Review: AMD Radeon HD 7750 and HD 7770 + XFX Black Super OC Edition - BeHardware
>> Graphics cards

Written by Damien Triolet

Published on February 17, 2012


Page 1


Following the release of the Radeon HD 7900s, AMD has attacked the mid-range segment with Cape Verde, the smallest of the three new generation GPUs. Two graphics cards designed around this GPU have been released: the Radeon HD 7770 and 7750. We have looked in detail at how the reference models and the XFX Radeon HD 7770 Black Super OC Double Dissipation Edition do.

Successors to the HD 6700s, not to the HD 5700s!
AMD planned three GPUs for its new family of graphics cards: Tahiti, Pitcairn and Cape Verde. They have been designed respectively for the high-end, the performance segment and the mid-range. AMD has not designed any new GPUs for the entry level, considering this segment to be sufficiently well provided for with its APUs and renamed Radeon HD 6600s/6500s/6400s for use in computers and other devices and no doubt for retail sale too soon.

Cape Verde and the Radeon HD 7700s were designed to succeed the Radeon HD 6700s, which are in reality renamed Radeon HD 5700s. The HD 5700s were introduced in the performance segment, which is the gamer segment that garners the highest sales, but have since been replaced by the Radeon HD 6800s. Remember that at launch of the 6800s, AMD modified its naming structure, moving them up by a hundred. The performance segment thus became the 800 series, the high-end the 900 series and the mid-range the 700 series.

Itís in this sense that Cape Verde is more the successor to the mid-range Radeon HD 6700s than the Radeon HD 5700s, which, while identical, represented the next segment up at launch. In other words we shouldnít expect Cape Verde to have the same impact as Juniper had when it was introduced with the Radeon HD 5700s but rather to represent a modest development in the mid-range AMD offer.

Before moving on, note that AMD has decided to highlight the fact that Cape Verde is the first GPU to clock at 1 GHz on series cards with the launch of the GHz edition label that will no doubt be used in the marketing communication on graphics cards clocked at this frequency or more.

Page 2
Cap Verde: GCN architecture and 28nm

Cape Verde
Cape Verde is a small GPU coming in at basically a third of Tahiti, used in the Radeon HD 7900s. Tahiti has a 384-bit memory bus and 32 fundamental blocks, the CUs, each of which contains 64 processing units and 4 texturing units. Cape Verde has just 10 CUs with a total of 640 processing units and 40 texturing units as well as a 128-bit memory bus.

This memory bus is fed by 16 ROPs charged with writing pixels to the memory. Tahiti has 32 ROPs and can output 2 triangles per cycle, which is no longer the case with Cape Verde. Hereís a visual of Cape Verdeís internal organisation:

Cape Verde adopts the specificities of the Graphics Core Next architecture that we described in detail here. This includes an efficiency gain linked to the scalar type functioning of the processing units, a more advanced cache architecture, DirectX 11.1 support, PCI Express 3.0 support and various developments designed with GPU Computing in mind.

Cape Verde includes all the Tahiti specificities, with two exceptions:

- Double precision processing is at the equivalent of 1/16th of single precision against 1/4th on Tahiti, which makes sense on GPUs that havenít been designed for professional processing.
- The L2 cache per 64-bit memory controller is up from 128 KB on Tahiti to 256 KB for Cape Verde, probably to prevent the total L2 cache from becoming too small.

Astonishingly, Cape Verde retains the 2 ACEs, which manage compute type tasks, as well as the 2 DMA engines loaded via the PCI Express port. It also supports PowerTune to control the cardís thermal envelope and ZeroPower, which almost turns off the GPU entirely in deep sleep mode (screen in standby).

Cape Verde and 2 of the 4 memory modules that supply it.

In spite of everything there are 1.5 billion transistors here. AMD has taken advantage of the 28nm fabrication process to fit them on a surface of more or less 125 mm≤, to be compared with 1 billion transistors engraved at 40nm that took up 172 mm≤ on Juniper, the GPU used on the Radeon HD 5700s and HD 6700s.

Page 3
Specifications, reference HD 7750 and 7770


The Radeon HD 7700s do not have as much processing power as the Radeon HD 5700s/6700s, but their processing units are more efficient. Their processing power however benefits from the increase in clock and gives them more texturing power than the older cards.

