ATI Radeon HD 5000 series

ATI Radeon HD 5870

A Radeon HD 5450 from Sapphire Technology

The Radeon HD 5000 series is a series of desktop - graphics chip company AMD , which in autumn 2009 as a successor to the Radeon HD 4000 series was introduced. All graphics processors in this series support the Shader model 5.0 according to DirectX 11 and OpenGL 4.4 as well as OpenCL 1.2 in the current version. AMD markets the ability of graphics cards for tasks apart from graphics calculation under the term "ATI Stream". The successor to the Radeon HD 5000 series is the Radeon HD 6000 series .

AMD also uses the code name "Evergreen" for the Radeon HD 5000 series. The mobile graphics solutions are marketed as the ATI Mobility Radeon HD 5000 series .

history

AMD presented the first executable models of the Evergreen series in July 2009 at Computex in Taipei . On September 10, 2009, AMD presented the Radeon HD 5000 series to the trade press. The official presentation of the Radeon HD 5850 and HD 5870 took place on September 23, 2009, although only test copies of the HD 5870 were supplied by the manufacturer. Test copies of the HD 5850 were delivered later, so that they could not be tested until September 30, 2009.

The Cypress core, which was sometimes still referred to as RV870 in the trade press, used the "Terascale 2 architecture" and, with approx. 2.1 billion transistors, was the most complex GPU on the market until then. The Cypress core is manufactured in a 40 nm manufacturing process and thus reaches a die area of ​​334 mm². It is divided into several shader clusters, each cluster consisting of 16 5D shader units and 4 texture units. It takes over the division of the RV770 and RV790 GPUs , but the number of shader clusters increases from 10 to 20. Compared to the previous generation , the number of stream processors on the Cypress core has doubled from 800 to 1600. Regardless of the shader clusters, the number of Raster Operation Processors (ROP) has doubled. However, since a 256-bit memory interface is still used, there are now two ROP clusters per memory controller, which are surrounded by a 128 KB L2 cache.

At its presentation, the Radeon HD 5870 had roughly the same performance as the dual graphics card Radeon HD 4870 X2 . In addition to the performance, AMD received praise in the trade press for the improved energy-saving mechanism after it only worked to a limited extent on the previous generation when using GDDR5 memory . Two shader clusters have been deactivated on the Radeon HD 5850 and the clock rates have also been reduced. Although both cards were already listed for the official presentation on September 23, none was initially available, which is why it was a so-called paper launch. The availability of the cards did not improve at first either, although this was primarily due to technical difficulties with the 40 nm production process at the contract manufacturer TSMC , in addition to high demand, favored by the lack of competing products from Nvidia . Only when TSMC was able to fix this in early 2010, according to its own information, was the Radeon HD 5850 and HD 5870 available across the board.

On October 13, 2009, AMD finally presented the Radeon HD 5750 and HD 5770, which are based on the Juniper chip. This is largely a halved version of the Cypress GPU, which has 1.04 billion transistors and a die size of 166 mm². In contrast to the Radeon HD 5870 and HD 5850, the HD 5770 and HD 5750 were already extensively available on the market for presentation, whereby AMD was again unable to deliver the test samples to the trade press in time for the latter. In terms of performance, AMD replaced the roughly equally fast predecessor models Radeon HD 4850 and HD 4870 with the card .

On November 18, 2009, AMD presented the Radeon HD 5970 dual GPU graphics card . Two Cypress GPUs with a PCIe switch are installed on the card. Graphics cards with two Cypress GPUs are also sold by AMD under the code name "Hemlock". In order to be able to adhere to the official PCIe specifications and to have a power consumption of less than 300 watts, AMD reduced the clock rates to the level of the Radeon HD 5850. There was no need to deactivate shader clusters. When the Radeon HD 5970 was presented, it had a significantly higher performance than the fastest graphics card on the market until then, the Geforce GTX 295 . Contrary to some expectations, AMD had not made any improvements in the area of ​​micro-stuttering. A special feature of the Radeon HD 5970 is its length of 30.5 cm, which means that it cannot be installed in all midi towers.

