Today we look at two AMD A78-based motherboards (MSI's A78M-E35 and Gigabyte's GA-F2A78M-D3H), to determine how they stack up against A88X.
An Introduction To AMD's A78 Chipset
It has been nearly a year since Tom's Hardware last reviewed a motherboard for AMD's processors. And with the recent Godavari APU announcement, we thought we should dust off our Kaveri APUs to see if we can find some additional value in the company's not-so-famous alternative chipsets.
Our own Thomas Soderstrom looked at AMD's A88X chipset from an enthusiast's standpoint and determined that an APU placed into this platform, along with a complementary graphics card, could provide plenty of performance and features to satisfy anyone looking for a mainstream Intel alternative.
But does the average desktop user really need CrossFire, an eight-phase voltage regulator or enough SATA 6Gb/s ports to load an ATX tower? Do they instead want an HTPC with enough bells and whistles for an out-of-the-box overclock, modest graphics horsepower and plenty of room for storage without busting the budget (or form factor)? I have good news, friends: AMD has a solution and you might not have even considered it.
|PCI Express||1x16 / 2x8||1x16||1x16||1x16 / 2x8||1x16|
|SATA (6Gb/s)||8 (8)||6 (6)||6 (0)||8 (8)||6 (6)|
|RAID||0, 1, 5, 10||0, 1, 10||0, 1, 10||0, 1, 5, 10||0, 1, 10|
|RAID Driver||Promise||Promise||Promise||Dot Hill||Dot Hill|
|USB (3.0)||14 (4)||14 (4)||14 (0)||14 (4)||14 (4)|
* Per AMD's website - FM2 easy upgrade path featuring latest USB and SATA technologies; FM2+ backwards compatible, future-ready and PCIe 3.0-ready.
Introducing the A78, AMD's "media-class" chipset. Though it doesn't have any fancy Xes or a three-digit designation, this platform is completely adequate for most folks. In fact, comparing AMD's FM2+-compatible offerings, the only difference between the flagship and A78 is an inability to split 16 lanes of PCIe into two x8 links, two missing SATA 6Gb/s ports and RAID 5 support.
Today, I will be comparing MSI's A78M-E35 and Gigabyte's F2A78M-D3H. I will also be using some of Thomas' data for the Asus A88X-Pro Eric's MSI H81M-E34 results for comparisons to a different AMD chipset and Intel's low- budget offering.
MSI's A78M-E35 is the smaller of our two A78 boards, measuring 8.9x8.5 inches. Though both samples are microATX-sized, MSI's offering is geared more toward smaller enclosures and entry-level builds. With its reduced size, I have access to only two DDR3 DIMM slots, and the bottom of the board does get cramped if you have a PCI card installed. There is only one available four-pin ATX12V CPU connector, so make sure your power supply has a four-pin lead or that it can split its eight-pin connector.
The board offers two four-pin PWM fan connectors (one system fan and one CPU fan); they're are located right where you would need them. For those of us with legacy peripherals, two PS/2 ports are available. There are enough USB ports available on the back panel and through headers to satiate mainstream users, though higher-end chipsets enable more.
MSI's UEFI definitely has more of a GUI-like feel than I am used to, and the home screen isn't as intuitive as I would have liked. Working through some of the pages, I noticed that the option selections were not consistent. Some of them required clicking for a list of settings, others demanded that you press + or - to increment up or down and some wanted you to key a number in specifically. Finally, since this is an entry-level board, there is no access to voltage biasing through the BIOS. If you intend to use OC Genie, MSI's one-button overclock, the software tool only supports 65W APUs with this chipset.
Looking at the board's packaging, "Military Class, "Top Quality" and "Stability" are the most readily apparent phrases. For less than $60 (at the time of this writing), that seems like a lot of marketing for such a low price tag. MSI has been a solid contender in the past, so I look forward to seeing the results.
