by Ilya Gavrichenkov
08/20/2013 | 10:23 AM
All the impressive innovations AMD has unveiled this year on the processor market are related to its new Jaguar microarchitecture in one way or another. Combined with GCN-based graphics cores, the Jaguar has made a breakthrough on the game console market and given birth to energy-efficient mobile APUs of the Kabini and Temash series which are targeted at tablet PCs and ultra-compact notebooks. However, the Jaguar microarchitecture is focused on low power consumption, so Jaguar-based APUs are not very fast and can hardly suit even midrange notebooks. In other words, the Jaguar is in no way a replacement to the Piledriver microarchitecture we find in the majority of desktop and a lot of mobile APUs from AMD. That’s why the wholesale transition of the model range to the completely new APU design has not affected AMD's top-end APU products. The company has been preparing the next generation of the Bulldozer design, codenamed Steamroller, for that market segment but we have to wait for Steemroller-based APUs yet. The release of such APUs with GCN graphics cores (known under the codename of Kaveri) is scheduled for the end of 2013 or early 2014.
Still, AMD’s current strategy implies annual updates of mobile platforms, so the Trinity series APUs have not been left unaffected by the changes. AMD has rebranded them, producing the new name "Richland" which denotes barely changed APUs made of Piledriver x86 cores and VLIW5 graphics cores. Although the Richland is identical to the Trinity internally, the improved 32nm tech process and the retuned Turbo Core technology have helped the manufacturer increase the clock rates of the new models while keeping them within the older TDP limits. So, the Richland is indeed a step forwards in an end-user's eyes, especially as the lack of architectural innovations is partially made up for by certain software optimizations.
We have been aware of these intricacies of AMD’s marketing policy for a few months already as the new mobile platform and its components were introduced by the company back in March. And now it is the desktop processors' turn. The Trinity-to-Richland transformation couldn’t but provoke similar changes in the desktop product range where APUs play an important role, forming the Socket FM2 infrastructure. So, a couple of weeks ago AMD released a few new Socket FM2 Richland-based APUs in addition to the Trinity-based series.
There are fewer innovations in desktop Richland-based APUs than in their mobile counterparts. They just have higher clock rates and a more aggressive setup of the Turbo Core technology. Still, AMD thought it appropriate to assign a new series of model numbers to the new APUs and their integrated graphics, which looks like a rather nasty marketing trick to us. Well, we may be wrong about the desktop Richland-based processors which may actually be much faster than we expect. To check this out, we are going to carry out a comprehensive testing of the entire desktop Richland-based APU series. In our today’s review, we will test their computing performance. The performance of the Richland's integrated graphics will be discussed in an upcoming review.
AMD’s high-performance hybrid processors for desktop PCs form a separate and closed ecosystem. It was originally the Socket FM1 platform with Llano series APUs. Then there was the Socket FM2 platform with Trinity APUs that featured the more progressive Piledriver microarchitecture and faster VLIW4 graphics cores. The 2013 model year APUs of the Richland generation do not bring about any dramatic improvements over their predecessors but the good side is that the platform remains stable. Older mainboards for Trinity APUs with A85X, A75 and A55 chipsets are perfectly compatible with the Richland series after a BIOS update.
AMD has successfully kept its promise that Socket FM2 is a long-lasting design but the upcoming Kaveri series APUs, built out of computing and graphics cores with new architecture and supporting a shared address space for all types of cores, will call for a new socket and, consequently, for new chipsets. In other words, the Kaveri APUs won’t be compatible with the existing Socket FM2 mainboards, putting an end to the lifecycle of that platform.
The Richland doesn’t change anything in the Socket FM2 ecosystem, just as expected. Although AMD manufactures Kabini and Temash processors on 28nm technology, the Richland is based on a 32nm semiconductor die. The die itself is exactly like a Trinity die: 246 sq. mm and 1.3 billion transistors. That’s a clear indication of the lack of any design modifications.
Anyway, like their mobile counterparts, the new desktop Richland-based APUs are somewhat optimized in terms of the Turbo Core technology. Additional sensors installed on the APU die provide more accurate temperature data while the frequency adjustment algorithm has been modified to support more intermediary states. As a result, the clock rate changes more responsively depending on the APU's temperature while the APU's cores spend more time working in the Turbo mode.