Note that the performance gap between the Radeon HD 7770 and 7750 is bigger than between the Radeon HD 5770 and 5750: -36% compared to -26%.

For this test, AMD supplied us with a reference Radeon HD 7770 and 7750:

The reference Radeon HD 7770

In a 21 cm long double slot format, the reference Radeon HD 7770 is relatively compact. It has the same connectivity introduced with the Radeon HD 7900s, namely: a DVI Dual Link out, an HDMI 1.4a 3 GHz out and two mini-DisplayPort 1.2 outs.

With a TDP of 100W, its cooling system is quite simple and is based on a rounded radiator cooled using an axial fan, all of which is covered in a plastic casing in the same successful style as the Radeon HD 7900s.

The PCB is rather pared down and of sufficient size for the positioning of the cooling system. There are three phases for the GPU and one 6-pin PCI Express power connector is required. Thereís an additional phase for the Hynix T2C GDDR5 certified at 1.25 GHz.

With respect to overclocking, we managed to modify the GPU voltage to obtain the following (GPU clock / memory clock @ GPU voltage):

1000 / 1125 MHz @ 1.075V: 63W
1000 / 1125 MHz @ 1.174V: 76W (by default)
1125 / 1525 MHz @ 1.174V: 88W
1150 / 1525 MHz @ 1.225V
1200 / 1525 MHz @ 1.250V: 111W

The reference Radeon HD 7750

The reference Radeon HD 7750 is just a single slot and measures 17cm long. Its connectivity has also been simplified: one DVI Dual Link out, a 3 GHz HDMI 1.4a out and a mini-DisplayPort 1.2 out. Just like the reference Radeon HD 7770, it can support up to six screens via a DisplayPort hub.

With a TDP announced at 75W, only a compact cooler is required: an aluminium plate, a few fins and a small low profile fan that looks as if it wonít be all that effective and with relatively high noise levels.

The card is powered via the PCI Express bus, drawing 75W in all, of which 66W maximum via the 12V. There are two phases for the GPU and an additional phase for the Hynix T2C GDDR5 certified at 1.25 GHz.

We werenít able to modify the GPU clock and the card proved uncooperative with respect to overclocking (GPU clock / memory clock @ GPU clock):

800 / 1125 MHz @ 1.093V: 45W (by default)
850 / 1200 MHz @ 1.093V: 49W

Moreover we noticed that the card is much more rapidly limited by PowerTune than the other Radeons equipped with this technology. Without overclocking, this doesnít have much of an impact on gaming performance. With overclocking, even if slight, we noted a few dips in the most demanding games. We wonder therefore if the TDP might not in reality be lower than the 75W communicated by AMD.

Page 4
XFX Radeon HD 7770 Black Super OC Edition DD

XFX Radeon HD 7770 Black Super OC Double Dissipation
XFX, visibly very active with the Radeon HD 7000 series, is marketing three versions of the Radeon HD 7770 using its Double Dissipation cooling system:

Radeon HD 7770 Double Dissipation Edition: 1000/1125 MHz
Radeon HD 7770 Black Edition featuring Double Dissipation: 1095/1125 MHz
Radeon HD 7770 Black Super OC Edition featuring Double Dissipation: 1120/1300 MHz

The last model was supplied to us by XFX:

The card uses the reference PCB, which is however made by XFX. It's therefore still in the relatively compact 21cm format, but is a double slot. It has the same connectivity introduced with the Radeon HD 7900s, namely: a DVI Dual Link, a 3 GHz HDMI 1.4a out and two mini-DisplayPort 1.2 outs.

The PCB is rather pared down and of sufficient size for the positioning of the cooling system. There are three phases for the GPU and an additional phase for the Hynix T2C GDDR5 memory certified at 1.25 GHz. It requires a 6-pin PCI Express power supply connector.

Note that the GPU packaging is slightly different on the XFX model: thereís a protection that you wonít see on the reference cards.

This model stands out for its cooling system. The Double Dissipation cooler used here consists of a copper base to which a wide aluminium radiator is fixed, cooled by two 8cm axial fans. A brushed aluminium casing has been built around the card and gives it a very nice look.

