Video cards. Video cards Appearance and dimensions

I installed ThermalRight Shaman and a heat spreader plate from the reference on the Radeon R9 290X and then lowered the operating voltage from 1.212 V (according to some sources from 1.250 V and I’m starting to believe in this value) to 1.085 V, winning back the PC consumption, in this case only from the video card, a decent 60 W, that is, if according to average indicators it consumed 250 W, which are given in the reviews, then it became 190 W. This will not only save a small penny, but most importantly, it will reduce the load on the R9 290X cooling system and the heating of components in the system.

First about the cons

What about overclocking?

Preparation

• i5-4670K processor overclocked to 4000 MHz
• PowerColor Radeon R9 290X video card with a modified cooling system
• Wattmeter Robiton PM-2
• Fluke 59 MAX IR Thermometer
The results were also carried out with two BIOS versions - reference, 015. 042 .000.000.000000 and the version from Split 390X For 290X (the last one generally available to the living and all mortals, number 015. 049 .000.000.000000). I've never seen anything fresh.

Results

1)

2)

3)

4)

For a snack

Power consumption of my PC in several modes

MSI Kombustor

The Witcher 3

Instead of a conclusion

It's just strange to me - why? reference voltage exposed already! equal 1.250 V!!! I remember the Radeon HD 6950 and 6970 models had 1.100 and 1.175 V and they felt good. If they were also set to 1.250 V, we would have received all the “horror stories” about heating and CO noise back in 2011. Unfortunately, I now do not have the opportunity to check how the reference cooling system will cope at 1.085 V, because I “teared it apart”, but I think I will see a much lower temperature and/or noise level there.
P.S.S. If anyone cares, I’ll accept/buy one of these as a gift and conduct another series of tests for you, comrades)))
At these frequency values, and, unfortunately, at this voltage, the thermal package for the 290X fits into 230-260 W, although a second question arises: in the BIOS, the thermal package is programmed as 208 W or 230 W for some copies. Therefore, the third question: how did some authors have an average power consumption of as much as 280-335 W??? They increased PowerLimit, but did not write about it?

What do I want to say? I personally liked the architecture of video cards codenamed Hawaii, it’s a pity that not everything is fully optimized, and the average consumer is not interested or does not know how or is afraid to edit something himself and is worried about the warranty, but in my opinion the cards were a success. The only thing that distinguishes them from the “Greens” is acceleration. It is smaller, leaner, and consumption and heat generation begin to grow not in direct proportion. Therefore, of the two ways - optimization or overclocking, I chose optimization) And if overclocking, then without increasing the voltage) Fortunately, overclocking the video memory from 1250 to 1500 MHz increased the consumption by only 8-11 W, and the chip from 1000 to 1120 MHz - by 12 Tue
If you are interested in new tests, tests or experiments - write, I will try to carry out all the necessary measurements.

Once again, I apologize for the fact that my thoughts were running wild, that I answered 1-2 questions in the material, but then immediately asked 2-3 new ones. I’m just very interested in figuring this out myself and I want to get answers to all of them as quickly as possible!

[Added]

The Radeon R9 290 video card is the youngest flagship of the new generation AMD of the ninth series. This is a mid-price gaming video camera that is perfectly balanced in terms of technical parameters and can handle most modern video games at maximum, ultra-high settings. Was released in 2013. The average cost in Europe is $280-290.

Overview, technical specifications

Radeon R9 290 is a gaming video card released on the new Hawaii processor. The crystal is manufactured using a 28 nm technical process. Contains 6.2 billion transistors, and its area is 438 mm². Unlike the AMD 290X, which has a bar frequency of 1000 MHz, this model has only 945 MHz. The memory capacity is 4096 megabytes. Connected to a 512-bit bus. It operates at a frequency of 1.250 MHz, which provides a throughput of 320 GB/s. Has support for all new features: TrueAudio, DirectSound 11.2 Tier 2.

In terms of technical parameters, the AMD 290 can be compared with the Nvidia 780 graphics adapter, although in some parameters it is inferior to its competitor. In addition, this model is not much different from the AMD 290X, although compared to the latter the Boost frequency was reduced and the number of stream processors was reduced from 2,816 to 2,560. The number of tactural units is 160, while the number of raster operation pipelines (ROPs) is also 64.

The Radeon R9 290, like the 290X, received a VID interface, thanks to which you can quickly and accurately adjust the voltage on the GPU, as well as a telemetry channel through which PowerTune receives voltage data from analog sensors.

The R9 290 PCB is the same as the R9 290X. The frame buffer is composed of SK Hynix H5GQ1H24AFR-R0C chips with a nominal clock frequency of 6 GHz. The components are powered according to a 5+1+1 scheme (GPU, memory chips, PLL).

The video adapter specifications show the maximum GPU frequency, but in practice it can be limited by power consumption limits and core temperature. Another variable is the fan speed limit: if the GPU temperature cannot be kept within normal limits at a given cooler speed, the frequency will decrease. Therefore, if you manually configure PowerTune parameters, you can give priority to either the performance or acoustic parameters of the video card. give priority to either productivity.

Temperature, power consumption

Performance

Radeon R9 290 is produced by the following manufacturers:

1. SAPPHIRE
2. GIGABYTE

Below is an overview of video cards from each manufacturer.

Testing Radeon R9 290

It was initially planned that the characteristics of the R9 290 would be similar to the GeForce GTX 770, but as a result, the video adapter received an upgrade in status to such a rival as the rival GTX 780, as evidenced by numerous tests.

Before the release of the Radeon R9 290, AMD decided to boost performance by changing the fan speed limit from 40% to 47%. The initial limit of 40% corresponds to the so-called Quiet Mode of the Radeon R9 290X, but actual rotation speeds vary: 1987 rpm for the R9 290X and 2172 rpm for the R9 290.

Under increased loads, the core temperature approaches the permissible 95 °C, which is considered normal for the Hawaii processor. But at the same time, the fan can no longer spin faster, which leads to active frequency throttling. During a test run in the game Crysis 3, a drop to 851 MHz was noticed from the original 947 MHz.

Considering that the frequencies of the AMD 290 and 290X differ by only 5%, the power indicators of both video cards are almost the same.

