It'd be easy to dismiss AMD's A10-7800 as just another APU after a quick look at the Kaveri-based chip's spec sheet. After all, the flagship A10-7850K, which sports an unlocked clock multiplier, has already been out for a while. And there are also several models at the lower end of the performance spectrum. So, why review the A10-7800? It sounds like an A10-7850K-light, and isn't even that much cheaper than the company's highest-end offering.
But the newest model emphasizes performance per watt, rather than trying to push just benchmark results, which put it on our radar. After setting aside all of AMD's marketing material associated with the -7800, we drilled down on the APU's main selling point, rather than writing the 1001st story filled with benchmarks about it. Is this the sweet spot of AMD's Kaveri design, making it the most efficient APU ever?
Our purpose is assessing the gaming and compute performance of the A10-7800, flanked by sophisticated power measurements.
This isn't going to be just another APU review. We'll underclock an A10-7850K and overclock the A10-7700K to -7800 levels. The experiments will be accompanied by detailed readings from expensive lab equipment and, in the end, we'll determine whether AMD’s A10-7800 deserves to assume that sweet-spot position.
But first, let’s look at a table of the four fastest Kaveri-based models:
| Products |
|
|
|
|
| Pricing |
|
|
|
|
| Cores/Threads | 4/4 | 4/4 | 4/4 | 4/4 |
| Clock (Base) | 3.7 GHz | 3.5 GHz | 3.4 GHz | 3.1 GHz |
| Maximum Turbo Core Frequency | 4.0 GHz | 3.9 GHz | 3.8 GHz | 3.8 GHz |
| L2 Cache | 4 MB | 4 MB | 4 MB | 4 MB |
| Graphics Engine | Radeon R7 | Radeon R7 | Radeon R7 | Radeon R7 |
| Compute Units | 8 | 8 | 6 | 6 |
| Shaders | 512 | 512 | 384 | 384 |
| Graphics Frequency | 720 MHz | 720 MHz | 720 MHz | 720 MHz |
| TDP | 95 W | 65 W | 95 W | 65 W |
| Street Price | $170 | $166 | $160 | $110 |
While the A8-7600 isn't going to be benchmarked today, it's included in the table for the sake of completeness.
We already covered the technical details of AMD’s Kaveri architecture in our launch story, AMD A10-7850K and A8-7600: Kaveri Gives Us a Taste Of HSA, so we'll just review the high-level points.
Kaveri’s x86 CPU cores are based on the Steamroller architecture, AMD’s most modern design, while the graphics engine employs Graphics Core Next, also a first for APUs. In addition, AMD revamped its Heterogeneous System Architecture to facilitate improved developer access to the platform's resources.
These APUs are manufactured on a 28 nm process at GlobalFoundries, and sport better efficiency than the Llano, Trinity, and Richland APUs preceding them. So far, so good.
Like the A10-7850K, the -7800 features a capable graphics unit with 512 GCN-based shaders. This newer model sports slightly slower base and Turbo Core clock rates, and doesn't include an unlocked multiplier.
This APU is highly complex. The 245 mm2 die hosts a whopping 2.41 billion transistors. But even though this is a fairly complete SoC, Kaveri requires a number of other platform components still.
We Need RAM, Fast RAM
It's not news that AMD's host and graphics processing subsystems benefit from fast system memory. The processor’s integrated memory controller supports DDR3-2133, and for our experiment, we went with AMD-branded memory rated for up to DDR3-2400. While we didn’t experience stability issues at 2400 MT/s, the performance increase over DDR3-2133 was marginal. Feel free to stick with the lower data rate and tighter timings.
A Simple Taste In Motherboards
Our German team is cycling through three different motherboards for APU testing, depending on form factor and focus. However, after extensive measurements, MSI's A88XM Gaming emerged as the most efficient. None of the APUs experienced stability issues with it, and its BIOS allows us to conveniently set underclocking and overclocking options to optimize the thermal ceiling.
A maximum of 8.2 W can be attributed to the motherboard under full load, which includes 16 GB of overclocked DDR3-2400 RAM. At idle, its power consumption drops to approximately 5 W. Needless to say, those values do not include the APU itself. Unsurprisingly, the motherboard derives the DIMMs’ supply voltage from the 12 V rails of the 24-pin connector.
While we adapted our graphics card test setup for the APU tests, the equipment didn't change. The HAMEG HZO 3054 (Rohde & Schwarz) is the core instrument, a fast four-channel DSO that can be remote-controlled via Ethernet and store up to 60,000 samples per channel.
We run the 12 V wires of the eight-pin (2 x four-pin) CPU power cable through a current probe, as shown in the left picture below, and also the 24-pin cable's 12 V wires (on the right). That makes four HAMEG HZO50 probes to measure the current without having to insert a series resistor into any cable. Simultaneously, each rail’s voltage is fed into a HAMEG HMC8012, which also has the deep storage and remote control options installed.
In order to tame the massive amount of data, we use a custom program and Excel. A measurement takes a full minute, and the sampling interval is 10 ms, which results in 6000 samples. Shrinking the sampling interval further wouldn't yield a tangible benefit, and would instead drown us in test data.
Unsurprisingly, at less than 100 W power draw, platform-oriented benchmarking isn't as wild as some of the graphics card-based results we've seen, and the motherboard doesn’t impose massive load spikes on the PSU. But there are still a few noteworthy observations.
As a preview to the following pages, let’s look at the motherboard's power draw in one second:

