Once upon a time, processor reviews were simpler. Isolating x86 performance made it easy for us to draw comparisons. Today, though, we have on-board graphics, integrated memory controllers, and ISA extensions to look at. The benchmarks we run play a huge part in exposing the benefits of each piece of a modern host processor, and we're naturally very careful about testing as thoroughly as possible.
AMD's Trinity architecture received a lot of attention from us when it launched:
AMD Trinity On The Desktop: A10, A8, And A6 Get Benchmarked!
AMD Desktop Trinity Update: Now With Core i3 And A8-3870K
Gaming At 1920x1080: AMD's Trinity Takes On Intel HD Graphics
AMD's Trinity APU Efficiency: Undervolted And Overclocked
In those stories, we discovered that AMD's Piledriver-based modules are faster than their predecessors based on the Bulldozer architecture. But because AMD never released a Bulldozer-based APU, we also had to face the fact that, in some cases, older Llano-based parts were as quick or quicker. Hardly a reason to toss your old Socket FM1 motherboard and upgrade to Socket FM2, right?
Fortunately, the new APUs do manage to nudge graphics performance forward. And because AMD's on-die graphics engines are already significantly faster than Intel's best effort, the extra speed only served to hammer that point home.
And so, you're faced with an interesting conundrum. On one hand, we wouldn't mind a Trinity-based APU specifically for its 3D capabilities. On the other, we're still compelled to build with an add-in graphics card in mind. Really, then, the ideal customer for an APU is someone able to enjoy the cost savings of an A10 or A8, who won't be stymied by the chip's limitations, and who won't then go get frustrated and buy a discrete GPU. At that point, you're just watering down the platform's value and compounding its power consumption.

With that in mind, we're curious about AMD's latest driver optimizations. We also wonder if motherboard vendors have any graphics performance-optimizing tricks up their sleeves. So, rather than dropping some unrealistically-expensive discrete board into today's test beds, we're evaluating six Socket FM2-equipped platforms using an AMD A10-5800K operating all on its own.

| Socket FM2 Motherboard Features | |||
|---|---|---|---|
| ASRock FM2A85X Extreme6 | Asus F2A85-V Pro | ECS A85F2-A GOLDEN | |
| PCB Revision | 1.04 | 1.01 | 1.0 |
| Chipset | AMD A85X FCH | AMD A85X FCH | AMD A85X FCH |
| Voltage Regulator | Ten Phases | Eight Phases | Five Phases |
| BIOS | P1.30 (10/12/2012) | 5104 (09/14/2012) | 10/12/12 |
| 100.0 MHz RCLK | Variable | Variable | Variable |
| Internal Interfaces | |||
| PCIe 3.0 x16 | None | None | None |
| PCIe 2.0 x16 | 3 (x16/x0/x4 or x8/x8/x4) | 3 (x16/x0/x4 or x8/x8/x4) | 2 (x16/x0 or x8/x8) |
| PCIe x1/x4 | 2/0 | 2/0 | 3/0 |
| USB 2.0 | 3 (6-ports) | 4 (8-ports) | 3 (6-ports) |
| USB 3.0 | 1 (2-ports) | 1 (2-ports) | 1 (2-ports) |
| SATA 6.0 Gb/s | 7 | 7 | 7 |
| SATA 3.0 Gb/s | None | None | None |
| 4-Pin Fan | 2 | 5 | 1 |
| 3-Pin Fan | 4 | None | 2 |
| FP-Audio | 1 | 1 | 1 |
| CD-Audio | None | None | None |
| S/PDIF I/O | Output Only | Output Only | Output Only |
| Internal Buttons | PWR, RST | BIOS Flash, MemOK, DirectKey | None |
| Diagnostics Panel | Numeric | None | None |
| Legacy Interfaces | 2 x PCI, Serial | 2 x PCI, Serial | 2 x PCI, Serial |
| I/O Panel Connectors | |||
| P/S 2 | 1 | 1 | None |
| USB 3.0 | 4 | 4 | 4 |
| USB 2.0 | 2 | 2 | 2 |
| IEEE-1394 | None | None | None |
| Network | 1 | 1 | 1 |
| eSATA | 1 | 1 | 1 |
| CLR_CMOS Button | Yes | No | Yes |
| Digital Audio Out | Optical | Optical | Optical |
| Digital Audio In | None | None | None |
| Analog Audio | 5 | 6 | 5 |
| Video | VGA, DVI-D, DisplayPort, HDMI | HDMI, DisplayPort, VGA, DVI-D | VGA, DVI-D, DisplayPort, HDMI |
| Other Devices | None | None | None |
| Mass Storage Controllers | |||
| Chipset SATA | 7 x SATA 6Gb/s 1 x eSATA 6Gb/s | 7 x SATA 6Gb/s 1 x eSATA 6Gb/s | 7 x SATA 6Gb/s 1 x eSATA 6Gb/s |
| Chipset RAID Modes | 0, 1, 5, 10 | 0, 1, 5, 10, JBOD | 0, 1, 5, 10 |
| Add-In SATA | None | None | None |
| USB 3.0 | ASM1042 PCIe (2-ports) A85X Integrated (4-ports) | ASM1042 PCIe (2-ports) A85X Integrated (4-ports) | ASM1042 PCIe (2-ports) A85X Integrated (4-ports) |
| IEEE-1394 | None | None | None |
| Gigabit Ethernet | |||
| Primary LAN | RTL8111E PCIe | RTL8111F PCIe | RTL8111E PCIe |
| Secondary LAN | None | None | None |
| Audio | |||
| HD Audio Codec | ALC898 | ALC892 | VT1819S |
| DDL/DTS Connect | Not Specified | Not Specified | Not Specified |
| Warranty | Three Years | Three Years | Three Years |