Performance, gaming tests

3DMark 2011, focused on DirectX 11, did not support AMD's ambitious plans: the Radeon R9 290X was ahead of the GeForce GTX 780, but the R9 290 was behind in some respects. A newer version of 3DMark also found a noticeable difference between the R9 290 and R9 290X. At the same time, the AMD 290 is significantly superior to the GeForce GTX 780 and almost reaches the GTX TITAN.

The victory of the Radeon R9 290 over the GTX 780 came at the cost of operating the cooling system at higher speeds. But if you limit the fan speed to 40% of the nominal value, according to the benchmark results, overclocking the cooler did not have a significant effect, and only in some tests leads to a slight increase in the frame rate. However, this was quite enough to confirm its unconditional superiority over the GeForce GTX 780.

Overclocking the Radeon R9 290 by the user to the same frequencies as the R9 290X has a very noticeable effect on performance both in synthetic benchmarks and in real modern video games, including in 4K resolution.

The Radeon R9 290's peak processing power (4.9 TFLOPS) is 87.5% of the R9 290X's (5.6 TFLOPS). The GeForce GTX 780 compares roughly with the GeForce GTX TITAN (89%), and the difference between them is in video games such as: Battlefield 3, Far Cry 3, Tomb Raider, Crysis 3, Metro: Last Light, Unigine Heaven 2, Batman: Arkham City is practically inconsequential. Users can safely run games at maximum settings.

If we talk about the overall performance of the tested model in games, then it will be enough to get acquainted with almost all new products both in multi-monitor configuration mode with a total resolution of 5760 x 1080, and on displays with a resolution of 2560 x 1600.

SAPPHIRE R9 290 4GB

SAPPHIRE R9 290 4GB graphics accelerator is a model with a “reference” cooling system. Occupies two slots for expansion cards. The video card has a reference frequency circuit without the use of any factory overclocking, which means that all overclocking is carried out by users.

The SAPPHIRE R9 290 4GB GDDR5 video adapter is equipped with an AMD Hawaii PRO graphics chip, which is produced using a 28 nm process technology. Includes 2560 universal shader processors, 160 texture units, 64 rasterization units. Depending on the load and heating, the clock frequency of the graphics core dynamically changes, reaching a maximum of 947 MHz.

Set of video card interface panel ports:

1. 2 x Dual-link DVI-D;
2. 1 x HDMI;
3. 1 x DisplayPort.

HDMI and DisplayPort connectors support image output in 4K Ultra HD resolution (4096×2160). In AMD Eyefinity mode, you can connect four monitors and TVs using the DisplayPort interface and special adapters, using all connectors of the interface panel directly.

The height of the printed circuit board is 112 mm. The length of the video card is 290 mm. In the center there is a graphics chip equipped with a metal frame for safe installation and removal of the cooling system. The video card's power subsystem is built according to a seven-phase circuit: five phases for the graphics core, one for memory and one for PLL. To power the graphics adapter, in addition to the PCI Express 3.0 x16 slot, two additional 6- and 8-pin PCIe connectors are used.

In terms of performance, the SAPPHIRE R9 290 is about 6% ahead of the NVIDIA GTX 770, but 9-12% behind the NVIDIA GTX 780. At the same time, gaming tests have shown that most new video games are stable on displays with a resolution of 5760 x 1080 and 2560 x 1600 run on maximum settings. When overclocking, we managed to increase the GPU frequency to 1190 MHz (25%;) which can be called a very good result. The video memory frequency was fixed at 1400 MHz (+12%).

The main advantage of this model is its high level of performance in computer games, which allows you to assemble a completely balanced gaming system that will satisfy both professional gamers and beginners. The downside is the high noise level of CO.

ASUS RADEON R9 290 OC

ASUS Radeon R9 290 OC (R9290-DC2OC-4GD5) is a perfectly balanced gaming video card that is suitable for comfortable gaming at maximum, ultramaximum settings.

The model is equipped with an open-type DirectCU II cooling system with two Everflow fans, a massive aluminum radiator with five heat pipes. Direct contact technology is used, with three out of five tubes used in the heat sink.

The main advantages of ASUS Radeon R9 290 OC include:

• high degree of productivity;
• fresh GPU architecture;
• relatively low cost;
• high-quality food system.

Disadvantages: modest equipment, average CO efficiency, increased noise at maximum load.

The length of the graphics accelerator is 287 mm. Two DVI-D connectors – HDMI, DisplayPort – are responsible for image output.

The attached elements of the printed circuit board are arranged as tightly as possible, which is explained by the large dimensions of the graphics processor, as well as a large number of different microcircuits, and a complex power system. The BIOS switch is located on the top edge of the PCB. Performance or quiet modes available.

The AMD Hawaii processor has a very large core surface area and is equipped with a protective metal frame. 2560 universal stream processors, which are located in the core housing, provide maximum video performance. The frequency of the graphics chip of the video card is increased compared to the reference and is 1000 MHz.

When the video card was loaded at 100% using the MSI Combuster utility, the AMD Hawaii video chip warmed up to 81 degrees Celsius. The level of heating under load shows that the installed DirectCU II does not look as confident on AMD Hawaii as the cooling systems in similar video cards from other manufacturers. During overclocking, the graphics chip frequency increased to 1150MHz, the memory frequency to 6000MHz.

In gaming tests, the model showed excellent results. The ASUS Radeon R9 290 OC has enough power for comfortable gaming in such heavy projects as Metro Last Light, Hitman Absolution.

GIGABYTE Radeon R9 290

When planning to build a productive gaming system, users can give preference to the GIGABYTE Radeon R9 290 graphics accelerator, which attracts with its good price-quality ratio. Thanks to the high degree of performance, you can run your favorite modern video games at maximum, ultra-high settings without dropping FPS.

The GIGABYTE Radeon R9 290 video adapter, manufactured on the revolutionary GCN architecture, is equipped with an excellent, efficient Windforce 3X cooling system, which keeps the device temperature low even at maximum graphics settings and at the same time ensures almost silent operation. AMD TrueAudio technology allows you to get the highest quality, clearest, surround sound.