On January 14, 2010, AMD presented the Radeon HD 5670, with which the Radeon HD 5000 series was expanded to the mainstream sector. The Radeon HD 5670 is based on the Redwood graphics processor, which in the broadest sense is a halved Juniper chip. Since the number of raster units has also been halved, but a 128-bit memory interface is still used, there is now only one ROP partition per memory controller. In terms of performance, AMD placed the Radeon HD 5670 between the Nvidia competitors Geforce GT 240 and Geforce 9800 GT .

On February 4, 2010, AMD released the Radeon HD 5450, the first low-end graphics card in the HD 5000 series. Based on the AMD reference design, the HD 5450 is a passively cooled card in the ultra low profile format. B. suitable for home theater PCs . The graphics card uses the Cedar graphics processor. In this case, the structure of the shader clusters is slightly different, only 8 shader units are coupled with 4 texture units, so that Cedar consists of 2 clusters with a total of 16 shader units, but 8 texture units. The HD 5570 followed on February 9, 2010. It is based on the same chip as the HD 5670, but has a lower clock rate and slower memory. The Radeon HD 5550 presented on the same day is based on a partially deactivated Redwood GPU, but had no market relevance due to the limited support from board partners. A little later, on February 25, the Radeon HD 5830 completed the product family for the time being. The model is priced between the Radeon HD 5770 and HD 5850. The graphics card uses the larger RV870 chipset, which has been severely restricted so that only 1120 stream processors, 56 TMUs and 16 ROPs of the Radeon HD 5830 are available. In return, the clock rates have been increased slightly compared to the Radeon HD 5850, but this results in a slightly higher power consumption.

technology

architecture

For the first time, AMD used its "Terascale 2 architecture" for the Radeon HD 5000 series, which is a further development of the unified shader architecture of the R600 GPU . Due to the design for DirectX 11, various changes had to be made, including: a. the support of the shader model 5.0. This includes the use of the new Compute, Hull and Domain shaders, made it necessary to improve texture compression (e.g. 6: 1 compression for 16-bit HDR textures) and requires the texture units to have the maximum texture resolution of 16k × 16k (previously 8k × 8k). The compute shader is used to support Direct Compute, i.e. the calculation of non-graphic data on the GPU (GPU computing). Direct Compute is in competition with OpenCL . The architecture has also been optimized for GPU computing and supports the IEEE 754-2008 standard , which is why the GPU can perform Fused Multiply Add (FMA). In addition to optimizing the caches, other commands such as Sum of Absolute Differences (SAD) as well as commands important for Direct Compute such as Bit count, insert, extract were implemented. Since the R600 graphics processor, ATI has been using so-called Tessellator units, which are generally necessary with DirectX 11 support. With them it is possible to increase the number of polygons in an object with relatively little computing effort . The link with the Hull and Domain shaders is new, while on the previous generation the tessellator largely only acted with the geometry shader. Regardless of the DirectX 11 implementation, AMD presents the option of controlling more than two screens natively under the name “ATI Eyefinity”.

The "Terascale 2 architecture" uses a new memory management system that now masters EDC ( Error Detecting Code ), whereby the errors are recognized by cyclic redundancy checks (CRC). This is primarily relevant in the area of ​​GPU computing tasks, which AMD treats under the term “ATI Stream”. With so-called "link retraining" it is also possible to change the clock speed and voltage of the memory quickly and reliably, which is why the GDDR5 memory, in contrast to the previous series, is automatically downclocked even when idling.