| Socket FM2 Motherboard Features | |||
|---|---|---|---|
| Gigabyte F2A85X-UP4 | MSI FM2-A85XA-G65 | Sapphire Pure Platinum A85XT | |
| PCB Revision | 1.0 | 1.1 | 1.0 |
| Chipset | AMD A85X FCH | AMD A85X FCH | AMD A85X FCH |
| Voltage Regulator | Eight Phases | Eight Phases | Eight Phases |
| BIOS | F3g (10/03/2012) | V1.1 (10/09/2012) | 0.40 (09/12/2012) |
| 100.0 MHz RCLK | Variable | Variable | Variable |
| Internal Interfaces | |||
| PCIe 3.0 x16 | None | None | None |
| PCIe 2.0 x16 | 3 (x16/x0/x4 or x8/x8/x4) | 2 (x16/x0 or x8/x8) | 3 (x16/x0/x4 or x8/x8/x4) |
| PCIe x1/x4 | 3 (1 shared with x4) / 0 | 3/0 | 2/1 |
| USB 2.0 | 4 (8-ports) | 3 (6-ports) | 2 (4-ports) |
| USB 3.0 | 1 (2-ports) | 1 (2-ports) | 1 (2-ports) |
| SATA 6.0 Gb/s | 7 | 8 | 7 |
| SATA 3.0 Gb/s | None | None | None |
| 4-Pin Fan | 5 | 3 | 1 |
| 3-Pin Fan | None | 2 | 5 |
| FP-Audio | 1 | 1 | 1 |
| CD-Audio | None | None | None |
| S/PDIF I/O | Output Only | None | Input andOutput |
| Internal Buttons | PWR, RST, CLR | PWR, RST, CLR, OC Genie | PWR, RST, CLR, ROM selector |
| Diagnostics Panel | Numeric | None | Numeric |
| Legacy Interfaces | PCI, Serial | 2 x PCI, Serial (mini) | 2 x PCI, Serial |
| I/O Panel Connectors | |||
| P/S 2 | 1 | 1 | 2 |
| USB 3.0 | 4 | 2 | 2 |
| USB 2.0 | 2 | 4 | 4 |
| IEEE-1394 | None | None | None |
| Network | 1 | 1 | 1 |
| eSATA | 1 | None | None |
| CLR_CMOS Button | No | No | No |
| Digital Audio Out | Optical | Optical | Optical |
| Digital Audio In | None | None | None |
| Analog Audio | 6 | 6 | 6 |
| Video | VGA, DVI-D, HDMI, DisplayPort | VGA, DVI-D, HDMI, DisplayPort | HDMI, DisplayPort, VGA, DVI-D |
| Other Devices | None | None | Bluetooth Transceiver |
| Mass Storage Controllers | |||
| Chipset SATA | 7 x SATA 6Gb/s 1 x eSATA 6Gb/s | 8 x SATA 6Gb/s | 7 x SATA 6Gb/s 1 x mSATA 6Gb/s |
| Chipset RAID Modes | 0, 1, 5, 10, JBOD | 0, 1, 5, 10 | 0, 1, 5, 10 |
| Add-In SATA | None | None | None |
| USB 3.0 | EJ168A PCIe (2-ports) A85X Integrated (4-ports) | A85X Integrated | A85X Integrated |
| IEEE-1394 | None | None | None |
| Gigabit Ethernet | |||
| Primary LAN | RTL8111F PCIe | RTL8111E PCIe | RTL8111F PCIe |
| Secondary LAN | None | None | None |
| Audio | |||
| HD Audio Codec | ALC892 | ALC892 | ALC892 |
| DDL/DTS Connect | Not Specified | Not Specified | Not Specified |
| Warranty | Three Years | Three Years | One Year |
The first thing we noticed about ASRock’s entry into our first Socket FM2 motherboard round-up wasn't the board itself, but rather its name. It follows a convention that some competing vendors make even more complex, adding socket and chipset identification to the series (Extreme6) name. We prefer nomenclature that's easy to remember, and ASRock gets us halfway there.
Opening the box, we were pleased to find ASRock's familiar Extreme6-class feature set extended to this A85X-based board, including the CLR_CMOS button and an extra pair of USB 3.0 ports added through ASMedia’s ASM1024 controller. AMD was an early proponent of DisplayPort, so we aren't surprised to find a full-sized DisplayPort output next to the platform's HDMI, VGA, and DVI connectors.
The A85X chipset originally surprised us by enabling eight SATA 6Gb/s ports, in spite of the FCH's 2 GB/s connection to the CPU. ASRock wires one of those connectors to an eSATA interface on the rear I/O panel and faces the eighth leftover connector outward, next to six forward-facing ports on the board's bottom-right edge.
ASRock also makes full use of the APU’s PCIe lanes by adding a third x16-length slot. PCIe 2.0 x16-x0-x4 lane connections automatically switch to x8-x8-x4 when a card is detected in the middle slot.
The A85X FCH’s four lanes connect two PCIe x1 slots, a gigabit Ethernet controller, and the previously-mentioned secondary USB 3.0 controller. In theory, the chipset’s UMI interface could be a pretty severe bottleneck. Fortunately, it'd be rare to see accesses to all attached devices at the same time.
Layout is good, with a few exceptions. Namely, there's a front-panel audio header on the bottom-rear corner and an ATX12V connector that’s a little too close to the CPU socket. If you own an oversized heat sink, you could have a tough time connecting or disconnecting that auxiliary power source. Moreover, if you're using an older case, you might find its built-in front-panel audio cable a little too short. Forward-facing SATA ports provide room for extra-long expansion cards, but could be blocked out by the hard drive cages of some older cases, too.
Top-side power and reset buttons, plus the two-digit diagnostics display, aid in motherboard troubleshooting. But while those features are handy for reviewers and competitive overclockers, they're typically inaccessible in a buttoned-up build.