The video card model is quite massive and weighty, which is explained by the size of the card and the casing made of metal. The heat pipes and cooling system casing protrude slightly beyond the printed circuit board. At the same time, thanks to the well-thought-out design of the casing, the weight is distributed more evenly. The cooling system consists of three proprietary 75mm fans and a massive radiator, consisting of two sections connected by five heat pipes with a diameter of 5 and 8 mm.

Video memory capacity – 4 GB. There is support for SLI/CrossFire, CrossFire X, DirectX 12, OpenGL 4.5 modes. Interfaces for connecting to monitors are represented by two DVI connectors, one HDMI and Display Port.

In benchmarks on the Intel i7-3930K, 4 GHz processor, Extreme profiles were used. In Crysis 3, Metro Last Light Redux, Sleeping Dogs, Tomb Raider 2013, and other modern games at maximum settings, FPS confidently remains at 70-80. The temperature under load increased, but the noise level was reduced.

Bottom line

As you can see, the difference with the 290X is insignificant. Even in its “lightweight” form, GPU Hawaii does not lose its ambitions for leadership positions in 4K resolutions. AMD does not disclose the TDP of the card, but if the Radeon R9 290X under load consumes no less than 280 W, then in this regard the Radeon R9 290 is not far behind.

In the midst of price wars in the top echelon of video cards, the release of a new product under the name AMD Radeon R9 290 is especially interesting. From below along the red line it is supported by the Radeon R9 280X with a recommended price tag of $299, and from above it is limited by its older brother R9 290X for $549. The competitor is also not asleep and has already expressed his opinion on prices: GeForce GTX 770 – 11,990 rubles, which is equivalent to $349, GTX 780 – 17,990 rubles or $499. The latter figure, in my opinion, is too close to the stated threshold of $549, so it is more logical to assume that the Radeon R9 290 will be priced at $449.

Yes, friends, a day before the announcement we still did not receive the recommended price of the new product, which means we will rely on its other properties, more precisely on its performance compared to the GeForce GTX 780, because it is the latter that will constitute the main competition.

Technical features

From the technical side of the issue, the Radeon R9 290 is nothing more than a simplified version of the full-fledged Hawaii core. There are slightly fewer stream processors and texture units, and the GPU frequency has decreased, which is a significant difference from the R9 290X.

As you already know, Hawaii's frequency formulation may differ from the declared value under load, especially in quiet mode. For the R9 290, the basic principle of GPU operation is exactly taken from the silent mode of the Radeon R9 290X. It served as a prototype, thanks to which the concept “up to 947 MHz” should be taken literally. Under prolonged load, due to the desire to fit the video card within a certain power consumption limit, it reduces the frequency and sometimes quite significantly.

Formally, the BIOS version switch is left in its rightful place, but now these are just two identical BIOSes, nothing more. An attempt to flash an image from an older model will only lead to an increase in power consumption and reconfiguration of the fan control algorithm. However, the frequency will almost always be 950 MHz.

The kernel itself is also complex - inside each GPU there are several execution and monitoring devices located at once: for voltages, power consumption, video decoding functions, and so on. We are most interested in the control units responsible for voltage.

Oh yes, I almost forgot, it’s time to wean ourselves off the single voltage, now inside the GPU the familiar Vgpu is just an incoming value, which is then distributed among the internal parts of the core. If it has not grown to the required quality, then by increasing the voltage in some of its areas it is possible to force the GPU to function correctly, but the power consumption increases.

On the other hand, AMD has no choice yet, and any Hawaii core installed on the Radeon R9 290 has worse overclocking and power consumption compared to the R9 290X. This is clearly visible on the energy consumption graph, and it’s obvious from the heating that the truncated Hawaii core likes to heat up. These things are partially offset by fewer stream processors.

Specifications

Appearance and dimensions

A– length of the printed circuit board, excluding the cooling system and video output port strip.
IN– width of the printed circuit board, excluding PCI-E contacts and cooling system.
WITH– height from the horizontal plane of the printed circuit board to the level of the upper surface of the cooling system.
D– diameter of the fan/s along the outer radius.

A1– the length of the printed circuit board, taking into account the cooling system (if it extends beyond the printed circuit board) to the video output port strip.
IN 1– width of the printed circuit board, excluding PCI-E contacts, but with measurement of the cooling system (if it extends beyond the boundaries of the printed circuit board).
C1– height, taking into account the back plate (if any)/radiator mounting screws to the level of the upper surface of the cooling system. If it is lower than the height of the rear strip of the video output ports, then the height to the top point of the strip is measured.

It's obvious that the reference version of the Radeon R9 290 will be based on the R9 290X PCB, but that won't always be the case. AMD partners have already started equipping their own video cards with other fans, leaving the basic PCB design. But this will not last long, for example, ASUS and MSI will certainly announce unique designs with a modified power unit.

In the meantime, you can see a copy of all the main qualities of the R9 290X, starting with the printed circuit board and ending with the cooling system.

Printed circuit board

Even before the R9 290 sample appeared in our laboratory, it was clear that AMD would not make a separate version of the printed circuit board exclusively for the Radeon R9 290. And all the interest boiled down to the answer to one question: will the number of phases be reduced? The answer was found very quickly. After dismantling the cooling system, we compared both printed circuit boards and found no differences.

Want to check it out for yourself? Please…

The only problem is that not everyone is able to accurately determine which video card is on which side.

And the GPU looks identical to the one installed on the R9 290X.

More details about the printed circuit board and cooling system were described in the material “AMD’s new flagship. Review and testing of the AMD Radeon R9 290X video card."

Test stand

The AMD Radeon R9 290 video card was tested with the following configuration:

• Motherboard: ASUS MAXIMUS VI HERO (Intel Z87, LGA 1150);
• Processor: Intel Core i7-4770K 4500 MHz (100 MHz x 45, 1.25 V);
• Cooling system: water cooling system;
• Thermal interface: Arctic Cooling MX-2;
• RAM: GeiL EVO Veloce 2400 MHz, 2 modules x 8 GB, (10-12-12-31-1T, 1.65 V);
• Hard drive: SSD Corsair Force Series GT, 128 GB;
• Power supply: Corsair AX1200i Digital, 1200 Watt;
• Audio card: ASUS Xonar HDAV 1.3;
• Operating system: Microsoft Windows 7 x64 SP1;
• Drivers: AMD 13.11 Beta8, NVIDIA 331.40, in both cases CrossFireX compensation, SLI was enabled.