With respect to overclocking, we managed to modify the GPU voltage to obtain the following (GPU clock / memory clock @ GPU voltage):

1120 / 1300 MHz @ 1.150V
1120 / 1300 MHz @ 1.174V: 79W (by default)
1150 / 1400 MHz @ 1.174V
1175 / 1400 MHz @ 1.200V
1200 / 1400 MHz @ 1.225V
1225 / 1400 MHz @ 1.250V: 106W

XFX supplies its card with an installation guide for the card and drivers, a CD for the drivers, a product registration code, a brand insignia, an HDMI to DVI adaptor, a DVI to VGA adaptor and a CrossFire X bridge. It should be available soon for Ä170.

Page 5
Noise, heat and IR thermography

Noise, heat and infrared thermography
For this test we used the new protocol as described here.

Here first of all is a summary of all the readings:

The reference Radeon HD 7750 is rather noisy at idle while the XFX Double Dissipation model is really quiet.

In load, the reference Radeon HD 7750 is still the noisiest of the three cards, while the XFX model is the quietest. Once the two Radeon HD 7770s have been overclocked, the XFX card has an even bigger advantage.

Hereís what our thermal imaging showed:

Reference Radeon HD 7750
Reference Radeon HD 7770
XFX Radeon HD 7770 Black Super OC
  [ Idle ]  [ Load ]
  [ Idle ]  [ Load ]  [ OC ]
  [ Idle ]  [ Load ]  [ OC ]

In spite of the significant factory overclocking, the XFX Radeon HD 7770 Black Super OC Double Dissipation is better cooled than the reference Radeon HD 7770.

Page 6
Energy consumption and performance/watt

Energy consumption
We used the test protocol that allows us to measure the energy consumption of the graphics card alone. We took these readings at idle on the Windows 7 desktop as well as with the screen in standby so as to observe the impact of ZeroCore Power. In load we opted for the readings in Anno 2070, at 1080p with all options pushed to maximum, as well as those in Battlefield 3, at 1080p in High:

The Radeon HD 7700s draw very little power at idle, two of them using under 10W on the Windows 7 desktop. With the screen in standby, the ZeroCore Power technology functions well and can reduce energy consumption to 1.2W!

In load, the Radeon HD 7770 draws about the same amount of power as the Radeon HD 5770 in Anno 2070 but consumes 10W less in Battlefield 3. The Radeon HD 7770s seem to be particularly well balanced for Anno 2070 with maximum details, which means that their processing and texturing units run at full speed and therefore give slightly higher performance and draw slightly more power than on average.

The Radeon HD 7750 has a significantly lower performance. Its energy consumption doesnít vary much between the two games, which can be explained both by the different processing unit balance and PowerTune.

If we increase the PowerTune limit by 20% in Anno 2070, the reference Radeon HD 7770ís power consumption is unchanged but the reference Radeon HD 7750 increases to 48W and performance is up from 15.3 to 15.4 fps. The impact of PowerTune is therefore minimal, even if the Radeon HD 7750 seems much closer to its limit than the other cards using it. In Battlefield 3, modifying the PowerTune limit doesnít change energy consumption for any of these cards.

Here's a graphical representation of the energy consumption readings, with fps per 100W to make the data more legible:

[ Anno 2070 1080p Max ]  [ Battlefield 3 1080p High ]

Thanks to the 28nm fabrication process, the energy yield has improved significantly. Note however that each game represents a particular case and that the mid-range GeForces lag somewhat in Anno 2070 when the settings are pushed to a maximum while the architecture of the Radeon HD 7700s brings a significant gain. Thereís not so much of a gap in Battlefield 3 but the trend is the same.

Page 7
Theoretical performance: pixels

Texturing performance
We measured performance during access to textures of different formats in bilinear filtering. for standard 32-bit (8xINT8), 64-bit ďHDRĒ (4x FP16), 128-bit (4x FP32), 32-bit depth (D32F) and 32-bit RGB9E5, an HDR format introduced with DirectX 10 which enables the storing of 32-bit HDR textures with a few compromises.

The GeForce GTX 500s can filter FP16 textures at full speed but the Radeons have such superior filtering power that even though they have to filter FP16 textures at half-speed, they arenít far behind the GeForces.