Seven of eight SATA ports are accessible internally, yet ASRock includes only four of the seven cables. Color us satisfied, though. That's two times more cables than most folks need, giving you the flexibility to attach an SSD, two storage disks, and an optical drive right out of the box.
ASRock dresses the FM2A85X Extreme6's firmware in stellar fashion, but doesn't allow new artwork to complicate its classic menu arrangement. Kudos for that. The OC Tweaker menu works the way we expect it to.

We chose 1.45 V for our overclocking attempts, and ASRock's Extreme6 board produced that voltage level at its 1.4375 V setting. AMD's APU was then overclockable to nearly 4.5 GHz with 100% stability.

The FM2A85X Extreme6 provides load-line calibration to stabilize CPU voltage under load, but we found it most consistent at the “0%” setting.

Since EPP is no longer mainstream, we were happy to find Intel’s competing XMP standard supported by ASRock’s AMD motherboard. XMP Profile 1 made it easier for us to nail down the highest stable DRAM frequency using our memory kit's rated timings.
XMP made things easy for us, but we still could have picked those same timings manually if we needed to. The FM2A85X Extreme6 displays XMP settings in its DRAM Timing Control menu, and provides a broad array of manual adjustments to match them.
ASMedia’s ASM1042 controller adds an additional pair of USB 3.0 ports to the rear panel of Asus' F2A85-V Pro, while a set of PCIe 2.0 pathway switches contributes a second graphics card slot for converting x16-x0 configurations to x8-x8. Asus uses the CPU’s remaining four lanes to add a third PCI Express 2.0 slot with lower bandwidth.
Though its layout appears familiar, a closer look at the F2A85-V Pro reveals better CPU power connector and front panel audio header placement. An even closer look reveals a few tiny buttons for the Asus-exclusive features MemOK, USB BIOS Flashback, and Asus DirectKey.
MemOK automatically underclocks poorly-programmed memory to at least facilitate boot-up into the UEFI for adjustment. USB BIOS Flashback gives you the ability to update the board's firmware with no other hardware installed. Finally, Asus DirectKey allows users to enter the UEFI at any time during motherboard initialization.
Asus is particularly proud of its Fan Xpert 2 feature, which facilitates control over and monitoring of the motherboard's fan headers. Attached coolers can be set to spin up or down automatically, or configured to operate at fixed speeds.
USB BIOS Flashback is particularly useful when purchasing a newer CPU and older motherboard simultaneously, since it circumvents the issue of hardware compatibility and a firmware mismatch. Previously, that might have required you to buy an older APU just to update the board's UEFI to properly recognize the new chip. On this platform in particular, BIOS Flashback isn't as necessary, though, since Socket FM2 and Socket FM2-compatible APUs are both new. This could change in the future if AMD maintains backward-compatibility with its next-gen architecture.
The secondary function of Asus DirectKey is that of a power button. Touching the DirectKey button when the system is already up and running in an operating system triggers a shut-down, and touching it while the system is off causes it to boot directly to UEFI. That function is increasingly practical as the window for hitting the Del key to enter the UEFI gets shorter with each new motherboard generation (and even more so with Windows 8).
We’ve noticed that AMD fans are more likely to be frugal, but perhaps the time has come to quit being frugal about cases? While a good case can last through several builds, many older designs lack space to connect SATA cables to the forward-facing ports common on modern motherboards. The F2A85-V Pro can be counted among these modern designs, with six of its seven internal SATA ports facing forward to allow increased card clearence.

The F2A85-V Pro has seven internal and one external SATA connector, but users will only be able to connect half of those out of the box. The installation kit includes four internal cables, though Asus also adds a pair of its quick-connecting cable extenders.
Asus Ai Tweaker remains familiar, without so much as a new theme to distract us from its navigational ease.
The company's D.O.C.P. mode reads our memory kit's DDR3-2666 XMP profile and, when enabled, attempts to achieve the target data rate.

Since AMD's integrated memory controller doesn't support the proper ratio, the F2A85-V Pro multiplied its DDR3-2133 ratio by a 125 MHz reference clock. Unable to boot at this high memory speed, we tried D.O.C.P. once more after restoring the APU's stock reference clock and increasing its memory ratio to 24x. Success!
The F2A85-V Pro’s combination of hardware and firmware optimization allowed our A10-5800K CPU to crest 4.5 GHz with no noticeable loss of stability, but some extra voltage would be needed to achieve that combination of frequency and stability. The motherboard’s 1.4375 V setting gave us around 1.42 V at idle, while its 1.625 V DRAM setting gave us around 1.65 V.

Our target core voltage of 1.45 V was only really needed at full load, and that was achieved by setting the board’s CPU Load Line Calibration to High. Other combinations of higher base voltage and less load line calibration produced less-precise results.