Gigabyte's F2A78M-D3H is a sleek black square measuring 9.6x9.6 inches and sporting the same back-panel connections as MSI's board. With the additional real estate, we get access to four DDR3 DIMM slots (supporting up to 64GB), an additional PCIe x16 slot and a little more space between the SATA connectors. A word of caution: the second PCIe x16 link is wired to only four lanes, so it's better suited to high-end storage and networking controllers.
Though I do enjoy the labeled front-panel wiring diagram printed out on the board, I am disappointed in some of Gigabyte's other layout decisions. All of the fan connections are PWM-capable, but the system and APU headers are right beside each other and directly below the cooler's mounting bracket. It was very difficult to connect fans after installing our large heat sink. Also, the other fan header is in the top-right corner above the 24-pin ATX connector, where traditionally you would be in hard drive bay territory. Needless to say, wiring around there can be tricky. Overclockers reading this might be glad to see the eight-pin EPS connector for additional power.
Gigabyte's UEFI feels more like a traditional BIOS, and to me is easier to navigate. After the lack of voltage biasing from MSI's board, I was pleased to see that Gigabyte's firmware adds both voltage biasing and load-line calibration options. For this comparison, I did not fully test those features, but I suspect that this platform could achieve overclocking comparable to what you'd get from AMD's higher-end APU-oriented chipset.
Gigabyte's packaging states that the product is "Ultra Durable." Will this extra durability help drive my APU to peak out-of-box performance? In data I trust!
How We Test
Test System Configuration
I was fortunate enough to have access to the same APU and heat sink/fan combo that Thomas used in his A88X article. This should show whether I can achieve similar overclocking results as well as similar CPU performance in the applicable benchmarks.
I am also using the same brand and part numbers for memory and solid-state storage to better compare system configurations between our Intel and AMD budget systems. This will help eliminate DRAM inconsistency when comparing memory controller performance.
Though this APU has enough graphics muscle to drive decent frame rates at lower detail levels, the motherboard review team decided to put in a low-power GPU to help put both AMD and Intel budget systems on a level playing field. For that, a Gigabyte GeForce GT 730 was installed for graphics workloads so that I can test system stability and repeatability across platforms. Both motherboards do support AMD Dual Graphics technology, allowing you to install a discrete GPU for CrossFire with your APU.
To power these test boards, I am using the same PSU used in Four LGA-1150 Motherboards Under $60, be Quiet!'s Straight Power 10 500W. Since it is 80 PLUS Gold-rated, this unit will prove to be more efficient in low-load conditions.
|Synthetic Benchmarks and Settings|
|PCMark 8||Version: 2.3.293|
Work, Home, and Creative Benchmarks
|SiSoftware Sandra||Version: 2015.01.21.15|
CPU Arithmetic, Multimedia, Cryptography
File System Bandwidth
|Cinebench R15||Version: R15.0 x64|
CPU Single- and Multi-Core
|3D Tests and Settings|
|3DMark||Cloud Gate Version: 1.1, Skydiver Version: 1.0|
Test Set 1: Cloud Gate, 1920x1080, Default Preset
Test Set 2: Skydiver, 1920x1080, Default Preset
|Unigine Heaven 4.0||Version 4.0, Built-in Benchmark|
DirectX 11, Low Detail, 1920x1080, No AA, No Tessellation
|Unigine Valley 1.0||Version 1.0, Built-in Benchmark|
DirectX 11, Low Detail, 1920x1080, No AA
Since I have three comparisons going on within these charts, I will do my best to describe what the numbers mean and how those results reflect the different scenarios. Though the primary focus is comparing the A78-based boards, I also want to explore which platform might be best for a particular budget and system configuration. Let’s jump in!
Both A78 boards demonstrate similar behavior, which is expected. With less than 1.0 percent variance between them in all three usage scenarios, it's safe to say that the components are configured properly. This also shows us that when I compare A78 to A88X or H81M, I do see a slight performance decrease, which could be attributed to memory performance and possibly APU limitations.
Again, I observe consistent results with the A78 boards, with the Whetstone Double AVX showing the largest deviation of 2.6 percent across all SiSoftware tests. It is interesting (though not entirely surprising) to see that the Intel platform has higher memory performance than A78 while using the same memory modules. Also, the AMD APU scores better than the dual-core Pentium G3258, though I did not quite get twice the performance.