It is questionable whether such changes are enough to regard the Richmond as a new APU generation, yet AMD has introduced a new series of model numbers for its new APU products. Although the Richland processors are identical to their Trinity counterparts and use Piledriver x86 cores with VLIW4 graphics cores, their model numbers are higher by 1000 and their graphics core is referred to as Radeon HD 8000D.
Comparing the specifications, the Richland APUs are about 300 MHz faster than their Trinity cousins in terms of the clock rates of their x86 cores. The graphics core frequency is higher by 40 to 84 MHz depending on the specific APU model. The number of Piledriver modules, the amount of cache memory and the number of shader processors in the graphics core have remained the same. The older Trinity-based A10, A8 and A6 APUs of the 5000 series and their 6000 series successors are very similar indeed. They even have the same overall structure of the model range which splits into two groups. The first group includes overclocker-friendly products with a TDP of 100 watts which have maximum x86 and graphics core frequencies and allow to overclock via the frequency multiplier. The second group includes 65-watt products which are somewhat slower but let you build rather economical computers. The Richland APUs are somewhat more expensive but they are supposed to add to rather than to replace the earlier Trinity APUs.
As a result, the new model names may be confusing, especially when it comes to the graphics core indexes. AMD doesn’t seem to be disturbed by the fact that its discrete graphics cores of the 7000 and 8000 series feature the GCN architecture whereas the graphics cores of the Trinity and Richland APUs are related to the Radeon HD 6900 in its design. As for the processors themselves, it is possible to use their model numbers within the A10, A8 and A6 series for making comparisons. Within each series, Richland-based APUs offer higher clock rates, so their higher model numbers do mean somewhat higher performance.
The performance benefits are not really high, though. They are going to amount to 5 to 7% in the majority of applications. Considering this, AMD has tried to ensure more speed by overclocking the memory controller, so the senior model A10-6800K is officially compatible with DDR3-2133 SDRAM. Well, this type of memory was supported by senior Trinity-based APU models as well, even though not officially, so it is hardly a serious innovation. Moreover, the rest of the Richland-based APUs do not support such fast system memory.
Having the full range of Richland-based APUs at our disposal, we will of course compare them with Trinity-based APUs in the first place. It's not so simple with finding appropriate opponents from Intel, though. AMD positions its APUs in the following way:
However, as we know from our earlier tests of the Socket FM2 platform, even top-end A10 series APUs are inferior to the Core i5 in computing performance. The official prices are indicative of this fact, too.
That’s why Intel will be represented by most of the Core i3 series models plus a Pentium and a junior Core i5. Since the Haswell microarchitecture has been announced recently, the Core i5 is the most up-to-date LGA1150 variant.
We don’t examine the 3D performance of integrated graphics cores in this review, so we also include a Socket AM3+ AMD FX-4350 processor. It lacks any integrated graphics and is tested together with a discrete graphics card Radeon HD 6670.
The other configurations are tested with their integrated graphics with one exception only: we install a discrete card Nvidia GeForce GTX 680 into them to check out the processor’s capabilities in today’s 3D games.
Fast DDR3 memory becoming more and more popular, we’ve decided to switch to DDR3-2133 SDRAM in our test configurations. Such modules are not promoted as overclocker-friendly anymore, so they come at no extra cost while most of today’s platforms can work with such memory without any limitations. As we see, the CPU developers begin to officially support such memory as well. We mean the A10-6800K which is specified to support DDR3-2133. As for the other processors in this review, the AMD A10-6700 and AMD A8-6500 are the only ones to be incompatible with DDR3-2133 SDRAM. We lower the memory frequency to DDR3-1866 for them.
As a result, our testbeds were built with the following hardware and software components:
As usual, we use Bapco SYSmark 2012 suite to estimate the processor performance in general-purpose tasks. It emulates the usage models in popular office and digital content creation and processing applications. The idea behind this test is fairly simple: it produces a single score characterizing the average computer performance. After the launch of Windows 8 SYSmark 2012 got updated to version 1.5, and this is exactly the version e are using in our test session.