Other Digi+ Power Control settings worked adequately at motherboard defaults.
The main reason we like XMP support is for its automatic timings configuration. The F2A85-V Pro supports those automatic timings through Asus D.O.C.P. mode. We also discovered that the motherboard compensates for higher frequency multipliers with appropriately-looser secondary timing when set to manual mode.
At long last, ECS is acting on our complaints about confusing the various models in its line-up by putting the word "Golden" into the name of this specific board. Previous examples simply used gold packaging to indicate the difference. Online searches are text-based, though, leaving way too much room for confusion. By naming the A85F2-A Golden more descriptively, it's a piece of cake to find on sites like Newegg.
Just because this board bears ECS' flagship designation doesn't mean it includes all of the same features as past golden-class boards, unfortunately. The Port 80 diagnostics display present on ECS' higher-end Intel boards is now missing, along with the third four-lane graphics card slot offered by most competing A85X-based products. ECS does add the same secondary USB 3.0 controller as the first two competitors in today’s round-up. It also keeps the same suite of display outputs and has a similar suite of seven internal and one external SATA 6Gb/s ports as ASRock and Asus.
A handy I/O panel CLR_CMOS button and a CPU power connector that’s too close to the CPU socket are both matched by ASRock, yet the A85F2-A Golden lacks the modestly-valued internal buttons of that competitor.
It's possible that ECS decided the A85F2-A Golden’s missing features weren’t worth the cost of implementation. More than likely, nobody is going to bother installing three high-end graphics cards on AMD's value-oriented Socket FM2 platform, and the on-board power and reset buttons offered by other motherboard companies aren't worth much once you button the side of your case up. With a new understanding that ECS' Golden label is relative (and not indicative of a certain feature set), we now have to hope that the company's flagship costs less than ASRock's Extreme6 model, which comes loaded with more features.

Surprising us with seven SATA cables for seven ports, it appears that ECS knows math a little better than its competition. Most customers will only use two or three of the seven cables, but we still count the complete installation kit in considering the A85F2-A Golden’s overall worth.
We wanted to give every product in today’s comparison equal coverage, but that’s difficult when ECS' A85F2-A Golden only has one page of firmware settings to address the enthusiast market.

Scaling back the number of overclocking-oriented settings doesn't help ECS avoid errors, though. We found the DDR3-2400 multiplier labeled DDR3-2600. Not that we were able to utilize those higher settings; we weren’t able to get our DDR3-2666-capable kit stable, even after dropping to the board’s DDR3-2133 setting. Instead, we had to start out with DDR3-1866 and work our way up using the CPU reference clock.
Smaller issues plagued our CPU overclocking effort. The A85F2-A Golden couldn't keep our APU stable at the expected 4.5 GHz, though it was able to sustain the next step down: 4.45 GHz at 44 x 101 MHz.
The setting labeled “CPU Over-Clocking Func” was supposed to enable CPU reference clock control, but it reverted to Disabled every time we entered the UEFI.
The lowest setting we found for CPU VCore was 1.50 V, but our target was only 1.45 V. We were surprised to find that the board applied the 1.45 V we wanted to see under load by default.
We don’t want ECS to feel shortchanged by the lack of UEFI imagery, but our attempts to find another overclocking-related page in its firmware came up empty. We weren’t even able to show you the board’s list of saved configurations, because it doesn’t support those. Without the ability to apply user profiles, overclockers have start over any time their changes force them to use the A85F2-A Golden’s CLR_CMOS button.
A short and straight naming scheme makes Gigabyte’s F2A85X-UP4 easier to write about and remember, though we haven’t tested enough UP4-series products to know how their features are supposed to stack up against -UP3 or -UP5 boards. Instead, we see the similarity in features between Gigabyte’s -UP4 and ASRock’s Extreme6 platforms.
Gigabyte includes a secondary USB 3.0 controller, but uses Etron’s EJ168A rather than ASMedia's chip. It puts the F2A85X-UP4’s on-board power, reset, and CLR_CMOS switches in a group on the board’s upper-right corner. Gigabyte also moves its eight-pin auxiliary power connector away from the Socket FM2 interface. But none of those changes detract from the board’s value. In fact, we see the power connector placement as an improvement that makes larger heat sinks easier to work around.
We also find three PCIe x16-length slots that automatically switch from x16-x0-x4 to x8-x8-x4 lane configurations whenever a card is detected in the middle slot. But Gigabyte gives up the second PCI slot in favor of a third PCIe x1 slot, and that’s probably not an improvement, since the x4 slot drops to x1 mode whenever a card is dropped into the third x1 slot. AMD's chipset simply runs out of PCI Express connectivity when all of this board's slots are populated.
If you complement AMD's APU with a double-slot add-in card for Dual Graphics operation, you'll lose access to the second x1 slot. So, Gigabyte’s decision to share the third x1 slot with its x4 interface is questionable at best. We would have rather sacrificed the less-useful second slot in the interest of retaining the x4 connector's full bandwidth.
The seventh internal SATA connector faces outward along the F2A85X-UP4’s bottom edge, which could also cause it to be blocked by a graphics card. Fortunately, it’s so far forward that most graphics cards (at least the ones you'd install on a Socket FM2 motherboard) are too short to hide it.

The F2A85X-UP4 installation kit includes a generous six SATA cables, which is only one shy of the seven ports this board boasts. The chipset's eighth SATA port is devoted to eSATA on the rear I/O panel.
Gigabyte’s M.I.T. menu is familiar, relying on various sub-menus to provide comprehensive configuration settings.

The first sub-menu provides reference clock and multiplier settings. The F2A85X pushes our A10-5800K CPU to 4.5 GHz at 1.45 V.

The M.I.T. Advance Memory Settings sub-menu provides redundant DRAM multiplier control, in addition to hosting another menu for memory timings. Gigabyte matches ASRock by providing Intel XMP settings on an AMD motherboard.
Setting DRAM Timing Selectable to “Quick” allows both channel timings to be set simultaneously. Expert mode allows per-channel timings, but we’ve never found these beneficial. Primary, secondary, and memory waveform controls are selectable.