Note that we have to increase the energy consumption limit of the Radeon HD 6900s and the Radeon HD 7700s to a maximum here, otherwise the clocks are cut during this test. By default the Radeons therefore seem incapable of fully benefitting from their texturing power! The good news is that this is not the case for the Radeon HD 7900s.

We measured the fillrate without and then with blending, and this with different data formats:

In terms of fillrate, the Radeons have an advantage over the GeForce GTX 580s, above all with FP10s, a format processed at full speed while with the GeForces itís processed at half-speed. Given the limitation of the GeForces in terms of datapaths between the SMs and ROPs, itís a shame that NVIDIA hasnít given its GPUs the possibility of benefitting from FP10 and FP11 formats.

Although the GeForces and the Radeon 6900s and 7000s can process FP32 single channel at full speed without blending, only the Radeons retain this speed with blending. With just a 128-bit memory bus, the speed of the Radeon HD 7700s with blending is seriously limited as this mode is very demanding in terms of memory bandwidth.

Page 8
Theoretical performance: geometry

Triangle throughput
Given the architectural differences between the various GPUs in terms of geometry processing, we obviously wanted to take a closer look at the subject. First of all we looked at triangle throughput in two different situations: when all triangles are drawn and when all the triangles are removed with back face culling (because they arenít facing the camera):

Although the Radeon HD 7900s and 6900s are indeed able to process 2 triangles per cycle, the GeForce GTX 580 retains the advantage with 4 triangles per cycle. When the triangles have to be rendered however, performance is reduced as NVIDIA has limited them to differentiate the Quadros and the GeForces.

Here the Radeon HD 7700s behave as expected: 1 triangle per cycle.

Next we carried out a similar test using tessellation:

While the Radeon HD 7700s donít really stand out from the Radeon HD 5700s without tessellation, their performance is significantly better with.

The architecture of the Radeons means that they can be overloaded by the quantity of data generated, which then drastically reduces their speed. Doubling the size of the buffer dedicated to the GPU tessellation unit in the Radeon HD 6800s meant they gave significantly higher performance than the Radeon HD 5000s. AMD has continued down this line with the Radeon HD 7000s.

Strangely, the Radeon HD 7900s donít seem to be able to benefit from their double tessellation unit when the triangles have to be rendered and donít have any advantage over the Radeon HD 7700s.

Page 9
Test protocol

The test
For this test, we used the protocol introduced for the report on the Radeon HD 7970 which includes some new games: Anno 2070, Batman Arkham City, Battlefield 3, F1 2011 and Total War Shogun 2.

We have decided no longer to use the level of MSAA (4x and 8x) as the main criteria for segmenting our results. Many games with deferred rendering offer other forms of antialiasing, the most common being FXAA, developed by NVIDIA It therefore no longer makes sense to organise an index around a certain level of antialiasing, which in the past allowed us to judge a card according to its effectiveness with MSAA, which can vary according to implementation.

At 1920x1080, we carried out the tests with a very high level of quality, which automatically included some antialiasing (either MSAA 4x, or FXAA/MLAA/AAA) and at a medium quality level.

We no longer show decimals in game performance results so as to make the graph more legible. We nevertheless note these values and use them when calculating the index. If youíre observant youíll notice that the size of the bars also reflects this.

The Radeons were tested with the beta 8.932.2 drivers and the GeForces with the 295.51 beta drivers.

Test configuration
Intel Core i7 980X (HT and Turbo off)
Asus Rampage III Extreme
Corsair 6 GB DDR3 1333
Windows 7 64-bit
GeForce 295.51 beta drivers
Catalyst beta 88.932.2

Page 10
Benchmark: Anno 2070

Anno 2070

Anno 2070 uses a development of the Anno 1404 engine which includes DirectX 11 support.

We used the Very High quality mode on offer in the game and then the Medium quality mode. We carried out a well defined movement and measured performance with Fraps.

Hold the mouse over the graph to classify cards by performance.

The Radeon HD 7700s do particularly well in this game and the Radeon HD 7770 outperforms the GeForce GTX 560.