List of used instrumentation and instruments

• Sound level meter: Center 320;
• Multimeter: Fluke 289;
• Microphone: Philips SBC ME570;
• Electricity tarifficator: E305EMG.

CO testing methodology

To correctly measure temperature and noise, the following conditions were used. A room containing an automatic climate control system. In this case, the temperature level was set at 24°C +/-1°C. The accuracy of compliance with the specified parameters was monitored by four sensors, one of which was located 5 cm from the fan of the video card cooling system and was the leading one. It was used for the main correction of the room temperature.

The noise was measured at a distance of 50 cm to the video card. The background level was less than 20 dBA. An SSD was used as a hard drive, and the power supply, pump, and radiator with fans were located outside the room during measurements. There were no other components at the stand that made any noise.

Sound and video recording of the cooling system was made at a distance of ~10 cm from the fan. The first 5-10 seconds without load in idle mode, then 100% load was turned on using the Unigine Heaven Benchmark v4.0 program. The highest noise level is reached at the end of the audio recording. The temperature conditions and noise were determined in advance so that during the recording of the audio track you could hear exactly the maximum noise. While watching videos, you can highlight the timbre and nature of the sounds produced by the cooling system. I warn you that the sound on them is greatly embellished, that is, it is felt stronger than it actually is.

The level of electricity consumption during idle time was assessed according to the readings of the E305EMG tarifficator immediately after loading the operating system. The values ​​displayed on the graph correspond to the minimum achieved figures from the device. The video cards were tested under load using Unigine Heaven Benchmark v4.0. After 10-15 minutes, the temperature and fan speed reached their theoretical maximum, after which the data was entered into the table.

The temperature of the power circuits was measured by installing a temperature sensor in the space between the radiator and the thermal pad in the most loaded place.

I will try to explain in detail the nuances that arose during the testing process as they arise.

Pages of material

• Introduction, technical features and characteristics, appearance and dimensions, printed circuit board, test bench, CO testing methodology
• Study of the potential of the cooling system, overclocking results
• Power consumption, performance testing methodology, test results: Sleeping Dogs, Sniper Elite V2, Far Cry III
• Test results: Crysis 3, Tomb Raider (2013), Metro Last Light, Company Of Heroes II
• Test results: Total War: Rome II, Battlefield III, final results, average number of frames per second and their cost, conclusion

Hawaii has smaller islands: the speed is a little slower, but the price is much lower!

• Part 2 - Practical acquaintance
• Part 3 - Game Test Results (Performance)

We continue the saga about the “crabs”. You can read the beginning. So, the “reddish crab” made a “big ball” and successfully rolled it into the “hole”, which caused some nervousness in the “greenish crab”, who became even greener with anger and went to do something with his “balls”.

However, the “reddish crab” decided that such a huge “ball” was not enough, and, having trimmed the existing “balls”, making them a little smaller, he began to roll them into the “mink” too. Smart - yes he understands, quick-witted - yes he understands. For the rest - read below :)

Part 1: Theory and architecture

Following the release dedicated to the current line of graphics solutions from AMD, consisting of the Radeon R7 and R9 families, as well as a full review of the senior solution in the line - Radeon R9 290X, which became the first video card based on the new graphics chip codenamed Hawaii, material followed about the junior model of the top line, based on the same chip, which AMD decided to release a little later.

The release of such a video card after the older version is atypical for AMD. At first, the release of a video card one step below the top one was scheduled for a week later, and then it was postponed again, this time due to a military stratagem. AMD still could not finally determine the strategy and tactics of the fight of the Radeon R9 290 family with its competitors from Nvidia, for which this company reduced prices after the release of the top model Radeon R9 290X, also announcing the release of a new top - Geforce GTX 780 Ti. In this regard, AMD decided to further refine the drivers, squeezing out a few more percent of performance and at the same time bringing the date of the announcement of the Radeon R9 290 closer to the date of the actual appearance of the announced video cards on sale.

Let us remind you once again which video cards are included in the new families of AMD Radeon video cards. The company's current line contains several series: R9 and R7. The company's new line contains the following models, covering most market segments:

The R7 250 and R7 260X are designed for the $90-$140 US price range, with the R9 270X selling for $200 and the R9 280X selling for $300. The recommended price for the flagship of the line - the R9 290X model - on the US market is $549, and for the R9 290 they decided to ask another hundred and fifty less, and it is located between the R9 280X and R9 290X (it is not on the diagram).

The Radeon R9 290 model is not much different from the top-end R9 290X, since it is based on the same Hawaii video chip with Graphics Core Next architecture. This GPU has increased the number of execution units while keeping the power consumption of the solutions within acceptable limits. The main goal when developing a new GPU for AMD was to increase its energy efficiency, which was achieved, as we saw in the corresponding review.

It is clear that the new solution supports all the company's technologies, including ZeroCore, updated PowerTune, Eyefinity, etc., etc. The R9 290 simply costs less than the top-end R9 290X and provides an even better performance-to-price ratio, since it is not much inferior in price speed, but has a noticeably lower price. So, let's take a look at the AMD Radeon R9 290 graphics card.

Let's see how AMD itself is positioning its new solution on the market using the traditional Fire Strike tool in two different versions (Performance and Extreme) from the latest version of the Futuremark 3DMark benchmark. This benchmark is excellent for modern AMD video cards, and the number of points scored by the Radeon R9 290 video card is slightly higher than the performance of its main competitor, the GeForce GTX 780.