Test Setup and Test Equipment:
| Method | Contact-free DC measurement at PCIe slot (using a riser card) Contact-free DC measurement at external auxiliary power supply cable Mass-free Voltage measurement at external auxiliary power supply cable |
|---|---|
| Equipment | 1 x HAMEG HMO 3054, 500 MHz digital multi-channel oscilloscope 4 x HAMEG HZO50 current probes 3 x HAMEG HZ355 (10:1 probes, 500 MHz) 1 x HAMEG HMC 8012 digital multimeter with real-time storage function |
| Test System | MSI A88XM Socket FM2+ AMD Radeon Memory Corsair H100i Closed-Loop Water Cooler Corsair Neutron 480 GB SSD SeaSonic X-Series PSU |
For many years, I was the lead programmer for custom business sector-specific applications, and in lieu of synthetic benchmarks, which may not be relevant to real-world applications, I created three tests using software from that world.
|
|
|
|
|
| Read the Review | ||||
|
|
|
|
|
|
One Thread Solving a Complex Problem
The first program optimizes building a complex brick wall. A 60-foot-long stretch contains eight windows and two door openings, including the frames and jambs. Small holes for the spools of horizontal blinds and three interlocking walls also have to be placed. The goals are minimizing the number of split bricks and optimizing re-use of split bricks so that waste is almost eliminated. Since placement of bricks depends on where other bricks are located, the task can't be parallelized easily. So, it runs in a single thread, while memory use is negligible.

Intel's Core i3 wins this benchmark comfortably, delivering performance that's 20% faster than AMD's A10-7850K. However, since we are only talking about a three-second run time, the absolute difference isn't significant. Then again, we've seen similar scaling in our single-threaded iTunes and LAME benchmarks, too.
4 Threads = 4 Jobs?
The next application optimizes solar panel placement by considering the sun's position throughout the day, each of the 365 days in a year, from sunrise to sunset, in one-hour intervals. Two trees, a neighbor’s house, and a few chimneys create shadows, and in some light conditions, front-row solar panels throw shadows on second-row panels as well. This program also optimizes the placement of wiring. Furthermore, based on historical meteorological data, the expected energy output for the whole year is estimated. This software can be easily parallelized, as the energy output corresponding to each sun position is calculated independently.

