Radeon R700

The Radeon HD 4000 series is a series of desktop graphics chip company AMD and the successor to the Radeon HD 3000 series. All graphics processors of this series support Shader Model 4.1 (SM 4.1 ) to DirectX 10.1. In this series of GDDR5 memory is used for the first time. The successor to the Radeon HD 5000 series.

• 5.1 Measurement of power consumption of a whole system

Description

The official presentation of the graphics chip used on the graphics card HD 4850 and HD 4870 RV770 was held on 25 June 2008. Following its early commercial availability and thereby early reports published by AMD buyers preferred the idea of the HD 4850 on 19 June, under the condition that only performance comparisons are made, but no details of the architecture will be announced.

The RV770 is based on the unified shader architecture of the R600 chips of the Radeon HD 2900 XT, which was, however, significantly modified. The most important difference is the relocation of the texture units ( TMUs ) in the shader clusters. In the R600 generation (including the RV6xx offshoots ), the texture units formed a cluster whose size is identical to the shader cluster size. This allowed, in principle, a slightly better utilization of TMUs at R600, but the design does not scale to larger TMU numbers without loss of efficiency due to large shader cluster. With the R700 generation there are always 4 TMUs in a shader clusters, each of which also has its own first-level cache is available (up to RV670 all TMUs have a single L1 cache divided ), the texture of L2 cache remains unified. Just this change enabled the significant increase in unit numbers with the RV770.

In contrast to the direct predecessor, the RV670, the number of 5D shader units was increased from 64 to 160. Although the number of texture units on a half times increased ( from 16 to 40, wherein the TAUs are now installed to TMUs in a 1:1 ratio ), the number of transistors increased only from 666 to 956 million transistors. In constant 55 - nm manufacturing process, the die area of 193 mm ² enlarged to 256 mm ². Since the competition chip from Nvidia, which was GT200 with 576 mm ² significantly larger (due to the significantly higher number of transistors as well as the older 65nm manufacturing process ), ATI RV770 could make the cost-effective. As a result, ATI Radeon HD 4870 was able to offer much cheaper on the market than the first about equal fast Geforce GTX 260 from Nvidia. To be very successful, the Radeon HD 4850 proved in the market area of ​​performance class because it is significantly more powerful than the almost equally expensive GeForce 9800 GT. ATI applied their new graphics card generation with the power saving function " PowerPlay 2.0". These came partly widespread criticism because they worked much less efficient than the previous version on the Radeon HD 3000 series. Especially in connection with the GDDR5 memory on the Radeon HD 4870 power consumption has been criticized at idle too high, which could not be resolved with driver updates.

With two RV770 chips that work together using CrossFire technology, the dual-GPU card HD 4870 X2 and HD 4850 X2 were presented in August 2008. While the latter map is exclusively distributed by Sapphire, the HD 4870 X2 is found in almost all board partners in the portfolio. In 2D mode, the power saving function " PowerPlay " helps to keep the power consumption within limits, under load, however, breaks the power consumption of the HD 4870 X2 all previous negative records, and therefore also a strong power supply is a prerequisite for this card. With the Radeon HD 4870 X2 ATI introduced the first time since the Radeon X1000 series again the most powerful graphics card on the market available to Nvidia in January 2009, the Geforce GTX 295 introduced.

On September 10, 2008, AMD introduced the Radeon HD 4600 series for the mainstream market, and on September 30, the Radeon HD 4500 and the Radeon HD 4300er series for the low- end market. The graphics cards based on the RV730 or RV710 GPUs, are characterized by a low power loss especially at idle. Since the Radeon HD 4670 quite successfully acted against Nvidia's GeForce 9600 GSO, but could not compete with the GeForce 9600 GT, ATI presented on 23 October 2008, the Radeon HD 4830th It is a Radeon HD 4850 with reduced clock rates and two disabled shader clusters.

On 2 April 2009, AMD introduced the Radeon HD 4890, which is based on the RV790. It is a RV770 derivative which has been prepared in a 55 nm improved manufacturing process to achieve higher clock speeds. This allows to reach the Radeon HD 4890 in about the performance of the Geforce GTX 280 is no longer produced at this time. Nvidia responded with the presentation of approximately the same fast Geforce GTX 275, which architecture requires cost-intensive to manufacture, however. On a dual- GPU graphics card based on the RV790 renounced AMD, since such a map would have a TDP of 300 watts (which would be beyond the permissible according to the PCI Express standard power ) and after AMD's assessment also " simply not necessary " be. However, shortly afterwards announced some graphics card manufacturers already such a card, it should - not standard - two 8- pin power connectors are used. Furthermore, AMD stated that they had reduced the power consumption of the Radeon HD 4890 compared to 4870 at idle. However, this was not confirmed by the first independent test measurements.