Since the new model of the Radeon R9 290 video card is practically a twin of the R9 290X video card, and the Hawaii graphics processor largely repeats the features of the GPU of the previous Radeon HD 7000 series, before reading this material it will be useful to familiarize yourself with detailed information about AMD’s early solutions:

• AMD Radeon R9 290X: Reach for Hawaii! You will achieve new heights of speed and functionality
• AMD Radeon R7 and R9 - updated line of video cards: new families so far without their flagship
• AMD Radeon HD 7790: senior representative of the middle class of accelerators
• AMD Radeon HD 7970: the new single-processor leader in 3D graphics

Radeon R9 290 graphics accelerator

• Chip codename: "Hawaii"
• Manufacturing technology: 28 nm
• 6.2 billion transistors
• A unified architecture with an array of common processors for stream processing of numerous types of data: vertices, pixels, etc.
• Hardware support for DirectX 11.2, including Shader Model 5.0
• 4 geometry processors
• 512-bit memory bus: eight 64-bit wide controllers, supporting GDDR5 memory
• Core clock up to 947 MHz (dynamic)
• 40 (out of 44 physically present on the chip) GCN computing units, including 160 (out of 176) SIMD cores, consisting of a total of 2560 (out of 2816) ALUs for floating point calculations (integer and floating formats supported, with FP32 precision and FP64)
• 160 (out of 176) texture units, with support for trilinear and anisotropic filtering for all texture formats
• 64 ROP units with support for full-screen anti-aliasing modes with programmable sampling of more than 16 samples per pixel, including with FP16 or FP32 frame buffer format. Peak performance up to 64 samples per clock, and in Z only mode - 256 samples per clock
• Integrated support for up to six monitors connected via DVI, HDMI and DisplayPort interfaces

Radeon R9 290 graphics card specifications

• Core frequency: up to 947 MHz
• Number of universal processors: 2560
• Number of texture blocks: 160, blending blocks: 64
• Effective memory frequency: 5000 MHz (4x1250 MHz)
• Memory type: GDDR5
• Memory capacity: 4 gigabytes
• Memory bandwidth: 320 gigabytes per second.
• Compute performance (FP32) 4.9 teraflops
• Theoretical maximum fill rate: up to 60.6 gigapixels per second.
• Theoretical texture sampling speed: up to 152 gigatexels per second.
• PCI Express 3.0 bus
• Two Dual Link DVI, one HDMI, one DisplayPort
• Power consumption up to 275 W
• One 8-pin and one 6-pin power connector;
• Dual slot design
• The recommended price for the US market is $399 (for Russia - 13,990 rubles).

We continue to get used to the new naming system for AMD video cards, but once again we express doubts about the correctness of the decision to name the Radeon R9 290 and 290X models, which differ only by an additional symbol in the name of the older model. Both video cards belong to the top-end R9 family and have the maximum serial number in the series - 290, but the older one has an additional “X” in its name, although in the previous family they made do with numbers. In our opinion, names like R9 2950 and R9 2970, as before, would be more logical and clearer.

As usual, the recommended price for the Radeon R9 290 is lower than the price of the corresponding solution from a competitor from the same price segment, even taking into account the price reduction Nvidia announced the other day. The Radeon R9 290 is clearly aimed at fighting the Nvidia Geforce GTX 780, based on the GK110 chip, which is the competitor’s junior top board (after all, the Geforce GTX Titan has been around for a long time, and the GTX 780 Ti has already been announced and will be released soon). The Nvidia model has a higher recommended price ($499 versus $399), but can provide better gaming performance - it's not the AMD-friendly Fire Strike from 3DMark. However, we will explore all such questions in the third part of the material.

The second top model of the AMD video card that we are considering today, like the older board, has GDDR5 memory with a capacity of four gigabytes. Since the Hawaii graphics chip has a 512-bit memory bus, which is not cut down in the Radeon R9 290, it was possible to install 2 GB, but for a top-end solution such a volume is too small, because modern projects are already starting to use large amounts of video memory, which will only get worse in the coming months, when multi-platform games designed for next-generation consoles: PS4 and Xbox One will begin to be released.

Unlike the older model, which has a special switch on the card that allows you to select one of two BIOS firmwares: (quiet mode and super mode), differing in power consumption and performance, the younger Radeon R9 290 model does not have this option; only one mode is available for it. It is clear that power consumption has decreased compared to the R9 290X, especially when compared with the latter's super mode, but official specific figures are unknown.

Architectural features

We're already talking about the new Hawaii graphics chip, which powers the AMD Radeon R9 290X and 290 graphics cards. This GPU is based on the long-known Graphics Core Next (GCN) architecture, which has been modified in terms of computing capabilities and support for all DirectX 11.2 features, similar to the Bonaire chip (Radeon HD 7790 and R7 260X). Architectural changes also include support for more concurrent threads and a new version of AMD PowerTune technology.

We already know that the Hawaii graphics chip differs from Tahiti in a large number of actuators. But how does the Hawaii version used in the Radeon R9 290 differ from the full version? Let's see exactly how AMD engineers “slowed down” the junior top model so that it would not be too strong a competitor for the R9 290X:

The Hawaii GPU consists of four Shader Engines, each of which physically contains 11 enlarged Compute Units, including texture modules, a geometry processor and rasterizer, as well as ROP units. In total, the Hawaii graphics chip includes: 44 Compute Units containing 2816 stream processors, 64 ROP units and 176 TMU units.

But this applies to a full-fledged chip with all active units, which is used in the Radeon R9 290X. The younger R9 290 received a chip with 40 active Compute Units, containing 2560 stream processors and 160 texture units. But the number of ROP blocks was not cut; there were 64 of them left. The same applies to the memory bus; it remains 512-bit, consisting of eight 64-bit controllers.

In our review of the Radeon R9 290X, we looked in detail at the internal structure of the graphics processor, the shader engines that make it up, and the Compute Units (CU). The only difference between the R9 290 and the R9 290X is that the shader engine in this case includes not 11 CU units, but fewer (for example, 10 pieces if one CU is disabled in each engine). And each CU computing unit contains functional blocks: 16 texture fetch modules, 4 texture filtering modules, a branch prediction unit, a scheduler, four vector and one scalar computing units, memory for vector and scalar registers, 16 KB of first-level cache and 64 KB of shared memory.

Note that the Hawaii chip has four geometry processing units and rasterization engines and can process up to four geometric primitives (triangles) per clock cycle. Also in Hawaii, the buffering of geometric data has been improved and the caches for the parameters of geometric primitives have been increased, which provides a significant increase in performance with large volumes of calculations in geometry shaders and the active use of tessellation.