While the Core i3-4330 is still in front, it’s a closer race since Hyper-Threading technology doesn't quite match the effectiveness of four integer units. The Haswell-based Intel CPU is barely faster than the A10-7850K, while beating the lower-clocked A10-7800 by 6%. The older Core i3-2100, which we included for comparison purposes, clearly shows its age.
We further increase the degree of parallelization by running a photo-realistic renderer on several computers on a network. For instance, all of a company's office PCs can be harnessed for a computation-intensive task like rendering. One PC serves as the controller, which farms out jobs to other PCs based on their hardware capabilities. In this test, each of the compute clients runs four worker threads.
The different CPU and APU models are even closer together due to communication overhead and increased memory footprint. The A10-7850K manages a win against the Core i3-4330, and the A10-7800 finishes a close third. What do we learn from this benchmark? Real-world applications involve more moving parts than video transcoding or gaming. While IPC is naturally an important consideration, it's not all-telling.
A Typical Consumer Application: Video Compression
When an application doesn't support OpenCL, or OpenCL acceleration is disabled, the Kaveri architecture's two modules address up to four threads in parallel. We decided to use HandBrake as a benchmark to test this. As expected, the A10-7800 winds up in the middle of our test field.

While we don’t want to rely on synthetic benchmarks, we'll use a couple to illustrate the potential gains of OpenCL and AMD's HSA initiative as those efforts start to take hold.
LuxMark 2.0
In order to assess the A10-7800's capabilities, we run three separate tests: CPU-only, GPU-only, and both combined. While we might have expected the more resource-laden APUs to win, Intel's Core i3-4330 and its superior IPC still puts in a strong showing. Let’s start with just the CPU:

We naturally expect the Haswell design to fare well in a CPU-only measure of performance. But while the win is unsurprising, the magnitude of Intel's advantage is fairly overwhelming. Favor is shifted to AMD by only using the on-chip graphics engine:

Intel's HD Graphics implementation is outclassed by Graphics Core Next in the Kaveri design. But what happens when CPU and GPU resources are utilized simultaneously?

Just as Bill Murray likes to photobomb pictures, the Core i3-4330 sneaks right in the middle of AMD’s family photo. One of the reasons appears to be that the Core i3's CPU and GPU scores are added together, yielding an aggregate, while the APUs are weighed differently. Perhaps one on-chip complex or the other isn't running at peak performance under a combined full load.
HSA: A Great Idea in Need of Software Support
Leading up to the Kaveri APU introduction, AMD put a lot of effort into evangelizing the benefits of its HSA (Heterogeneous System Architecture) initiative. Again, if you want to know more, read our Kaveri launch article. However, even now, several months later, the number of applications exploiting HSA remains small. That's a disappointing state of affairs, since the fundamentals of HSA are enticing.
Moving on from theory to practice, the LibreOffice benchmark originally provided by AMD impressively documents the value proposition of HSA. Once again, we conduct three separate test runs: CPU-only, OpenCL-only, and HSA. Since Intel CPUs do not support HSA, they aren't represented in the third chart. And we didn’t bother benchmarking the outdated Core i3-2100 at all.
Let’s start with the CPU-only chart, which, as we might expect, the Core i3-4330 dominates:

Thirty-seven percent higher performance. Such an advantage cries for an OpenCL-based rematch. Said and done. Surprisingly, Intel’s relatively modest graphics engine is slower than its CPU. Or perhaps the company's drivers are to blame.

It'd be easy to guess that a seeded test like this one would favor AMD's hardware. And it does, demonstrating what a highly parallelized workload can do on hardware with the necessary support. Those gray bars are for comparison only. They depict the execution times of Intel's Core i3-4330 and AMD's A10-7800 in software and OpenCL modes.