As expected, the Richland-based CPUs are not much faster than their Trinity-based predecessors. The average difference is about 6 to 7% and depends directly on their clock rate. The same can be inferred from AMD’s price policy: the new models of the A10, A8 and A6 series are more expensive by the same percentage. It means the market positioning of Socket FM2 APUs has not changed. Their computing performance is still rather unassuming in everyday applications in comparison with the competing products from Intel. In fact, the A10 and A8 APUs with two Piledriver modules can only compete with Intel’s Pentium series whereas the A6 APUs with only one Piledriver module are even slower than that.
Let’s take a closer look at the performance scores SYSmark 2012 generates in different usage scenarios. Office Productivity scenario emulates typical office tasks, such as text editing, electronic tables processing, email and Internet surfing. This scenario uses the following applications: ABBYY FineReader Pro 10.0, Adobe Acrobat Pro 9, Adobe Flash Player 10.1, Microsoft Excel 2010, Microsoft Internet Explorer 9, Microsoft Outlook 2010, Microsoft PowerPoint 2010, Microsoft Word 2010 and WinZip Pro 14.5.
Media Creation scenario emulates the creation of a video clip using previously taken digital images and videos. Here they use popular Adobe suites: Photoshop CS5 Extended, Premiere Pro CS5 and After Effects CS5.
Web Development is a scenario emulating web-site designing. It uses the following applications: Adobe Photoshop CS5 Extended, Adobe Premiere Pro CS5, Adobe Dreamweaver CS5, Mozilla Firefox 3.6.8 and Microsoft Internet Explorer 10.
Data/Financial Analysis scenario is devoted to statistical analysis and prediction of market trends performed in Microsoft Excel 2010.
3D Modeling scenario is fully dedicated to 3D objects and rendering of static and dynamic scenes using Adobe Photoshop CS5 Extended, Autodesk 3ds Max 2011, Autodesk AutoCAD 2011 and Google SketchUp Pro 8.
The last scenario called System Management creates backups and installs software and updates. It involves several different versions of Mozilla Firefox Installer and WinZip Pro 14.5.
The gap between same-class Richland and Trinity-based APUs remains stable irrespective of their usage scenario. The increased clock rate being the only source of computing performance improvements, it is natural that the 6000 series models are only 5 to 7% faster than their predecessors. That’s why the Richland series show the same weak spots as we noticed about Socket FM2 APUs earlier. For example, their computing performance is only competitive at multithreaded loads. Like any other processors based on the Piledriver microarchitecture, the Richland series are not efficient at single-threaded loads. As a result, the A10 and A8 APUs can only challenge the Intel Core i3 series in the 3D Modeling scenario since rendering tasks can be easily carried out in parallel on any number of execution cores. System Management is another example of such a scenario.
To test the processors performance during data archiving we resort to WinRAR 5.0 archiving utility. Using maximum compression rate we archive a folder with multiple files with 1.7 GB total size.
When it comes to data compression, AMD’s A10 and A8 APUs, irrespective of whether they belong to the second or third APU generation, fit in between the Core i3 and Pentium processors. The dual-core A6 APUs look very weak by today’s standards. The new Richland design can only ensure a 5% increase in performance thanks to higher clock rate.
To check out our configurations with Microsoft’s office applications, we use a special test script from Futuremark which simulates a typical user working in Word 2013, Excel 2013 and PowerPoint 2013.
Office applications are generally single-threaded ones, so they are not a good choice for highlighting the advantages of processors with Piledriver microarchitecture. It is no wonder then that most of the Socket FM2 APUs fall behind even the Pentium G2130. So again, the A10, A8 and, especially, A6 are not optimal for office work.
We use Futuremark’s Peacekeeper browser benchmark as an online application test. It makes use of cutting-edge and resource-consuming web technologies and runs in Google Chrome 27.
We can see the same picture as in the office applications. Unfortunately, AMD won’t have any solutions with good x86 computing performance under single-threaded loads until the arrival of Kaveri-based APUs. The situation may only be improved by the upcoming Steamroller microarchitecture.