The Hawaii chip includes 16 Render Back End units (twice the Tahiti) containing 64 ROP units capable of processing up to 64 pixels per clock. This GPU has 1 megabyte of L2 cache, which is divided into 16 64 KB sections, and their internal bandwidth has also been increased by a third. Memory is accessed using eight 64-bit controllers, and since the memory chips in the Radeon R9 290 also operate at 5.0 GHz, like the older model, the total memory bandwidth remains unchanged at 320 GB/s.

Functional Features

It is quite logical that the new Radeon R9 290 video card supports exactly the same technologies as the older model. We have already written about all the new technologies supported by the Hawaii graphics chip in the corresponding review of the Radeon R9 290X, and we will repeat this information here in a very brief form.

The solution presented today has support for the new graphics API Mantle, which is able to take full advantage of the hardware capabilities of AMD GPUs, without being limited by the software limitations of existing graphics APIs like OpenGL and DirectX and using a thinner software wrapper between the game engine and GPU hardware resources, similar to what is done on game consoles. Mantle was developed at AMD with significant input from leading game developers at DICE, and Battlefield 4 is the first project to use Mantle.

Theoretically, using Mantle can provide up to a nine-fold advantage in the execution time of rendering function calls compared to other graphics APIs, but such an advantage is only possible in artificial conditions; in real games it’s good if it’s ten percent (which is also not bad). Mantle API support is expected to appear in Battlefield 4 in December, when a free update optimized for AMD Radeon video cards will be released. On systems with GCN architecture video cards, the Frostbite 3 engine will use this API to reduce the load on the CPU by parallelizing work across eight cores, and will also introduce special low-level performance optimizations for AMD video chips.

Another important technology that the Radeon R9 290 supports is “sound” technology. AMD TrueAudio, which we have also already talked about. With the release of the Radeon R7 and R9 series, AMD introduced a programmable audio engine, which is supported only by the AMD Radeon R7 260X, R9 290 and R9 290X video card models, since the Bonaire and Hawaii chips are the only company developments that have received the latest technologies .

TrueAudio is an embedded programmable audio engine in AMD GPUs that provides real-time processing of audio tasks regardless of the installed CPU. Hawaii and Bonaire chips integrate several Tensilica HiFi EP Audio DSP cores, and their capabilities are accessed using popular audio processing libraries, the developers of which use the resources of the built-in audio engine using a special AMD TrueAudio API.

The new top-end video cards have improved and PowerTune power management technology. We wrote about these improvements back in our review of the Radeon HD 7790. For more efficient power management, the latest AMD graphics chips have multiple states with different frequencies and voltages, which allows for higher clock speeds. The Hawaii video chip integrates a second-generation serial voltage regulator interface and the GPU always operates at the optimal voltage and frequency for the current GPU load and energy consumption, on which switching between states is based.

In addition, the transitions between states themselves have become very fast, the GPU switches PowerTune states 100 times per second, and the Radeon R9 290 series solutions simply cannot have a single operating frequency, there is only an average over a certain period of time. Based on these global changes, modified overclocking and PowerTune settings have appeared in the Catalyst Control Center driver settings in the OverDrive tab:

The Power Limit and GPU clock are linked together in the power consumption and heat dissipation diagram, which makes overclocking settings more logical and understandable. In addition, the diagram reflects the fully dynamic control of GPU frequency that has appeared in the solutions of the series. The increase in GPU frequency is now indicated by increasing the corresponding value (GPU Clock) by a certain percentage, and there is no longer the ability to specify a specific frequency.

Also in the new OverDrive interface, the fan speed control has been seriously changed. Unlike the previous ability to set a fixed fan speed, the new interface has a maximum fan speed setting (“Maximum Fan Speed”), which only sets the upper speed limit for the fan. In this case, the fan rotation speed will change based on the GPU load and its temperature, and will not remain fixed, as was previously the case.

The most interesting innovation in the AMD Radeon R9 290 series video cards is support for the technology AMD CrossFire without the need to connect video cards to each other using special bridges. Instead of dedicated communication lines, the new GPUs communicate with each other over the PCI Express bus, using the built-in hardware DMA engine. The performance and image quality are ensured exactly the same as with connecting bridges.

In the case of a pair of Radeon R9 290 graphics cards, CrossFire technology provides excellent scaling in multi-chip systems when taking into account average frame rates. The following chart shows the comparative performance of a single AMD Radeon R9 290 and two such cards running AMD CrossFire rendering technology:

The games shown in the diagram provide a very good increase in the average frame rate when connecting a second video card - almost double. In the worst case, these applications show 80% CrossFire efficiency, and on average it is more than 85%.

About changes in image output to displays, technologies AMD Eyefinity We also already wrote about support for Ultra HD resolution. The new video card provides connection to two possible options for Ultra HD displays: TVs that support only 30 Hz and below at a resolution of 3840×2160 and connect via HDMI or DisplayPort, as well as monitors whose image is divided into two halves with a resolution of 1920×2160 at 60 Hz .

To support split monitors, the new VESA Display ID 1.3 standard has been introduced, which describes additional display capabilities. The standard allows for automatic image stitching for such monitors if it is supported by both the monitor and the driver. And in the future, AMD Radeon video cards will support a third type of UltraHD display, which requires only one thread to operate in ultra-high resolution at a refresh rate of 60 Hz.

The Radeon R9 290 delivers high enough 3D performance for multi-monitor configurations, which is essential when running at maximum gaming settings and the highest rendering resolutions on such systems. The new product has a clear advantage over the Nvidia Geforce GTX 780, expressed in a larger amount of video memory, which is important in ultra-high resolutions, as can be seen when comparing performance in Ultra HD resolution:

As you can see, the performance of the new solution makes it possible to play at maximum settings in this resolution, getting an acceptable average frame rate above 30 FPS (and you won’t be able to see more in the case of some displays) even in such demanding games as Crysis 3 and Battlefield 3.

The technology in the Radeon R9 series of video cards has been updated AMD Eyefinity. The new top-of-the-line Radeon R9 290 graphics card supports configurations of up to six displays using all available connectors, and up to three HDMI/DVI displays. For the latter case, without using DisplayPort connectors with hubs, you will need three displays that support identical timings. Output configuration is only done at system startup and does not support display hot-plugging for a third HDMI/DVI connection.

