards were in April 2007, and thus, presented and sold before the release of the HD 2000 series.

Unlike the competition, AMD introduced on 14 May 2007 then not only the high-end model ( HD 2900 XT) before, but also presented the technical data of the mainstream graphics cards ( HD 2600 ) and the low-end graphics cards (HD 2400). However, readily salable was only the Radeon HD 2900 XT.

Surprisingly, the top model in the launch of the AMD HD 2000 series, the Radeon HD 2900 XT not against the top models of Nvidia, the GeForce 8800 Ultra or GTX, but in terms of performance and priced only slightly slower against the 8800 GTS. In this regard, the Radeon HD 2900 XT was not convincing in all respects. In addition, the Radeon HD 2900 XT has a very high power consumption, all of which contained at the time of conception on the market - even faster - desktop graphics cards exceeds. The published in August 2007 Catalyst 7.8 driver brought a little more power in AA and AF. The Catalyst 7.10 ( October 2007) eliminated more driver error ( though not all ) on Windows Vista and increased performance with Crossfire systems. One of many hoped or expected " miracle driver " brings the power of the HD 2900 XT to a level corresponding to the high power consumption, but never materialized.

On 28 June 2007 was followed by the complete idea of the mainstream and low -end graphics cards, which, inter alia, in the case of disclosure agreements ( NDAs ) reflected for benchmarks. The first produced in 65nm manufacturing technology GPUs were in terms of power consumption exactly the opposite of the Radeon HD 2900 XT: presented for DirectX 10 graphics cards, the HD 2400 series sets new records in terms of low power on, but the HD -2600 series was very frugal. In the used on these series GPUs Unified Video Decoder is also installed for the first time, can be decoded on the footage without much processor load on the graphics card.

In autumn 2007, then followed by further additions to the product range. First was 2600 XT X2 presented by the board partners in September, the Radeon HD. For this graphics card a completely new board layout was developed because it consists of a graphics card with two GPUs. This was also reflected in the sale price, which is why relatively few batches were produced. As with the AMD Radeon HD 3870 X2 placed a dual graphics chip card in the high-end in early 2008, are also likely to have been won an argument experiences through the development to produce these.

Since the end of 2007 with the RV670 ( HD 3800 series) decency the successor of R600, AMD introduced in the fall of 2007 or the Radeon HD 2900 Pro and Radeon HD 2900 GT one over which the stocks of R600 chips should be sold. They were based on the same PCB layout as the Radeon HD 2900 XT and were also equipped with the same cooler design. Due to the placement as a product to sell out of the R600 both graphics cards were very reasonably priced set and only briefly on the market.

Technical

Unified shader

Come As with Nvidia's competition series with the Radeon HD 2000 series the first time unified shaders for use. The evolution of technology has meant that you can really differ in the traditional sense no longer after quads, since there is no rendering pipelines more but composite units still exist. The tasks of the pixel pipelines, vertex and pixel shader from the rendering pipelines now take over with so-called stream processors (SPs ).

These differ in comparison to the competition, however, clear: The five stream processors are combined into a shader unit. From this quintet in turn four to 16 pieces depending on the model form a shader cluster, which thus consists of 20, 40 or 80 stream processors. Within a shader unit all stream processors multiply and add operations ( MAD) can perform, but one also has the ability to perform transcendental functions (eg sine and Logarithmusberechnungen ). These five stream processors can independently execute scalar instructions what AMD called " superscalar ". At the same time they also work in parallel, with the result that no stream processor can calculate an operation that is the result of an operation of a different stream processor in that shader unit dependent. This can have the consequence that individual stream processors idle, unless the driver is no other meaningful computations, and thus the theoretical maximum power of the GPU can not be exploited.

The texture units ( TMUs ) are arranged parallel to the shader clusters and form a cluster - TMU. The number of TMUs corresponds to the size of the shader clusters. This means that the GPU with Shader small clusters ( RV610 ) has 4 TMUs that. Cluster size medium 8 TMUs and finally R600 16 TMUs This allows to select the shader / TMU ratio quite fine or to make textures and shaders performance independently within certain limits in the R600 architecture.

GPUs

Within the Radeon HD 2000 series graphics processors different are used, which differ in terms of 3D capabilities.

Naming

All graphics cards are designated by a four-digit number, which generally begin with " HD 2 ". The second digit then divides the family into different market segments. The third and fourth digits are used to further diversify. Furthermore, these models are categorized into different variants. These are marked with an appropriate symbol behind the model number.

Distribution:

• HD 29xx: performance and high-end
• HD 26xx: Mainstream
• HD 24xx: low-end

Abbreviation:

• GT - budget version of a chip with disabled stream processors
• Pro - standard version of a chip in comparison to the XT low clock rates and partly slower memory
• XT - Powerful version in all segments

Model data

Notes:

• The specified clock rates are fixed or recommended by AMD. However, is the final determination of the clock rates in the hands of the respective graphics cards manufacturers. Therefore, it is quite possible that there are graphics cards models or will be having different clock speeds.
• Mean specifications in parenthesis, that there are common retail versions from different manufacturers with this equipment, but these are not listed as a reference design in public AMD document.
• With the stated date of the appointment of the public presentation is given, not the date of availability of the models.
• The TDP - the manufacturer may differ from the real maximum power consumption.

Performance

For each model, the following theoretical power data:

Notes:

• The above performance data for the computing power of the stream processors, the pixel fill rate, the Texelfüllrate and memory bandwidth are the theoretical maximum values. The overall performance of a graphics card depends, among other things, on how well the available resources can be exploited or utilized. In addition, there are other not listed here, factors that affect the performance.
• The computing power of the stream processors are not directly related to the performance of the Nvidia Geforce 8 series comparable, since this is based on a different architecture, which scales differently.