We benchmark CPUs in Adobe Photoshop CS6 using our custom test that is based on the Retouch Artists Photoshop Speed Test and consists of typical processing of four 24-megapixel images captured with a digital camera.
AMD is no better in the Photoshop test, either. Each of the company’s quad-core APUs is slower than Intel’s dual-core solutions. The Richland doesn’t change anything except for the minor increase in clock rate. Comparing the A10-5800K and the A10-6700, which work at similar clock rates, it is clear that the new APU doesn’t have any hidden advantages in terms of its x86 speed. The senior Socket FM2 processor A10-6800K is only as good as the AMD FX-4350, which is not enough to compete with the LGA1155 configurations.
We use Xilisoft Audio Converter 6.5 utility to test audio transcoding speed into mp3. During this test we transcode the audio album saved in flac format.
Now we’ve come to applications that can use four x86 cores in parallel, so the standings are different. The A10-6800K is ahead of the Core i3-3240 whereas the A10-6700 and A8-6600K are comparable to midrange Core i3 models. So, AMD’s quad-core APUs are equal to Intel’s dual-core CPUs with Hyper-Threading technology when it comes to handling the multithreaded load of transcoding audio files. As for the Core i5, none of the Socket FM2 APUs can get anywhere near its performance, so the illustration of the Richland’s positioning you could see above (where the A10 series is opposed to the Core i5) is just AMD’s wishful thinking.
In order to measure how fast our testing participants can transcode a video into H.264 format we used x264 FHD Benchmark 1.0.1 (64 bit). It works with an original MPEG-4/AVC video recorded in 1920x1080@50fps resolution with 30 Mbps bitrate and measures the time it takes x264 r2334 coder to convert the video. I have to say that the results of this test are of great practical value, because the x264 codec is also part of numerous popular transcoding utilities, such as HandBrake, MeGUI, VirtualDub, etc.
Here, AMD’s processors look good, too. Although they are inferior to the Core i5, each of the quad-core Socket FM2 APUs beats the Core i3 series, which is quite an achievement considering the other test results. The difference between the A10-6800K and A10-5800K and between the A8-6600K and A8-5600K is 7 to 8%, just as expected.
We use special Cinebench 11.5 benchmark to test final rendering speed in Maxon Cinema 4D suite.
Although graphics rendering, like media content transcoding, is a computing task that can be easily carried out on multiple execution cores, AMD’s quad-core APUs with Piledriver microarchitecture are only comparable to the Intel Core i3 series in Cinebench, and only in their senior modifications. For example, the A8-6500 and A8-5600K are slower than the Core i3-3210 whereas the dual-core A6 series APUs are more than 1.5 times inferior to the Pentium.
The next diagram shows one of the intermediate results of the Futuremark 3DMark11 Fire Strike – Physics Score benchmark. This parameter shows how fast the testing participants can cope with a special physics test emulating the behavior of a complex system with a large number of objects.
The 3DMark Fire Strike physics test is multithreaded, so it produces typical results. AMD’s A10 and A8 series are comparable to the Core i3 in their x86 computing performance whereas the A6 APUs with one dual-core Piledriver module are too slow by today’s standards. This refers to both the Richland and the Trinity designs which differ by a mere 9% here. We can note, however, that the senior Richland-based APUs can compete with quad-core Socket AM3+ processors. The A10 6000 series APUs are both ahead of the FX-4350 that is similar to them in microarchitecture but has an additional L3 cache.
Even when we benchmark their conventional computing performance, we shouldn’t forget that AMD’s Socket FM2 processors are hybrid devices that combine x86 and graphics cores. The Devastator graphics core employed in the Richland and Trinity series supports the OpenCL framework which allows using its shader pipelines for general computing. The use of heterogeneous resources for a single task is the key idea of the APU concept as promoted by AMD. The company wants to influence the software market so that OpenCL were used everywhere.
There has been some progress in this area indeed. The number of software applications that can utilize the graphics core’s computing resources is on the rise, so AMD can proudly post a rather long list of OpenCL-compatible programs.