Gigabyte appears to have accepted our feedback on its inflated menu count, combining a collection of Advanced Voltage Settings that had formerly been spread across three sub-menus. Our target 1.45 V CPU core at full load was achieved by selecting a 1.425 V baseline and “Extreme” Loadline Calibration. Any combination of higher baseline and lower loadline settings produced less accurate results under full load.
It’s been a while since we’ve seen a -G65 board from MSI, the most recent being a microATX model. We expect a larger board to have more slots, and AMD's newer chipset helps enable the FM2-A85XA-G65’s expanded feature set.
The A85X Fusion Controller Hub even lets MSI give up an add-in controller, since the core logic integrates four USB 3.0 ports. The chipset also integrates support for DisplayPort output, and MSI chooses to ditch eSATA in favor of more internal ports.
Sporting a pair of second-gen PCI Express x16 slots that automatically switch from x16-x0 to x8-x8 transfers whenever the second is filled, the FM2-A85XA-G65’s closest layout match in today’s comparison comes from ECS. Both boards lack a x4 slot that competing boards do offer, so we're hoping to see lower prices accompany the feature reductions of both products.
The FM2-A85XA-G65 adds a few minor features compared to its closest rival, including power, reset and CLR_CMOS buttons. MSI OC Genie is also present, alongside a handy row of voltage detection points. Those additions are most useful to reviewers and the hardcore overclockers who use open test beds on a regular basis. Because AMD's Socket AM2 platform is decidedly budget-focused, competitive overclockers probably won't be spending much time tuning APUs. These features would come in handy if MSI ever decided to to sponsor a budget-oriented overclocking event.
One minor layout issue is that the front-panel audio connector might be a little too far away from the associated jacks for the cables of certain cases to reach. That’s really a chassis problem though, and it still affects many older enclosures. As a result, you see us pointing out when a motherboard manufacturer moves this connector into a more accessible position. Forward-facing SATA ports can also be problematic for certain older cases, though most folks prefer this design for its ability to avoid conflicts with long graphics cards.

A small bag of extender leads for the FM2-A85XA-G65’s voltage detection points adds a little value to MSI's installation kit, but four SATA cables is barely adequate for this eight-port board.
MSI continues to mix up the settings we'd rather be looking at, hiding the most important options with small fonts underneath large time and temperature readings.

The expected 4.5 GHz never materialized for this board, though a 102 MHz reference clock and 44x CPU multiplier pushed the same APU to a still-respectable 4.48 GHz.

Our target of 1.45 V for the processor was approximated at the board’s 1.45 V setting, but our memory reached its 1.65 V ceiling at the board’s 1.62 V setting. Like most of its competition in today's round-up, MSI isn't being accurate with its memory voltage. Could it be that all of these vendors are trying to unfairly secure a win in our DRAM overclocking charts? Regardless of the actual explanation we're correcting the settings of every board in today's competition to give us the proper voltage.

The FM2-A85XA-G65 provides primary and secondary memory timing control through its Advanced DRAM Configuration sub-menu. XMP modes aren’t selectable, but tuners willing to make manual adjustments will be happy to find a full timings report within its associated Memory-Z menu.

“CPU Core Vdroop Offset Control” within the FM2-A85XA-G65’s Digital Power sub-menu allows users to compensate for voltage changes due to changes in CPU load. The 60% setting kept our processor’s voltage almost perfectly stable under four threads of Prime95.
The first company to offer us a retail-ready motherboard with AMD's A85X FCH, Sapphire presents a couple unusual features on its Pure Platinum A85XT. An open-ended PCIe x4 slot is the most noteworthy change, though the fact that it's up top means you probably won't find a graphics card there. The PCIe x16 slot beneath it loses half of its connectivity to the second x16-length slot whenever a card is installed there.
The Bluetooth controller found on the Pure Platinum A85XT’s I/O panel is an unusual bonus among value-priced motherboards. Similarly unusual is the mini-PCIe x1 slot right in the middle of the Pure Platinum A85XT’s PCB, labeled for both PCIe and mSATA connectivity. One of the chipset’s eight SATA connections is tied up by the slot, reducing the number of standard SATA ports to seven.
Power, reset, and CLR_CMOS buttons along the Pure Platinum A85XT’s bottom edge aren’t unusual, nor is its Port 80 diagnostics display. On the other hand, we're far less accustomed to seeing a BIOS selector switch among those bottom-edge features. We've previously had little luck with auto-switching functionality, so we're ecstatic to find a manual switch on the Pure Platinum A85XT.
It almost appears that Sapphire designed this board to support some degree of form factor modularity. For example, a front-panel audio connector located above the second PCI Express x16 slot suggests a microATX-oriented design. As a result of its placement, you'll have to drag your audio cable up over the top of the board. Fortunately, even if your case suffers from short cable syndrome, it should still reach without a problem.
A bottom-edge USB 3.0 connector is really our only true layout concern. It’s far enough beneath the second graphics card slot to clear most GPU coolers, and far enough beneath the bottom PCI slot to clear most on-card circuits, but its distance from typical front-panel ports is a bit of a stretch for the cables included with some cases.

Anyone who’s unable to use the Pure Platinum A85XT’s front-panel USB 3.0 header with their case’s integrated ports might instead try installing the included 3.5” bay adapter. Failing that, the company also includes a slot bracket for the USB ports. Six internal SATA cables are also included in the Pure Platinum A85XT’s installation kit.
We don’t see too many motherboards from Sapphire (the company is best known for its graphics cards). So, we were pleasantly surprised at the simplicity of its Q-BIOS.

Unfortunately, we weren’t able to increase the multiplier ratio to overclock our APU, nor were we able to reboot after changing the processor core and northbridge ratios simultaneously. Instead, all of our overclocking efforts were limited to the APU's stock 38x ratio, and we used nothing other than the platform’s reference clock to control frequency.