Page 11
Benchmark: Batman Arkham City

Batman Arkham City

Batman Arkham City was developed with a recent version of Unreal Engine 3 which supports DirectX 11. Although this mode suffered a major bug in the original version of the game, a patch (1.1) has corrected this. We use the game benchmark.

All the options were pushed to a maximum, including tessellation which was pushed to extreme on part of the scenes tested. We measured performance in Extreme mode (which includes the additional DirectX 11 effects) and Very High.

Hold the mouse over the graph to classify cards by performance.

In Very High mode, the GeForces are the most effective, but the Radeon HD 7700s make up ground in Extreme mode.

Page 12
Benchmark: Battlefield 3

Battlefield 3

Battlefield 3 runs on Frosbite 2, probably the most advanced graphics engine currently on the market. A deferred rendering engine, it supports tessellation and calculates lighting via a compute shader.

We tested High and Normal modes and measured performance with Fraps, on a well-defined route. Note that a patch designed to improve performance on the Radeon HD 7000s came out on the 14th February. Naturally we installed it and noted a gain of between 1 and 2%.

Hold the mouse over the graph to classify cards by performance.

Once overclocked, as the XFX is, the Radeon HD 7770 easily outperforms the Radeon HD 6850 and the GeForce GTX 460 in Battlefield 3.

Page 13
Benchmark: Bulletstorm


Although only in DirectX 9 mode, the rendering is pretty nice, based on version 3.5 of Unreal Engine.

All the graphics options were pushed to a max (high) and we measured performance with Fraps.

Hold the mouse over the graph to classify cards by performance.

With antialiasing on, the Radeon HD 7700s suffer a little more than the other Radeons .

Page 14
Benchmark: Civilization V

Civilization V

Pretty successful visually, Civilization V uses DirectX 11 to improve quality and optimise performance in the rendering of terrains thanks to tessellation and to implement a special compression of textures thanks to the compute shaders. This compression allows it to keep the scenes of all the leaders in the memory. This second usage of DirectX 11 doesnít concern us here however as we used the benchmark included on a game card. We zoomed in slightly so as to reduce the CPU limitation which has a strong impact in this game.

All settings were pushed to a max and we measured performance with shadows and reflections. The latest patch was installed.

Hold the mouse over the graph to class cards by performance.

The Radeon HD 7700s improve on the rather poor showing of the Radeon HD 6000s in this game.

Page 15
Benchmark: Crysis 2

Crysis 2

Crysis 2 uses a development of the Crysis Warhead engine optimised for efficiency but adds DirectX 11 support via a patch and this can be quite demanding. As, for example, with tessellation, implemented abusively in collaboration with NVIDIA with the aim of causing Radeon performance to plummet. We have already exposed this issue here.

We measured performance with Fraps on version 1.9 of the game.

Hold the mouse over the graph to class cards by performance.

In High mode, the GeForces are the most efficient while the Radeons make up ground in Extreme mode.

Page 16
Benchmark: F1 2011

F1 2011

The latest Codemaster title, F1 2011 uses a slight development of the F1 2010 and DiRT 3 engine, which retains DirectX 11 support.

We pushed all the graphics options to a max and we used the gameís own test tool on the Spa-Francorchamps circuit with a single F1.

Hold the mouse over the graph to classify cards by performance.

The Radeon HD 7750 is on a par with the Radeon HD 5770 as well as with the GeForce GTX 550 Ti.

Page 17
Benchmark: Metro 2033

Metro 2033
Still one of the most demanding titles, Metro 2033 forces all recent graphics cards to their knees. It supports GPU PhysX but only for the generation of particles during impacts, a rather discreet effect that we therefore didnít activate during the tests. In DirectX 11 mode, performance is identical to DirectX 10 mode but with two additional options: tessellation for characters and a very advanced, very demanding depth of field feature.

We tested it in DirectX 11, at very high quality with tessellation on and in normal mode without tessellation.

Hold the mouse over the graph to classify cards by performance.

Itís difficult to play Metro 2033 with a high level of detail on these cards which therefore have to make do with Normal mode at which the GeForces perform best.

Page 18
Benchmark: Total War Shogun 2

Total War Shogun 2

Total War Shogun 2 has a DirectX 11 patch, developed in collaboration with AMD. Among other things, it gives tessellation support and a higher quality depth of field effect.