Looking at the HSA-based results after the numbers generated with OpenCL turned on reminds us of a quote attributed to General LeMay. When asked about the purpose of a nuclear second strike capability, he replied, "To make the cinders dance". However, not every application is suitable for HSA optimizations, and adoption of the initiative by the software industry is thus far disappointing.
We want to test AMD's APU as it sits on the motherboard. I see no point in buying a processor that emphasizes on-die graphics and then adding a Radeon R7 265X, even in CrossFire. Such a configuration makes little sense from the cost and technical angles. Yes, AMD officially recommends it and yes, we tried it out. But the much faster discrete card is a mismatch for the on-die engine. We even experienced detrimental effects like stuttering from this odd couple. If you want to go the add-in route, take a look at the Athlon X4 750K (or Pentium G3258) instead.
Let's instead stick to a setup better suited to the strengths of an APU: a more highly integrated all-in-one system with as few external components as possible.
Metro: Last Light
Both upper-end APUs will run at 1080p in this game, but they barely manage to average over 30 FPS. You're better off at 720p, where even the A10-7700K generates playable performance.
Battlefield 4
Battlefield 4 is another AAA title that can't be ignored. In single-player mode, however, where the graphics subsystem is emphasized, the APUs have a tough time maintaining adequate frame rates. You really need to run this one at 720p. Fortunately, the graphics look better than they sound.
BioShock Infinite
Since the third game in the BioShock franchise is a thinly disguised console port, its frame rates are adequate to good at 1080p.
The A10-7800 lets you play older games at medium quality settings and newer titles at lower details without discrete graphics. If you run into frame rate issues, consider lower resolutions. In other words, a Kaveri-based APU by itself is suitable for low-end gaming PCs. Since AMD's processors also drive both of the latest game consoles, we don’t expect this situation to change any time soon. We didn’t mention Mantle because most games don’t use Mantle.
We collect tons of data from each test run, which we then analyze in different ways. Today, we're going to present the power measurements at idle, under a gaming load, and under the most stress possible. Then, we'll dig into more specific details. The APU's power consumption is measured at the eight-pin 12 V connector. Simultaneously, we get the motherboard and memory draw at the 24-pin ATX connector. Total consumption is the sum of all power rails leading to the motherboard, without the SSD.
AMD promised us an efficiency surprise and, spoiler alert, it delivers. What does that achievement look like in detail, though?
Power Draw at Idle
MSI's motherboard and AMD's APU form an exciting low-power team. At a mere 6.4 W for the APU and 11.7 W for the total system, the platform's draw is impressively low.
While the 19 W measured at the wall socket is quite low as well, PSUs with an 80 PLUS Gold rating fail to hit our efficiency expectations at such a low power draw, even if they still technically comply with the specification. Plug in a DC-DC converter (which we can no longer call a picoPSU for legal reasons) and an efficient wall wart-style PSU; you can achieve less than 15 W at the outlet.
| Power Draw Idle | Minimum | Maximum | Average |
|---|---|---|---|
| CPU +12 V | 4.8 W | 24.0 W | 6.4 W |
| Motherboard +12 V | 0.0 W | 14.4 W | 4.8 W |
| Motherboard +3.3 V | 0.0 W | 0.1 W | 0.0 W |
| Motherboard +5 V | 0.1 W | 1.1 W | 0.4 W |
| System Total | 4.9 W | 38.9 W | 11.7 W |



Power Draw during Gaming
The next efficiency surprise is already waiting for us in the gaming benchmarks. While the A10-7800-based system is just a bit slower than one built using an A10-7850K, its power draw is quite a bit lower (32 W for the APU on average, and 40.3 W for the whole system). Similar to graphics cards, we observe peaks under load as high as 55 W. But the average proves that AMD operates the A10-7800 closer to its architecture's sweet spot. We used Unigine Heaven 4.0 at moderate quality settings to max out the APU, finding the results to be easily reproducible.
| Power Draw Gaming | Minimum | Maximum | Average |
|---|---|---|---|
| CPU +12 V | 7.2 W | 55.2 W | 32.0 W |
| Motherboard +12 V | 0.0 W | 24.0 W | 7.1 W |
| Motherboard +3.3 V | 0.0 W | 0.1 W | 0.1 W |
| Motherboard +5 V | 0.3 W | 1.9 W | 1.1 W |
| System Total | 12.8 W | 70.8 W | 40.3 W |