Both A78 boards perform well when I compare them to the multi-threaded results from Intel. However, single-threaded performance is clearly better on the Pentium. It's also important to note that Intel’s multi-thread to single-thread ratio is 1.91 (close to two since it’s a dual-core processor), whereas AMD's ratio is 3.58, quite a way off from the ratio of four for a quad-core processor. This shows how important CPU architecture is.
The Cloud Gate benchmark is geared toward the Windows notebook and home PC market, and I see in the results that extra CPU horsepower helps the AMD-based platforms attain an extra 13 percent on the combined score. This lead disappears once I run the mid-range PC workload, where the physics component becomes less important. Both A78 results are right on top of each other, and it is starting to show how consistent the platform is.
Unigine Valley also reveals consistency across all three platforms, with a slight edge to Intel for average and minimum frame rates. However, at this resolution, anything below 30 FPS is marginal. Unigine Heaven does show more variation, and I finally see an instance where the MSI A78 board has higher performance than Gigabyte's offering, even if it is only a few frames.
In every test condition, Intel's platform draws less power than the A78-based boards, which doesn't surprise us given each processor's TDP. The Gigabyte board burns a meager 2W more when running idle on the Windows desktop and continues to keep this delta under load with FurMark and Prime95.
Clearly, Intel has the advantage due to its lower-power part and more efficient architecture. Both A78 motherboards report the same CPU load temperature, but the MSI board has a nine-degree advantage over Gigabyte's solution when we measure VR temperatures. This is very beneficial for MSI's target segment, which might not have as much airflow. This is also interesting because the Gigabyte board dissipate just two additional watts of power in this test condition.
For the sake of an equal comparison, both A78-based samples were overclocked using the firmware's ratio multiplier option. Without adjusting CPU voltage, I was unable to reach the clock rate that Thomas achieved. The MSI board completed an eight-hour Prime95 stress test at 4200MHz, while the Gigabyte board reached a 4100MHz overclock under the same test conditions. At first glance, it seems like the MSI board is better out of the box, but Gigabyte's board has much more potential, offering a wider range of BIOS options and more current through its EPS connector.
As always, overclocking is very stressful on components, and these boards are no exception. The MSI and Gigabyte A78-based boards reached regulator temperatures in the 80- to 90-degree Celsius range, which I do not feel comfortable with over extended periods. In a smaller enclosure or in an area with insufficient airflow, I could see this being a hazardous situation. Since these boards are not true enthusiast-grade parts, I recommend backing off to a 40x multiplier. Both boards were able to run the Mushkin memory modules in their XMP profiles flawlessly, but I did not further tweak the timings to improve performance.
As the data shows, the A78 platform is consistent. From a chipset perspective, both of our samples perform well, making it hard to pick a favorite. Setting the Intel versus AMD debate aside, I really like AMD's A78. When I compare what Thomas did with A88X to my experiences with A78, the A88X does appear faster. And the flagship's added features open the door to a more sophisticated CrossFire configuration, more storage in RAID and better overclocking options. But are those features worth an extra $30 or so? Can you live without aggressive overclocking or CrossFire? For a lower price, I believe the A78 is a solid choice for many builders.
Rather than give any awards in this article, I will consider this data once I get a chance to look at some competing hardware. In the meantime, I'll give you my final thoughts on each product.
If your goal is to build an HTPC or small office box and you have no intentions of overclocking aggressively, MSI's A78M-E35 will work well. I enjoyed its layout more than Gigabyte's, though the UEFI was harder to navigate. Of the two boards I reviewed, MSI's motherboard is more in-line with what AMD's APU was designed for.
If you are building an A88X clone but don't need the additional PCIe lanes, Gigabyte's board shows some real potential. Given the right tuning, air flow and memory selection, I think this system could compete with similarly-equipped A88X systems. If the durability claim holds water, this would be even better for a tuner's system. This board also works in larger HTPC cases, but the odd fan header placement might make cable management a challenge.