Graphics processors

Graphics chip	archi- tecture	production			units						API support			Video pro- cessor	Bus interface stelle
		production process	transis- interfere	The - area	ROPs	Unified shaders			Texture units		DirectX	OpenGL	OpenCL
						Stream processors	Shader units	Shader cluster	TAUs	TMUs
Cedar (RV810)	Terascale 2	40 nm	0292 million	063 mm²	04th	0080	016 × 5D- VLIW	02	08th	08th	11.0	4.4 (Linux Mesa 18+: 4.4+, Mesa 19+, 4.5+)	1.2 (Linux Mesa: 1.1 almost complete, 1.2 in progress)	UVD 2.2	PCIe 2.0
Redwood (RV830)			0627 million	104 mm²	08th	0400	080 × 5D-VLIW	05	20th	20th
Juniper (RV840)			1040 million	166 mm²	16	0800	160 × 5D-VLIW	10	40	40
Cypress (RV870)			2,154 million	334 mm²	32	1600	320 × 5D-VLIW	20th	80	80

Naming

The Evergreen series uses the same labeling system as has been used since the Radeon HD 3000 series . All graphics cards are labeled with "ATI Radeon HD" and an additional four-digit number that generally begins with a "5" (for the series). The second and third digits are used to subdivide into different models.

division

• HD 5450 and 5550: low-end
• HD 5570 and 5670: mainstream
• HD 57xx and 5830: Performance
• HD 5850/70 and 5970: high-end

Due to the general drop in prices on the market and currency fluctuations, ATI's original classifications do not generally apply.

Model data

A Radeon HD 5870 from Sapphire Technology

model	Official launch	Graphics processor (GPU)						Graphics memory
		Type	Active units				Clock rate (MHz)	Size ( MB )	Clock rate (MHz)	Type	Storage interface
			ROPs	Shader - cluster	Stream processors	Texture units
Radeon HD 5450	Feb. 4, 2010	Cedar	04th	02	0080	08th	650	0512	0500	DDR2	064 bit
									0800	DDR3
Radeon HD 5550	Feb 9, 2010	redwood	08th	04th	0320	16	550	0512	0400	DDR2	128 bit
									0667 ... 800 0800 (1600)	DDR3 GDDR5
Radeon HD 5570	Feb 9, 2010	redwood	08th	05	0400	20th	650	0512	0900 0900 (1800)	DDR3 GDDR5	128 bit
Radeon HD 5670	Jan. 14, 2010	redwood	08th	05	0400	20th	775	0512	2000 (1000)	GDDR5	128 bit
Radeon HD 5750	Oct 13, 2009	Juniper	16	09	0720	36	700	0512 1024	2300 (1150)	GDDR5	128 bit
Radeon HD 5770	Oct 13, 2009	Juniper	16	10	0800	40	850	1024	2400 (1200)	GDDR5	128 bit
Radeon HD 5830	Feb 25, 2010	Cypress	16	14th	1120	56	800	1024	2000 (1000)	GDDR5	256 bit
Radeon HD 5850	23 Sep 2009	Cypress	32	18th	1440	72	725	1024	2000 (1000)	GDDR5	256 bit
Radeon HD 5870	23 Sep 2009	Cypress	32	20th	1600	80	850	1024 2048	2400 (1200)	GDDR5	256 bit
Radeon HD 5970	Nov 18, 2009	2 × Cypress (Hemlock)	2 × 32	2 × 20	2 × 1600	2 × 80	725	2 × 1024	2000 (1000)	GDDR5	2 × 256 bits

Performance data

The following theoretical performance data result for the respective models:

model	Total computing power of the stream processors in GFlops		Graphics processor (GPU)			Graphics memory
			Polygon - throughput in million triangles / s	Fill rate		Data transfer rate in GB / s
	SP (MAD)	DP (FMA)		Pixels in GP / s	Texel in GT / s
Radeon HD 5450	0104	k. A.	650	02.6	05.2	008.0
						012.8
Radeon HD 5550	0352		550	04.4	08.8	012.8
						021.3 ... 25.6
Radeon HD 5570	0520		650	05.2	13.0	028.8
Radeon HD 5670	0620		775	06.2	15.5	064
Radeon HD 5750	1008		700	11.2	25.2	073.6
Radeon HD 5770	1360		850	13.6	34.0	076.8
Radeon HD 5830	1792	358.4	800	12.8	44.8	128
Radeon HD 5850	2088	417.6	725	23.2	52.2	128
Radeon HD 5870	2720	544	850	27.2	68.0	153.6
Radeon HD 5970	2 × 2320	2 × 464	2 × 725	2 × 23.2	2 × 58.0	2 × 128