Theoretical performance evaluation and conclusions

Let's look at the theoretical performance indicators of the new product, estimating how slower the new Radeon R9 290 video card should be than the top model Radeon R9 290X in reality and how comparable it is to the R9 280X, previously known as the Radeon HD 7970 GHz. Let’s compare all the most important theoretical indicators for GPUs:

The peak geometry processing speed of solutions based on Hawaii (R9 290 and R9 290X) is almost twice as high as that of solutions based on Tahiti (R9 280X and HD 7970 GHz), and in terms of computational and texture performance, the new product is located exactly in the middle between the R9 290X and R9 280X . The video memory bandwidth of both top boards is the same and it is 20% higher than that of the R9 280X, and the fill rate of the R9 290 is almost the same as that of the R9 290X (on the plate it is equal, but this is an error - in the case R9 290 this figure is 60.6 gigapixels per second).

To confirm the capabilities of the new product, especially in terms of memory bandwidth and fill rate, AMD cites the average frame rate achieved in the newest game Battlefield 4 - the continuation of the most popular series developed by DICE. The game is part of the AMD Gaming Evolved partner program, and therefore has no problems with optimizing Battlefield 4 for GCN architecture GPUs. Let's see what happens in the version of the game without Mantle support:

In this game, the new Radeon R9 290, albeit slightly, is still ahead of the more expensive competing Nvidia Geforce GTX 780 in both modes with different (but high in both cases) resolutions. Perhaps the Nvidia video card in such high resolutions is hampered by the lack of video memory, which it has only 3 GB, in contrast to 4 GB for the Radeon R9 290, so the larger amount of video memory is another undeniable advantage of AMD's new product.

We will test the performance of the new solution in the following parts of the article. To summarize the theoretical part, we note that the AMD Radeon R9 290 video card model announced today should become one of the most profitable offers among single-chip 3D accelerators, and one of the most successful acquisitions in the top price segment intended for enthusiasts. It is also very important that the new model should not lag much behind the Radeon R9 290X in speed, which is the most expensive and powerful single-chip solution in the company’s line.

AMD offered the market a video card with a very competitive price and capabilities. Already based on theoretical indicators, it can be argued that this video card model has an excellent price-performance ratio. And the functionality of the new product is all right, it has a built-in DSP sound engine, supports TrueAudio technology and the new Mantle graphics API, as well as all the other technologies that are already familiar to us, which we are already tired of listing.

After we got acquainted with the characteristics and theoretical capabilities of the Radeon R9 290 video card, it’s time to move on to the next part of our material, which is traditionally devoted to studying the rendering speed of the new AMD video card in a set of synthetic tests. It will be interesting to compare the performance of the new product from the upper price segment with the top-end Radeon R9 290X, as well as with the speed of competing Nvidia video cards.

AMD Radeon R9 290 - Part 2: video cards and synthetic tests →

To measure system power, a stand with a Corsair AX1200i power supply is used. Energy-saving CPU technologies are disabled in all tests. The PCI-Express bus operates in 3.0 mode. To activate PCI-E 3.0 on GeForce 600 and 700 series video cards in a system based on the X79 chipset, a patch from NVIDIA is used.

In the NVIDIA driver settings, the CPU is always selected as the processor for PhysX calculations. In AMD settings, the Tesselation setting is always transferred from the AMD Optimized state to Use application settings.

⇡ Test participants

The following video cards took part in performance testing:

• AMD Radeon R9 290 (947/5000 MHz, 4 GB)

⇡ Overclocking, temperature, power consumption

As we already said, AMD decided in the last days before the launch of the Radeon R9 290 to boost its performance by changing the fan speed limit from 40% to 47%. Before we begin to analyze the performance, let's see how this decision affected the GPU temperature, clock frequency fluctuations and the acoustic properties of the video adapter.

The initial limit of 40% corresponds to the so-called Quiet Mode of the Radeon R9 290X, but the actual rotation speeds differ: 1987 rpm for the R9 290X and 2172 rpm for the R9 290.

As a result, as soon as the core temperature approaches the allowed 95 °C under load (remember that this is officially considered the norm for the Hawaii chip), the fan can no longer spin faster - and active frequency throttling begins. During a test run of Crysis 3, drops of up to 851 MHz were recorded from the original 947 MHz.

Overclocking the impeller to 47% of the nominal value raises the real rotation speed to 2893 rpm, which is close to the Uber Mode level of the Radeon R9 290X (50% of the nominal value). In this case, throttling of the GPU frequency completely disappears, although the card becomes noisier than with a limit of 40% of revolutions. This is AMD's secret sauce that boosts the performance of the Radeon R9 290.

For uncompromising overclocking using all the tools that the new version of PowerTune puts at the user's disposal, you need to raise the TDP limit to the maximum 50% and, covering your ears, run the fan at 100% speed, and this is no less than 5300 rpm. However, for everyday use it is certainly possible to select a less noisy mode, in which the GPU will still be cooled quite effectively during overclocking without reducing the frequency.

Under these conditions, the Radeon R9 290X graphics processor and memory were overclocked to the same level as the R9 290X components - 1100 and 6000 MHz, respectively. By the way, either it’s because of the four disabled Compute Units, or because the “turbine” of the R9 290 we received reached a speed of 200 rpm more than the Radeon R9 290X cooler (apparently, there is a variation between individual devices) , but the GPU of the 290 showed a maximum temperature several degrees lower, and its frequency graph is more uniform than that of the R9 290X: there is practically no throttling when overclocked to the same frequency. The minimum registered frequency under load is 1084 MHz.

The Radeon R9 290 actually differs little in power from its older brother, the R9 290X. And it’s not surprising, considering that their frequencies differ by only 5%.

Since the default GPU temperature limit is 95 °C, when the cooler speed is automatically controlled, the core warms up to exactly this level under load. However, the power reserve of the cooling system is colossal - if you run the turbine at maximum speed, the temperature can be reduced to 66 °C, despite significant overclocking.

Performance: Synthetic Benchmarks

3DMark 2011 did not support AMD's ambitious plans: the Radeon R9 290X was ahead of the GeForce GTX 780, but the R9 290 was no longer there.

And yet, the R9 290 is superior to the GeForce GTX 780 by a margin. Just a few points were not enough to catch up with the GTX TITAN.