Power Draw at Full Load
Even at full load, the test system doesn’t exceed AMD's 65 W TDP unless you increase voltage for an overclock. We briefly played around with these motherboard settings, but found that the system took a huge efficiency hit, while performance barely increased. Hence, we went back to the default values and enjoyed an easy-to-cool machine with decent performance.
| Power Draw (Full Load) | Minimum | Maximum | Average |
|---|---|---|---|
| CPU +12 V | 7.2 W | 88.8 W | 56.2 W |
| Motherboard +12 V | 0.0 W | 19.2 W | 6.5 W |
| Motherboard +3.3 V | 0.0 W | 0.1 W | 0.1 W |
| Motherboard +5 V | 0.1 W | 1.7 W | 0.8 W |
| System Total | 13.0 W | 95.0 W | 63.6 W |



Next, we’ll use the A10-7700K and -7850K's unlocked multiplier to approximate the performance level or power draw of the A10-7800, one at a time.
Overclocking AMD's A10-7700K To Approximate An A10-7800's Speed
We put the lower-end APU into overdrive by overclocking its host and graphics processing blocks, simultaneously increasing the TDP ceiling to achieve similar performance in our three benchmarked games. But while we can get power consumption to rise, the -7700K just can't match the -7800's frame rates. One explanation is that we had to boost the x86 cores to 4.2 GHz to compensate for the pruned GPU, but we couldn't overclock the graphics engine enough to make up for the missing shaders.
| Minimum | Maximum | Average A10-7700K Overclock | Average A10-7800 | |
|---|---|---|---|---|
| CPU +12 V | 16.8 W | 64.8 W | 39.6 W | 32.0 W |
| Motherboard +12V | 0.0 W | 19.2 W | 8.2 W | 7.1 W |
| Motherboard +3.3 V | 0.3 W | 0.4 W | 0.4 W | 0.1 W |
| Motherboard +5 V | 0.5 W | 2.3 W | 1.4 W | 1.1 W |
| System Total | 23.1 W | 76.3 W | 49.6 W | 40.3 W |

Apart from the fact that we couldn’t match the A10-7800’s performance level, power draw increased by a whopping 9 W (or 24%).
Underclocking The A10-7850K To Approximate The A10-7800
Now we're reducing the A10-7850K's clock rate to match the -7800, and in a second test adjusting the TDP target to match the -7800’s power consumption.
It turns out that the A10-7850K is 2-5% faster during gaming. Similarly, its CPU performance is up to 5% higher. However, power draw is up to 29% higher! This is proof that the A10-7850K was tuned more for performance than efficiency. AMD wanted to put its best foot forward in the benchmarks, and as as a result the -7850K operates beyond Kaveri's sweet spot at the expense of efficiency.
| Minimum | Maximum | Average A10-7850K | Average A10-7800 | |
|---|---|---|---|---|
| CPU +12 V | 7.2 W | 69.6 W | 41.3 W | 32.0 W |
| Motherboard +12 V | 0.0 W | 19.2 W | 8.1 W | 7.1 W |
| Motherboard +3.3 V | 0.3 W | 0.5 W | 0.4 W | 0.1 W |
| Motherboard +5 V | 0.5 W | 2.3 W | 1.5 W | 1.1 W |
| System Total | 18.1 W | 81.3 W | 51.3 W | 40.3 W |

Comparison After Underclocking
When we underclock the A10-7850K to the -7800’s power level, we experience 1.5%-lower gaming performance. This is a small difference though, which could be the result of any number of variables.
What we really show is that the A10-7800 is basically a factory-underclocked A10-7850K and not a brand new APU. Besides lower stock clock rates, it lacks the -7850K’s unlocked multiplier.
| Minimum | Maximum | Average A10-7850K @ A10-7800 | Average A10 7800 | |
|---|---|---|---|---|
| CPU 12V. | 12.6 W | 58.0 W | 32.2 W | 32.0 W |
| Motherboard +12 V | 0.0 W | 21.6 W | 7.4 W | 7.1 W |
| Motherboard +3.3 V | 0.0 W | 0.2 W | 0.1 W | 0.1 W |
| Motherboard +5 V | 0.1 W | 1.9 W | 1.1 W | 1.1 W |
| System Total | 13.0 W | 76.1 W | 40.7 W | 40.3 W |