Unfortunately, not all of these programs offer full support. Some of them only use the graphics core for certain tasks, which is due to the specifics of the graphics core design. It is only with simple parallel operations that a graphics core can really be efficient, so the APU concept can’t really push the performance bar of AMD’s hybrid processors to a whole new level across a variety of applications. On the other hand, there are quite a lot of scenarios, such as image or video processing, where the graphics core can be of huge help to the conventional x86 cores.
Ideally, we wouldn’t use special tests to benchmark OpenCL performance. It would be better if the popular applications we have on our ordinary test program supported OpenCL and heterogeneous computing. But since they do not, we have to check out such processors in specific tasks.
The first of them is Luxmark 2.0, which is based on the LuxRender rendering engine. We use the average-complexity Sala scene, rendering it on both graphics and x86 cores of the tested processor.
As you can see, things hardly get different when we make use of the graphics cores’ computing resources. Like AMD’s APUs, modern Intel processors offer full support for OpenCL. So they remain in the lead, even though their graphics core is less advanced. The superior integrated graphics of the Richland and Trinity processors doesn’t help them win even at heterogeneous loads. It is the Core i5-4430 with Intel HD Graphics 4600 and the Core i3-3225 with Intel HD Graphics 4000 that occupy the top places. The AMD FX-4350 is on top, too, but mostly due to our using it together with a discrete graphics card Radeon HD 6670.
Like before, the A10 and A8 series APUs are only comparable to the Core i3 series whereas the A6 series is still inferior to the Intel Pentium. And again, we don’t see any big difference between the Richland and the Trinity: the A10-6800K is a mere 4% ahead of the A10-5800K.
The introduction of OpenCL support into the popular archiver WinZIP shows that the APU concept is embraced by the software market, so we can’t help testing our processors in WinZIP 17.5. We compress a folder with files with a total size of 2.53 GB.
Besides their OpenCL compatibility, the latest versions of WinZIP work well on multicore processors, so the benefits of a specific graphics core are masked by the conventional x86 performance. In other words, WinZIP doesn’t differ much from OpenCL-incompatible archivers despite its using graphics core resources.
Image-editing and video content processing applications are a more traditional type of software that support OpenCL acceleration. Corel AfterShot Pro, a popular tool for batch-processing of digital photos, is an example of that. We use a scenario in which two hundred 12-megapixel RAW-format images are post-processed and exported as JPEG files.
AMD’s hybrid processors seem to be more efficient here. The A10 and A8 models beat the Core i3 series irrespective of what graphics cores they have and are very close to the junior Core i5. The AMD A6 series is brilliant as well, outperforming the Intel Pentium.
Another example of a popular OpenCL-compatible application is the professional video editing tool Sony Vegas Pro 12. When rendering video, it can distribute the load among all the computing resources of hybrid processors.
Intel CPUs with HD Graphics 4600 and HD Graphics 4000 can offer better performance than the various Richland and Trinity products from AMD. So, promoting the heterogeneous computing concept, AMD actually helps its archrival as well because Intel also implements OpenCL in its products and equips them with fast graphics cores. As a result, senior Socket FM2 solutions don’t look superior even if we use APU-optimized applications with OpenCL support. The Richland is only 5% faster than the Trinity, so we don’t see any breakthroughs here. Unfortunately, the modern GCN graphics architecture is not implemented in the Richland.
The second OpenCL benchmark we used was SVPMark 3. It is a specialized performance benchmark for the SmoothVideo Project software which improves video playback smoothness by inserting new intermediary frames into the video stream. This software makes active use of GPU resources via OpenCL.
The Richland-based APUs fail to compete with the junior Core i5 but look good enough in comparison with the Core i3 that have the older Ivy Bridge microarchitecture. Even the Core i3-3225 with HD Graphics 4000 falls behind all of the A10 and A8 series APUs. The A6-6400K beats not only the Pentium but also the Core i3 models with the junior version of the integrated graphics core.
So we can see that OpenCL optimizations can produce miraculous results in some situations, but there are too few such examples. AMD’s much-touted concept of heterogeneous computing on Socket FM2 processors doesn’t look like a killer feature. We’ve never seen A10 series APUs beat the junior Core i5 even under the most favorable conditions. The desktop Richland-based products can only compete with Intel’s Core i3 series.