Unigine Heaven 2

Radeon R9 290 and GeForce GTX 780 are approximately equal.

• The new product also differs little from the Radeon R9 290X.

⇡ Performance: Gaming Tests

Crysis Warhead

• A benchmark classic, and a major success for the Radeon R9 290: the new product differs little from the R9 290X and is even faster than the GeForce GTX TITAN.

Battlefield 3

The Radeon R9 290 is once again ahead of the GeForce GTX 780 and close on the heels of the GeForce GTX TITAN.

Batman: Arkham City

The difference between the Radeon R9 290 and R9 290X is small.

But this doesn’t matter, because in Batman: AC the advantage is always on the side of NVIDIA.

DiRT Showdown

On the contrary, this is a game with a clear tilt in favor of AMD. As a result, the Radeon R9 290 effortlessly dealt with the GeForce GTX TITAN.

Far Cry 3

The difference between the Radeon R9 290 and R9 290X is quite noticeable.

Because of this, the R9 290 was defeated by the GeForce GTX 780 and only slightly outperformed the GTX 770.

Tomb Raider

And again - the Radeon R9 290X is approximately equal to the GeForce GTX 780, but the Radeon R9 290 is inferior to both.

Bioshock Infinite

Luck has returned to the Radeon R9 290: at a resolution of 1920x1080, the new product outperforms the GTX 780, and at 2560x1440 it outperforms the GTX TITAN.

Crysis 3

The difference between the R9 290 and R9 290X is again small.

At 1920x1080 the GeForce GTX 780 is faster than the R9 290, but parity is established at 2560x1440.

Metro: Last Light

The results of the Radeon R9 290 are noticeably different from what the R9 290X can achieve.

And yet, the R9 290 can beat even the GTX TITAN here, not to mention the GeForce GTX 780.

⇡ Performance: at different fan speeds, overclocked

Since the victory of the Radeon R9 290 over the GeForce GTX 780 came at the cost of operating the cooling system at higher speeds and the accompanying noise, it is interesting to check what will happen if you limit the fan speed to 40% of the nominal value, as was originally intended for the R9 290. As a result, we can say that overclocking the cooler did not have a significant impact on the results of most benchmarks, and only in certain tests leads to a significant increase in frame rate. However, it was precisely this small contribution of the Radeon R9 290 that was not enough to confirm its unconditional superiority over the GeForce GTX 780. We also note that we are not ready to give a hand to cutting off the results of the card with a 40% rev limit, since “throttling” is guaranteed to appear GPU frequencies, some of the tests take longer than expected in our standard methodology, and it was too late to change anything in the short time allotted for preparing the review. But there is no doubt about the data of the Radeon R9 290 with a fan speed of 47%: a long test run in Crysis 3 shows that even when the maximum temperature is reached, there is no throttling.

Overclocking the Radeon R9 290 by the user to the same frequencies that the R9 290X previously conquered, as in the case of the latter, has a very noticeable effect on performance both in synthetic benchmarks and in real games.

3DMark(note: once again, we note that this is still unofficial data, although a refutation is unlikely). There is a certain irony in the fact that NVIDIA, having lowered the prices for the GeForce GTX 770 to $299 and the GTX 780 to $499, which in itself was a completely logical and strong move, improved the position of the Radeon R9 290. AMD slightly improved the performance of the card in the last days before launch, and as a result, from a “plug” in the range between $300 and $500, where NVIDIA has no offers yet, the Radeon R9 290 turned into an analogue of the GeForce GTX 780 with a convincing “only cheaper” tag.

In general, we are now witnessing an interesting pricing game between NVIDIA and AMD, similar to a “fool’s game.” NVIDIA compensated for the defeat of the GeForce GTX 780 by the Radeon R9 290X by lowering it to a lower price category, but even there it was overtaken by a powerful rival. Continuation depends on NVIDIA's willingness to make new sacrifices. As an option, the GTX 780 will go up due to the new version of the GK110 silicon, which will provide higher operating frequencies.

But let's return to the Radeon R9 290. In pursuit of the GeForce GTX 780, this adapter came close to the leader of its column - the R9 290X. The difference in results between them in most games does not exceed 10-15%, which means no matter how good the Radeon R9 290X is as a device on a fully unlocked Hawaii chip, its version “without X” has a hefty bonus in terms of price/features ratio. Imagine the performance of a giant GPU, previously only achievable with two smaller cores in an SLI or CrossFireX configuration, now available for just $449. In this case, it would be unfair to deprive the Radeon R9 290 of its well-deserved “best buy” medal.

At the last moment before publishing the review, a letter from AMD finally arrived with official prices. In the US market, the Radeon R9 290 will cost $400, in Europe - €289. This is $50 less than previously thought.

After such news, the Radeon R9 290 is doubly worthy of its award, but now the hand is reaching out to take the medal away from the Radeon R9 290X, which so impressed us two weeks earlier. In fact, the difference between these two cards is sometimes so small that it resembles the case of “GeForce GTX 780 vs GTX TITAN” and allows us to talk about the so-called cannibalization within the AMD line of discrete video adapters.

However, if for some reason the manufacturer is ready to jeopardize the position of the flagship card in order for a younger model to outshine the GeForce GTX 780 at any cost, then that’s its problem. The only pity is that buyers, to some extent, have become victims of AMD's zeal. From the Radeon R9 290X, its younger version inherited serious power consumption, which requires a certain compromise between performance and acoustic characteristics. But if owners of the R9 290X (at least in the reference version) initially have a choice between Quiet Mode and Uber Mode, which significantly affects the noise level, then the R9 290, with the light hand of AMD marketers, is doomed to work in a state close to Uber Mode, unless the user manually tightens the wick on the cooling system.

However, the situation with performance and noise also depends on the efficiency of the cooler itself. Turbine designs, which are traditionally used in reference design adapters, are generally not very quiet (especially AMD), and open-type cooling systems that partner companies will introduce in the near future may improve this. By the way, when choosing the original Radeon R9 290/290X, you will now have to read the reviews especially carefully, because not only the GPU temperature now depends on the cooling efficiency (which is mostly a problem for the machine, not the user), but also the performance of the video card, all other things being equal - standard frequencies and noise levels.