We tested it in DirectX 11 mode, at max quality, with the gameís "balanced" mode.

Hold the mouse over the graph to classify cards by performance.

The GeForces lag here as does the Radeon HD 7750 which is behind the Radeon HD 5750.

Page 19
Performance recap

Performance recap
Although individual game results are obviously worth looking at when you want to gauge performance in a specific game, we have also calculated a performance index based on all tests with the same weight for each game. We set an index of 100 to the Radeon HD 5770 / 6770:

Hold the mouse over the graph to classify cards by performance.

The Radeon HD 7750 is just a little behind the Radeon HD 5770 and the GeForce GTX 550 Ti. It gives a 15% gain on the Radeon HD 5750, while the Radeon HD 7770, which is a good deal faster, has a 21% advantage over the Radeon HD 5770.

The XFX Radeon HD 7770 Black Super OC benefits from GPU and memory clocks that have been increased by 12 and 16% respectively to overtake the Radeon HD 6850 and the GeForce GTX 460 1 GB.

Page 20
Gains with GPU and memory overclocking

Performance with overclocking
We wanted to observe the gains we could get with a 20% GPU overclocking on the Radeon HD 7770, followed by a 20% memory overclocking and finally with both the memory and the GPU overclocked by 20%.

We also included the results for the Radeon HD 6850.

These results were obtained at 1920x1080 at the very high quality mode:

Hold the mouse over the graph to show results in fps.

The 20% GPU overclocking is translated into a 5 to 15% gain in games and the memory overclocking into a more modest gain of 3 to 11%.

With both the GPU and memory overclocked, we observed gains of between 16 and 20% with an average gain of 18%.

Page 21

The arrival of the Cape Verde GPU and the Radeon HD 7700s offers AMD the opportunity to make its new architecture more readily available along with all the latest technologies that come with it: PCI Express 3.0, DirectX 11.1, OpenCL 1.2, HDMI 1.4a 3 GHzÖ

Moreover, the 28nm fabrication process has significantly improved the energy yield and it has to be said that Cape Verde does very well here, particularly with the Radeon HD 7750. In addition to the benefits of the finer engraving, AMD has also introduced ZeroCore Power, which brings the energy consumption of the card with the screen on standby down to a ridiculously low level.

While all this is positive, we arenít wholly convinced by these Radeons. There are lots of models in the mid-range segment and we were hoping that AMD would shake it up more and offer us more bang for our buck. Unfortunately, this isnít the case with the current pricing levels. To position its Radeon HD 7700s, AMD seems to have focussed on the GTX 550 Ti, the only recent GeForce to be released in the Ä100-Ä150 segment. As we said when it came onto the market, Nvidia didnít try to make this card competitive but rather looked to take advantage of its high-end naming (GTX Ti) to make as much from it as possible.

Positioning products alongside it is therefore hardly a way of giving value for money. Nevertheless, this is what AMD seems to have decided to do, at the same time completely ignoring the positioning of its own products already on the market, such as the Radeon HD 5700s/6700s and HD 6850.

More specifically, the Radeon HD 7770 offers a decent gain of over 20% on the Radeon HD 6770/5770, but at 40% higher cost. In fact the HD 7770 costs more than the Radeon HD 6850 in spite of giving a lower level of performance. The Radeon HD 7750 settles for getting close to the Radeon HD 6770/5770 for an identical price. Worse still, its reference design doesnít take advantage of its very low energy consumption to reduce noise levels sufficiently. At 7W at idle, a graphics card shouldn't have any noise issues at all.

Partner cards, such as the XFX Radeon HD 7770 Black Super OC Double Dissipation, suffer even more when you compare their pricing with already existing cards as, price wise, they're competing in the next segment up, against the Radeon HD 6870s and GeForce GTX 560s, but aren't able to compete in terms of performance.

While what are some very good products do have potential, AMD and its partners will need to review the current positioning to give more value for money. In the meantime, the older generation Radeons remain a better bet for most gamers. Itís time for Nvidia to come out of its extended hibernation and bring some competition back to the market!

Copyright © 1997-2015 BeHardware. All rights reserved.