AMD positions its A10-7800 perfectly, hitting the right balance between power and performance for Kaveri. The chip's default frequencies constitute the best compromise between performance and power consumption. The -7850K's disadvantage against the -7800 comes from the x86 cores, which are more efficient at the -7800's lower clock rate, while not limiting the APU's graphics component.
The following diagrams illustrate the A10-7800's power consumption in three different scenarios. They're mostly intended for experts and hardware enthusiasts, and we didn’t want to clutter the story with a data overload. Still, such a detailed power analysis, which tracks all rails simultaneously, is a rare treat that can be only achieved with a sophisticated arrangement of expensive test equipment.
Power Draw at Idle




Power Draw during Gaming





Power Draw during a Stress Test





Is this just another APU from AMD? Not quite. The A10-7800 has a different purpose than the flagship -7850K, and that's good news. AMD found the right spot between moderate performance and modest power consumption. As a result, the A10-7800 is a winner if you're building a well-balanced, low-cost PC where price, performance, and efficiency are well-balanced. Yes, you could easily build a faster system using a CPU and discrete graphics card, but it'd almost certainly pull more power from the wall, and would probably be pricier, too.
That combination of attributes, culminating in commendable efficiency, is the A10-7800's unique selling point. You get this from AMD's factory settings. As a result, we don't believe that attaching discrete graphics is a good idea. If gaming is your bag, then find a dedicated host processor and build from there. Otherwise, you're losing the APU’s only advantage, while suffering the compromises of a multi-GPU machine. Besides that, you’d affect the price/performance equation in a negative way. For instance, compare the power draw of a Radeon R7 250 with this APU. The processor gives you a comparably-fast graphics engine with x86 cores thrown in at lower consumption levels. This earns the A10-7800 our Tom's Hardware Smart Buy recognition. If you're considering an APU, the -7800 is the family's purpose-built solution (more so than the -7850K or -7700K).

And of course, if you yearn for high efficiency, the other system components also matter. This APU isn't cheap, and it's too easy to jeopardize those balanced figures we measured with platform parts that aren't complementary. An Athlon X4 760K and Radeon R7 250 would cost about the same. The APU is self-contained, though. And because it takes up less room, your case can be both smaller and easier to cool. The A10-7800 sets itself up to be the engine at the heart of a truly compact entry-level computer.
For such an application, you'd probably want a well-built mini-ITX motherboard. Short of that, MSI's A88XM Gaming proves itself to be an energy efficient microATX alternative, drawing a mere 4-8 W. It's armed with a feature-rich BIOS that lets you optimize the APU's behavior further. Given a shortage of exemplary Socket FM2+-based boards, we're glad we picked this one for our experiment.
If you want to build a home theater or general-purpose PC for casual gaming and office work, the A10-7800 is a smart choice. Just add fast memory (to feed the graphics engine) and a good motherboard. AMD hits the sweet spot, closing the gap between its A10-7700K and -7850K.
In fact, the A10-7800 is an even better option than the -7850K, since that APU's slim performance gain comes at a significant cost to efficiency. You won't notice the -7800's slightly lower performance. What you should be impressed by is the fact that we ran the whole system, including a 480 GB SSD, from a diminutive DC-DC converter and an efficient external AC adapter rated for 65 W. With the addition of a closed-loop liquid cooler, the system draws less than 55 W at the wall. I think it's possible to get below 50 W with an air cooler and some more tweaking. That's an amazing feat and a welcome surprised from a company commonly critiqued for power-hungry processors.

