Remarks

• The specified performance values ​​for the computing power via the stream processors, the pixel fill rate, the texel fill rate and the memory bandwidth are theoretical maximum values. The overall performance of a graphics card depends, among other things, on how well the available resources can be used or fully utilized. There are also other factors that are not listed here that affect performance.
• The computing power via the stream processors is not directly comparable with the performance of the Nvidia Geforce series, as it is based on a different architecture that scales differently.

Power consumption data

The measured values ​​listed in the table relate to the pure power consumption of graphics cards that correspond to the ATI reference design. A special measuring device is required to measure these values; Depending on the measurement technology used and the given measurement conditions, including the program used to generate the 3D load, the values ​​can fluctuate between different devices. Therefore, measured value ranges are given here, each representing the lowest and highest measured values ​​from different sources.

model	Type	Consumption ( watt )			additional power plug
		TDP	Readings
			Idle	3D load
Radeon HD 5450	Cedar	019th	07 ... 08	015 ... 016	no
Radeon HD 5550	redwood	040
Radeon HD 5570	redwood	043	09 ... 12	037 ... 042
Radeon HD 5670	redwood	061	12 ... 14	060 ... 062
Radeon HD 5750	Juniper	086	15 ... 17	077 ... 079	1 × 6 pin
Radeon HD 5770	Juniper	108	18 ... 22	102 ... 106
Radeon HD 5830	Cypress	175	20 ... 25	120 ... 159	2 × 6-pin
Radeon HD 5850	Cypress	170	18 ... 19	149 ... 150
Radeon HD 5870	Cypress	188	19 ... 21	186 ... 214
Radeon HD 5970	2 × Cypress (Hemlock)	294	43 ... 45	291 ... 337	1 × 6-pin 1 × 8-pin

Much more common than measuring the consumption of the graphics card is determining the power consumption of an entire system. For this purpose, a reference system is compiled in which the various graphics cards are installed; Then the measurement takes place directly at the socket with the help of an energy cost meter or a comparable device . However, the meaningfulness of the measured values ​​is limited: It is not clear what consumption comes from the graphics card and what can be ascribed to the rest of the PC system. With this measurement method, the difference in consumption between idle and 3D load operation does not only depend on the program with which the load was generated; the utilization and efficiency of the rest of the PC system including the power supply unit, mainboard and processor also influence the measured difference. Since the tested systems usually differ from your own PC system at home, the values ​​given there cannot be mapped to your own system. Only measurement data from otherwise identical systems are (to a limited extent) suitable for comparison with one another. Because of this dependency, total system measured values ​​are not listed in the table here. However, since they can give a better picture of the practical power consumption of a specific system with a specific graphics card, websites that made such measurements are listed under the web links .

Web links

Commons : ATI-Radeon-HD-5000-Series  - collection of pictures, videos and audio files