On 28 April 2009, AMD introduced the Radeon HD 4770, which is based on the RV740 and the Radeon HD 4830 to replace. When RV740 chip is the first manufactured in 40 - nm manufacturing process GPU. This enabled AMD to offer the card with a good price-performance ratio on the market and to successfully place against about the same expensive, but slower GeForce 9800 GT from Nvidia. The new manufacturing process progress in power consumption has been achieved, which is idle in Windows, however, were not as great as hoped, which is attributed to the use of GDDR5 memory. After the launch of AMD was the unexpectedly high demand for the Radeon HD 4770 initially not cover because the new 40nm manufacturing process had insufficient yield rate at TSMC. That's why AMD dropped his board partners PowerColor 4730 announce on May 28, 2009, the Radeon HD for June 8th. It is an overclocked version of the Radeon HD 4830 with only half the memory interface, but which uses GDDR5 memory. Therefore, the theoretical memory bandwidth remains unchanged. First test reports show a surprisingly low 3D performance, which suggests that the number of ROPs at the Radeon HD 4730 was halved to 8.

In August, presented board partners from AMD, the Radeon HD 4750 and 4860 for the Asian market before. Both cards come along with disabled Shader clusters, the Radeon HD 4750 produced in 40nm RV740 the graphics processor used, whereas the HD 4860 is based on a teildeaktivierten and to date only on the Radeon HD 4890 RV790 used. The clock rates are, however, as suggested in the model name, between which the models HD 4850 and HD 4870, significantly lower than those of the HD 4890th

GPUs

Naming

All graphics cards are referred to as " ATI Radeon HD" and an additional four-digit number that generally begins with a "4" ( for the series ). The second digit divides the family into different market segments. The third and fourth digit is a subdivision in the various models. Suffixes are not used as in the previous series, the performance of classification within the series or the market segment is only about the number.

Distribution:

• HD 42x0: IGP
• HD 43x0: low-end
• HD 45x0/46x0: Mainstream
• HD 47x0/48x0: Performance
• HD 48x0 X2: High-End ( models with two GPUs on one card)

Model data

Notes:

• The specified clock rates are recommended by AMD or specified, the memory clock of the I / O clock is specified. However, may differ by several clock a few megahertz of the exact timing, further 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.
• With the stated date of the appointment of the public presentation is given, not the date of availability of the models.
• It may be possible with the graphics chips, which are equipped with on-board graphics memory Side port, " Port Side and UMA " be switched to dynamically increase the graphics memory, or even to improve the graphics performance low.

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 is not directly comparable with the performance of Nvidia Geforce series, since this is based on a different architecture, which scales differently.

Power consumption data

The measured values ​​listed in the table refer to the pure power of graphics cards that comply with the ATI reference design. In order to measure these values ​​, it requires a special measuring apparatus; depending on the employed measurement technique and given measurement conditions, including the shared program, with which the 3D load is generated, the values ​​between different equipment may vary. Therefore, here measured value ranges are specified, each representing the lowest and highest measured values ​​from different sources.

Far more common than the measurement of the consumption of the graphics card is to determine the power consumption of a whole system. To this end, a reference system is assembled, in which the various graphics cards to be installed; then finds the measurement using an energy cost meter or similar apparatus rather than directly to the outlet. However, the significance of the measurements is limited: it is not clear which consumption comes from the graphics card and what is attributable to the rest of the PC system. The difference in consumption between idle and load 3D operation of this measuring method is not only dependent on the program with which the load was produced; the utilization and efficiency of the rest of the PC system, including the power supply, motherboard and processor affect the measured difference also. Because different the tested systems are usually of one's own home PC system, the values ​​given there can not be mapped to its own system. Only data from otherwise identical systems are good (conditional) for comparison with each other. Because of this dependence overall system measurement values ​​are not listed in the local table. But since they can provide a better picture of the practical power of a concrete system with a certain graphics card, web pages are listed under the links that are undertaking such measurements.