Without a way to determine our memory kit's secondary timings, we were forced to manually configure them using screenshots from another motherboard’s firmware. That’s not going to be possible for most of Sapphire’s customers.

Our CPU core operated at 1.45 V under full load with the Pure Platinum A85XT set to 1.425 V and “CPU Loadline Enabled." We only reached 4.4 GHz on a CPU known to run at 4.5 GHz, but we already know our problem was a result of multiplier control issues, rather than voltage. If you want to alter the APU's multiplier without running into the same issue we encountered, switch to software-based overclocking or wait for Sapphire to release a firmware update.

We haven’t focused much on user profiles lately, except to note when they’re missing. Sapphire doesn’t lack this feature entirely, but does limit it to a single register.

| Test System Configuration | |
|---|---|
| CPU | AMD A10-5800K (Trinity): 3.8 GHz Base, Up to 4.2 GHz via Turbo Core, 8 MB Shared L3 Cache, Socket FM2 |
| CPU Cooler | Sunbeamtech Core-Contact Freezer w/Zalman ZM-STG1 Paste |
| RAM | Kingston KHX1600C9D3K2/8GX: 8 GB DDR3-1600 CAS 9-9-9-27 |
| Graphics | AMD Radeon HD 7660D: 800 MHz GPU, DDR3-1600 |
| Hard Drive | Samsung 470 Series MZ5PA256HMDR, 256 GB SSD |
| Sound | Integrated HD Audio |
| Network | Integrated Gigabit Networking |
| Power | Seasonic X760 SS-760KM: ATX12V v2.3, EPS12V, 80 PLUS Gold |
| Software | |
| OS | Microsoft Windows 7 Ultimate x64 SP1 |
| Graphics | AMD Catalyst 12.11 |
| Chipset | AMD Platform Driver 8.947.0.0 |
We used the Windows 7 software configuration from our recent Bulldozer performance story for today’s benchmarks, updating only the drivers and one game. We altered our game settings by one quality level as well to help maintain more playable performance from the A10's on-die graphics engine.

Our G.Skill DDR3-2200 memory kit defaults to DDR3-1600 CAS 9. Unfortunately, both the Sapphire and ECS boards wouldn't boot with those modules installed. We were instead forced to use an older, slower DDR3-1600 CAS 9 kit for our benchmarks, which those two boards defaulted down to DDR3-1333. Manual configuration gave us the proper performance we were expecting, but you can be sure we're going to take compatibility into consideration at the end.

ECS and Sapphire didn’t have the same compatibility issues with our G.Skill DDR3-2666 kit, probably because this kit defaults to DDR3-1333 settings. Lacking SPD values for DDR3-1600, this second set of memory was reserved exclusively for our overclocking tests.

We also threw our full range of keyboards and mice at these boards to make sure that they all worked in UEFI and required no special attention in Windows. After watching every board pass that test, we tried DOS mode through a boot disk and discovered mouse issues on the Asus and ECS platforms. In DOS mode, our Logitech mouse would only move vertically, while our Razer mouse moved inconsistently, switching randomly between normal and reverse movements.

The need for DOS mode compatibility is mostly limited to old DOS mode-based games. We’re not sure how many readers are affected, but those who do should take note of the issues we're seeing, which often take firmware updates to rectify.
| Benchmark Configuration | |
|---|---|
| 3D Games | |
| Battlefield 3 | Campaign Mode, "Going Hunting" 90-Seconds Fraps Test Set 1: Medium Quality Defaults (No AA, 4x AF) Test Set 2: High Quality Defaults |
| F1 2012 | In-Game Benchmark using Adrenaline Tool 1.0.0.12 Test Set 1: Medium Quality Preset, No AA Test Set 2: High Quality Preset, 4x AA |
| Skyrim | Update 1.5.26, Celedon Aethirborn Level 6, 25 Seconds Fraps Test Set 1: Medium Details Defaults Test Set 2: High Details Defaults |
| Adobe Creative Suite | |
| Adobe Photoshop CS6 (64-bit) | Version 13 x64: Filter 15.7 MB TIF Image: Radial Blur, Shape Blur, Median, Polar Coordinates |
| Adobe After Effects CS 6 | Version: CS5.5: Tom's Hardware Workload, SD project with three picture-in-picture streams, source video at 720p, Render Multiple Frames Simultaneously |
| Adobe Premiere Pro CS 5.5 | Video length 2min 21s, Export to H.264 Blu-ray Source 960x720, Output 1280x720 |
| Adobe Acrobat X Professional | V10.0.0, PDF Creation from PowerPoint 2010 Presentation (3.6 MB) |
| Audio/Video Encoding | |
| iTunes | Version 10.4.1.10 x64: Audio CD (Terminator II SE), 53 minutes, default AAC format |
| Lame MP3 | Version 3.98.3: Audio CD "Terminator II SE", 53 min, convert WAV to MP3 audio format, Command: -b 160 --nores (160 Kb/s) |
| HandBrake CLI | Version: 0.98, Video: Video from Canon EOS 7D (1920x1080, 25 frames) 1 Minutes 22 Seconds, Audio: PCM-S16, 48 000 Hz, Two-Channel, to Video: AVC1 Audio: AAC (High Profile) |
| MainConcept Reference | Version: 2.2.0.5440: MPEG-2 to H.264, MainConcept H.264/AVC Codec, 28 sec HDTV 1920x1080 (MPEG-2), Audio: MPEG-2 (44.1 kHz, Two-Channel, 16-Bit, 224 Kb/s), Codec: H.264 Pro, Mode: PAL 50i (25 FPS), Profile: H.264 BD HDMV |
| Productivity | |
| Blender | Version: 2.62: Syntax blender -b thg.blend -f 1, Resolution: 1920x1080, Anti-Aliasing: 8x, Render: THG.blend frame 1, Cycles renderer and internal tile renderer (9x9) |
| Visual Studio 2010 | Compile Chrome project (1/31/2012) with devenv.com /build Release |
| Autodesk 3ds Max 2012 | V14.1.0.328 x64: Space Flyby Mentalray, Frame 248, 1440x1080 |
| WinZip | Version 16.5 Pro: THG-Workload (1.30 GB) to ZIP, command line switches "-a -ez -p -r" |
| WinRAR | Version 4.20.0.0: THG-Workload (1.30 GB) to RAR, command line switches "winrar a -r -m3" |
| 7-Zip | Version 9.28: THG-Workload (1.30 GB) to .7z, command line switches "a -t7z -r -m0=LZMA2 -mx=5" |
| ABBYY FineReader | Version 10.0.102.82: Read PDF save to Doc, Source: Political Economy (J. Broadhurst 1842) 111 Pages |
We use averages for performance comparisons, but our Battlefield 3 benchmarks also produce minimum and maximum FPS results. We checked through the test notes and found that all six motherboards played the game smoothly (above 20 FPS minimum) at 1600x900 and medium details, while none of them could reliably push resolution or detail levels any higher.
When considering minimum performance, the difference between the top-performing Asus board and bottom-performing Sapphire platform card drops to 1 FPS.