We’re going to discuss the 3D performance of the integrated graphics cores in our next review, so today we focus on computing performance. That’s why we want to benchmark the Richland-based APUs together with a top-end discrete graphics card Nvidia GeForce GTX 680.
First we run the synthetic benchmark 3DMark Fire Strike.
The diagram doesn’t show anything good for the Socket FM2 platform. According to 3DMark Fire Strike, the Richland-based APUs, including the top-end A10 and A8 models, are slower than the junior Core i3 model. The transition to the new APU design only provides a 3% increase in performance in this benchmark.
As you know, it is the graphics subsystem that determines the performance of the entire platform equipped with pretty high-speed processors in the majority of contemporary games. Therefore, we select the most CPU-dependent games and take the fps readings twice. The first test run is performed without antialiasing and in far not the highest screen resolutions. These settings allow us to determine how well the processors can cope with the gaming loads in general and how the tested CPUs will behave in the nearest future, when new faster graphics card models will be widely available. The second pass is performed with more real-life settings – in FullHD resolution and maximum FSAA settings. In our opinion, these results are less interesting, but they demonstrate clearly the level of performance we can expect from contemporary processors today.
We can see the same picture in every game. The Socket FM2 configurations with discrete graphics card are generally slower in games than any Core i3 configurations. This also refers to the Athlon X4 760K, which is based on the Richland design. AMD’s new APU generation is a mere 5% faster than their predecessors, which is not enough for a real breakthrough. The A6-6400K is especially poor in these tests: its single Piledriver module can’t cope with modern games at all.
The Richland series are faster than the Trinity thanks to a small increase in clock rates but there are no changes on the microarchitecture level or in their manufacturing technology, so we have some apprehensions about their power draw, especially as the Trinity series were not very economical in comparison with their Intel opponents. AMD promises that the new APUs have the same power consumption, specifying their TDP at 65 and 100 watts.
To find out more about the power consumption and heat dissipation, we performed a round of special tests. The new digital power supply unit from Corsair – AX760i – allows monitoring consumed and produced electrical power, which we use actively during our power consumption tests. The graphs below (unless specified otherwise) show the full power draw of the computer (without the monitor) measured after the power supply. It is the total power consumption of all the system components. The PSU's efficiency is not taken into account. The CPUs are loaded by running the 64-bit version of LinX 0.6.4 utility with FMA instructions support (for AMD processors) and AVX instructions support (for Intel processors). Moreover, we enabled Turbo mode and all power-saving technologies to correctly measure computer's power draw in idle mode: C1E, C6, Enhanced Intel SpeedStep and AMD Cool’n’Quiet.
Intel’s LGA1155 and LGA1150 platforms are the most economical in idle mode. The Socket FM2 platform with AMD APUs needs a few watts more. There are no changes between the Trinity and Richland designs but we can note that the Richland APUs drop their clock rate to 2.0 GHz instead of 1.4 GHz when idle.
It is at full load on the x86 cores that the lower energy efficiency of the Socket FM2 platform becomes apparent. Even AMD’s 65W APUs need about 50 watts more than similar configurations with Core i3 processors that deliver the same performance. The 100W A10 and A8 series APUs are even less economical. Thus, AMD’s Socket FM2 products cannot compete with Intel CPUs in terms of computing performance per watt. Hopefully, AMD’s upcoming Kaveri design and 28nm tech process will improve this situation because the Richland is no better than its predecessor in this respect.
To show you the full picture, we carry out another test, loading both the x86 and graphics cores. We use Luxmark 2.0 in CPU+GPU mode for that.
The overall picture remains the same but we should note one thing about it. AMD’s 100W APUs require about 15 watts more under heterogeneous load than at the maximum load on their x86 cores. The 65W APUs, on the contrary, consume less. This behavior is explained by the fact that the higher temperature of the integrated graphics core turns off the Turbo technology to keep the APUs within the specified thermal thresholds. That’s why the performance of the 100W and 65W APUs from AMD at heterogeneous loads differs more than in ordinary applications.