• Official product page of the Radeon HD 5000 series at AMD

Power consumption measurements

• Measurements of the entire system at ComputerBase

Individual evidence

1. ↑ AMD: ATi Stream Technology ( Memento of the original from August 27, 2009 in the Internet Archive ) Info: The archive link was automatically inserted and not yet checked. Please check the original and archive link according to the instructions and then remove this notice. , accessed January 31, 2010 
2. ↑ ComputerBase: AMD shows working DirectX 11 graphics card (update) , message from June 3, 2009, accessed on January 31, 2010
3. ↑ ComputerBase: ATi Radeon HD 5870 - Technology in Detail Part 1 , test report from September 23, 2009, accessed on January 31, 2010
4. ↑ ComputerBase: ATi Radeon HD 5870 - performance rating , test report from September 23, 2009, accessed on January 31, 2010
5. ↑ ComputerBase: Test: ATi Radeon HD 5870 - Conclusion , test report from September 23, 2009, accessed on January 31, 2010
6. ↑ Hardwareluxx : Page 27: ATI Radeon HD 5870 - Conclusion , test report from September 23, 2009, accessed on January 31, 2010
7. ↑ ComputerBase: ATis Radeon HD 5850 & HD 5870 listed , message from September 23, 2009, accessed on January 31, 2010
8. ↑ ComputerBase: ATi's Cypress-GPU does not want to become available , message of November 25, 2009, accessed on January 31, 2010
9. ↑ Hardwareluxx: TSMC: Manufacturing problems resolved , message from January 23, 2009, accessed on January 31, 2010
10. ↑ Radeon3D.org Technical data of the Radeon HD 5770 ( Memento of the original from January 15, 2010 in the Internet Archive ) Info: The archive link was inserted automatically and has not yet been checked. Please check the original and archive link according to the instructions and then remove this notice. , accessed January 31, 2010 
11. ↑ ComputerBase: Test: ATI Radeon HD 5750 - performance rating , test report from October 24, 2009, accessed on January 31, 2010
12. ↑ ComputerBase: Test: ATi Radeon HD 5970 - performance rating , test report from November 18, 2009, accessed on January 31, 2010
13. ↑ ^{a b c} PC Games Hardware: Test Radeon HD 5970 (Hemlock): The fastest DirectX 11 graphics card in the world , test report from November 18, 2009, accessed on January 31, 2010
14. ↑ ComputerBase: Test: ATi Radeon HD 5670 - performance rating , test report from January 14, 2010, accessed on January 31, 2010
15. ↑ Hardtecs4u: AMD Radeon HD 5570 - smallest mainstream representative with HTPC ambitions? , Test report from February 9, 2010, accessed on March 1, 2010
16. ↑ TechPowerUp: Club 3D Unveils Radeon HD 5550 and HD 5570 Accelerators , message dated February 10, 2010, accessed March 16, 2010
17. ↑ Radeon3D.org: Unveiled Radeon HD 5830 will be released the day after tomorrow for the retail market ( Memento of the original from February 25, 2010 in the Internet Archive ) Info: The archive link was automatically inserted and not yet checked. Please check the original and archive link according to the instructions and then remove this notice. , News dated February 23, 2010, accessed March 1, 2010 
18. ↑ ^{a b} Hardwareluxx: ATI Radeon HD 5870 - RV870 architecture , test report from September 23, 2009, accessed on January 31, 2010
19. ↑ ComputerBase: ATi Radeon HD 5870 - Technology in Detail Part 2 , test report from September 23, 2009, accessed on January 31, 2010
20. ↑ http://amd-dev.wpengine.netdna-cdn.com/wordpress/media/2012/10/Heterogeneous_Computing_OpenCL_and_the_ATI_Radeon_HD_5870_Architecture_201003.pdf
21. ↑ ^{a b c d e f} PC Games Hardware: Test Radeon HD 5570 (Redwood LE): All-rounder with DirectX 11 , test report from February 9, 2010, accessed on February 11, 2010
22. ↑ ^{a b c d e f g h i j k l m} HT4U.net: AMD Radeon HD 5570 - smallest mainstream representative with HTPC ambitions? , Test report dated February 9, 2010, accessed on February 11, 2010
23. ↑ ^{a b c d e f g} PC Games Hardware: Test Radeon HD 5750: Juniper GPU for the second time , test report from October 20, 2009, accessed on January 31, 2010
24. ↑ ^{a b c} HardTecs4U AMD ATI Radeon HD 5830 - The last in the group
25. ↑ AMD develops medium-Graphics ATI Radeon HD 5830  ( Page no longer available , search in web archives )  Info: The link was automatically marked as defective. Please check the link according to the instructions and then remove this notice.  
26. ↑ ^{a b c d} TechPowerUp: HIS Radeon HD 5850 1 GB - Power Consumption , test report from October 2, 2009, accessed on February 11, 2010