Pushing details and resolutions beyond playable settings only helps us to confirm that all six manufacturers are competing fairly. Large wins would indicate something wrong, and large losses would indicate configuration problems. We're happy to see none of that going on.
We're testing F1 2012 at medium and high details to retain consistent settings across all three benchmarks, in spite of our knowledge going in that sim games like this one are generally easier on performance than the first-person shooter we just benchmarked.
It's hardly a surprise, then, to see all six motherboards yielding playable frame rates at medium details and 1920x1080, though the minimum 37 FPS rate indicated in our test notes is nearly two times higher than we were expecting.


Frame rates once again indicate very similar performance between all six platforms. We turn to our test notes and find the lowest minimum frame rate of 22 FPS to be only 3 FPS slower than that of the top performer. They’re all playable from integrated graphics at high quality and 1080p.
We keep turning to our test notes to check playability, and again find that every board plays the game smoothly from the APU's Radeon HD graphics core at 1600x900 and medium quality settings. Higher settings weren't smooth on any board, so we’re left to judge each based on non-gaming criteria.


A difference of 3 FPS does look fairly large when charted on a narrow performance scale, but we’re not concerned, since none of these resolutions are playable at Skyrim’s High quality preset. These averages become additional data for each board’s overall performance index.
Who would have expected ECS and Gigabyte to battle each other for the top spot in Lame? Differences this small primarily serve to keep us awake as we wade through pages of similar benchmark numbers.



Sapphire falls one second behind in Lame and HandBrake, but catches the pack in MainConcept.
Five of the six boards in today’s comparison yield identical Photoshop performance, while the sixth board falls only a second behind in this two-and-a-half minute test.


The two boards that finished at the top of our single-threaded Lame encoding results fall to the bottom of our After Effects bench.


ECS stumbles a little in the single-threaded Acrobat X test, so we retested and saw the A85F2-A Golden stumble yet again. We’re not sure the cause of that minor benchmark gap, but we do like to confirm the consistency of any atypical results.
Gigabyte edges out the rest of the field in our 3ds Max benchmark, but ASRock catches Gigabyte in ABBYY FineReader.




Asus has a small lead in our Visual Studio 2010 compilation of Google Chrome.
Sapphire falls behind in our 7-Zip file compression test, while ASRock and MSI surge ahead. MSI retains its lead through WinRAR, and Sapphire retains its deficit.



ASRock clips a second off the average WinZip compression time, while Sapphire adds a second.
ASRock's FM2A85X Extreme6 takes a fairly big lead when it comes to energy consumption under full load, but there’s a trick behind that reading: each board attempts to cram our CPU into a specific power envelope through dynamic underclocking. The other boards would eventually idle down by around 10 W, and give up around 400 MHz while doing that. ASRock’s FM2A85X Extreme6 idled down a little faster, and we weren’t able to get its “real” full-load reading.

MSI's FM2-A85XA-G65 runs the coolest during our extended full-load test, though each board throttles down AMD's APU during the run. Disabling power management features is one way to combat this behavior. However, in the real-world, you're going to want to run with them enabled to avoid wasting power. That's why we leave them enabled throughout the benchmarks.

Using an average of averages to compare performance to power consumption, we find ASRock’s 7% power savings worth an 8% efficiency increase. Noting that the board had already began to throttle back before we could get a load reading, we estimate that the FM2A85X Extreme6 would normally place around mid-pack. This is one of those rare situations where Chris’ extended power readings (total watts required to complete the entire benchmark series) would have been more-enlightening.