The Trinity series was not very overclocker-friendly. Even though, like the new Richland APUs, they included special K-indexed models with unlocked frequency multipliers, their Piledriver modules already worked at high frequencies, over 4 GHz. With the same manufacturing process and semiconductor design, the Richland APUs are clocked at even higher frequencies. Considering that Socket FM2 processors of the Trinity generation used to speed up to 4.5 GHz with air cooling, we might expect the same overclockability from the Richland generation.
Well, the actual results are somewhat better than expected. After increasing its voltage to 1.5 volts and enabling Load-Line Calibration, we made our A10-6800K stable at 4.7 GHz. It was cooled by an air cooler NZXT Havik 140.
The clock rate of 4.7 GHz, even though definitely better than the results of the previous-generation processors, doesn’t look impressive at all, though. It only sets the frequency of the A10-6800K higher by 300 MHz in the Turbo mode, so there are but minor performance benefits compared to the default frequency.
It is the high temperature of the 32nm APU die that prevents us from getting better results. To ensure stability at high clock rates, it is necessary to step the voltage up quite aggressively. The temperature grows as the consequence, with a threshold at 120°C. Upon reaching that threshold, the protection mechanism is triggered. So, efficient cooling is most important for successful overclocking of Richland-based APUs.
Our second unlocked processor, A8-6600K, turned out to be identical to its senior cousin in terms of overclocking potential. It was stable at 4.7 GHz with a voltage of 1.5 volts.
Thus, our A10 and A8 series APUs have the same overclocking potential.
Our A6-6400K processor, based on a single Piledriver module, could be overclocked somewhat better. Being simpler in design, it doesn’t get as hot as the quad-core A10 and A8. So, after setting its voltage at 1.525 volts, we made it stable at 4.9 GHz.
Generally speaking, the Richland series overclocks in the same way as other modern products from AMD. Every processor with Piledriver microarchitecture has about the same frequency potential. With air cooling, they cannot reach 5 GHz or higher. On the other hand, the 32nm manufacturing process is steadily being optimized, pushing the frequency bar higher. The Richland is a good example, differing from the previous generation with increased clock rates in the first place. Similar things can be observed with the Socket AM3+ platform for which AMD has recently rolled out a couple of FX-9000 series models.
We are generally not very optimistic about the Richland-based APUs offered by AMD as an update to their Socket FM2 ecosystem. We just don’t see any significant innovations. The microarchitecture of the x86 cores and the integrated graphics core have been borrowed from the Trinity design without any changes, using the same Piledriver+VLIW4 formula. Most of the specs, such as the number of x86 cores, the number of shader processors, cache memory amount, etc., have remained intact, too. The Richland APUs are actually just slightly overclocked Trinity ones but with much higher model numbers.
According to our tests, the new 6000 series is a mere 6-7% faster than the corresponding 5000 series APUs on average. The maximum advantage of the Richland over the Trinity design can be observed at multithreaded loads. It amounts to 9-10% only. The new A10, A8 and A6 APU models do not change anything in the market positioning of the Socket FM2 platform, so this is all just a cosmetic facelift.
When it comes to x86 computing performance, the quad-core A10 and A8 series models can be viewed as competitors to the Core i3 series, but only at multithreaded loads. At single-threaded loads, the A10 and A8 slow down and fit in between the Core i3 and the Pentium. The dual-core A6 series, in its turn, looks absolutely uncompetitive against LGA1155 CPUs that cost the same money.
The support for OpenCL-based heterogeneous computing touted by AMD doesn’t change anything. As we’ve made sure in our tests, Intel CPUs are as compatible with OpenCL as their Richland and Trinity counterparts, so they ensure the same performance boost in applications which can use graphics cores for general computing purposes. So even in the most favorable situations for hybrid processors, the senior models of the AMD A10 series cannot match the speed of Intel’s junior quad-core CPUs but offer the same performance as Core i3 processors with HD Graphics 4000.
Summing everything up, we can say that the Richland-based APUs, like their predecessors, are not good for mainstream computers. They are more appropriate for entry-level configurations thanks to the good 3D performance of their integrated graphics core which may let you do without a discrete graphics card. Well, the Richland’s 3D performance is actually the topic of our next review!