We don’t have to make an estimation to understand why MSI doesn't do well in terms of efficiency, since the power consumption chart shows high idle wattage.
| BIOS Frequency and Voltage settings (for overclocking) | |||
|---|---|---|---|
| ASRock FM2A85X Extreme6 | Asus F2A85-V Pro | ECS A85F2-A GOLDEN | |
| Reference Clock | 100-136 MHz (1 MHz) | 90-300 MHz (1 MHz) | 90-200 MHz (1 MHz) |
| CPU Multiplier | 14-63x (1x) | 8-63x (1x) | 5-63x (1x) |
| DRAM Data Rates | 800-1866 (266.6 MHz) | 800-2400 (266.6 MHz) | 800-2400 (266.6 MHz) |
| CPU Vcore | 0.60-1.55V (6.25 mV) | 0.68-2.08V (6.25 mV) | 1.50-1.98V (20 mV) |
| CPU NB | 0.60-1.55V (6.25 mV) | 0.50-1.90V (6.25 mV) | 1.30-1.80V (20 mV) |
| A85X Voltage | 1.10-1.40V (100 mV) | 1.10-1.40V (10 mV) | 1.11-1.21V (~35 mV) |
| DRAM Voltage | 1.17-1.80V (5 mV) | 1.35-2.135V (5 mV) | 1.20-2.01V (10 mV) |
| CAS Latency | 5-14 Cycles | 5-16 Cycles | 5-14 Cycles |
| tRCD | 5-19 Cycles | 2-19 Cycles | 2-19 Cycles |
| tRP | 5-19 Cycles | 5-19 Cycles | 5-19 Cycles |
| tRAS | 15-40 Cycles | 8-42 Cycles | 8-42 Cycles |
| BIOS Frequency and Voltage settings (for overclocking) | |||
|---|---|---|---|
| Gigabyte F2A85X-UP4 | MSI FM2-A85XA-G65 | Sapphire Pure Platinum A85XT | |
| Reference Clock | 100-200 MHz (1 MHz) | 90-190 MHz (1 MHz) | 100-300 MHz (1 MHz) |
| CPU Multiplier | 8-79x (1x) | 8-63x (1x) | 16-58x (1x) |
| DRAM Data Rates | 800-2400 (266.6 MHz) | 800-2133 (266.6 MHz) | 800-2133 (266.6 MHz) |
| CPU Vcore | 0.80-2.10V (6.25 mV) | 1.20-1.90V (12.5 mV) | 1.20-1.70V (6.25 mV) |
| CPU NB | 0.80-2.10V (6.25 mV) | 1.00-1.50V (12.5 mV) | 1.20-1.54V (6.25 mV) |
| A85X Voltage | 1.00-1.40V (10 mV) | 0.91-1.51V (~11.5 mV) | 1.10-2.30V (10 mV) |
| DRAM Voltage | 1.10-2.62V (10 mV) | 1.29-2.01V (15 mV) | 1.30-2.30V (10 mV) |
| CAS Latency | 5-16 Cycles | 5-16 Cycles | 5-14 Cycles |
| tRCD | 2-19 Cycles | 2-19 Cycles | 2-19 Cycles |
| tRP | 5-19 Cycles | 5-19 Cycles | 2-19 Cycles |
| tRAS | 8-42 Cycles | 8-42 Cycles | 8-40 Cycles |
Some motherboards have far broader frequency ranges than others, but a 300 MHz reference clock limit on Sapphire’s motherboard is about as realistic as a 300 MPH speedometer on a Honda Civic. PCIe and integrated GPU overclocking limits are far more restrictive.

Asus, ASRock, and Gigabyte support our 45 x 100 MHz setting, with the difference between each vendor's board attributable to reference clock rounding errors.

Gigabyte reaches the highest reference clock with its integrated GPU intact. The second-place Asus F2A85-V Pro achieves a 156 MHz base clock with full CPU stability, but reference clocks over 130 MHz destabilize the APU’s integrated Radeon HD engine. Similar problems on the PCIe bus would affect graphics cards in a similar manner.

The F2A85-V Pro take top honors in memory overclocking, allowing us to use its highest 24x memory multiplier with four DIMMs installed. ASRock doesn't support a high memory ratio, but the board is able to reach a high data rate through increases in reference clock.
Asus provides the best overclocking on its exorbitantly-priced F2A85-V Pro, while ASRock offers the best price on its FM2A85X Extreme6, which doesn't overclock as well. Charting performance-per-dollar might it easier for some folks to choose a winner, but we also need to discuss the feature set each platform offers at its price point.

Setting a $100 baseline, the FM2A58X Extreme6 includes a x16-length PCIe x4 slot, an extra USB 3.0 controller, internal power and reset buttons, a Port 80 diagnostics display, and an I/O panel-based CLR_CMOS button. All of those features have gotten cheaper over the past few years, but that combination should still be worth at least $20 over a board that lacks them. Unfortunately, we don’t have any $80 baseline boards to compare.
Instead, we jump up $10 in price to the MSI FM2-A58XA-G65. MSI buyers gain the firm’s OC Genie button, but lose out on ASRocks’ integrated Port 80 diagnostics display. MSI buyers gain a PCI slot, but lose a PCIe x4 slot. MSI buyers gain a cooler-running voltage regulator, but lose out on ASRock’s extra pair of I/O panel-based USB 3.0 ports. These boards offer similar apparent worth, so the extra $10 hurts MSI's value proposition.
Gigabyte’s F2A85X-UP4 appears to be a higher-quality part than ASRock’s. Its feature set is nearly identical though, and that makes it hard for us to justify a $25 price premium.
The A85F2-A Golden from ECS actually loses features compared to ASRock’s model, and we can’t use its gold-plated connectors to excuse the $27 price increase.
Compared to ASRock’s sample, the Sapphire Pure Platinum A85XT gains a PCI slot, moves its PCIe x4 slot to a place where it’s useful only for single-slot cards, gains an mSATA/mini-PCIe combo connector, and adds dual BIOSes with manual switching. This editor would love to give Sapphire an award based on that manual firmware selector alone, but the tiny switch and the spare ROM cannot justify the board’s $35 higher price. Even if Sapphire were to drop its price by 10%, a mere one-year warranty would continue to diminish its award worthiness.
And so, we pick up at the end of the conclusion where we left off at the beginning, trying to choose between the best-in-class Asus F2A8F-V Pro and the best-value ASRock FM2A85X Extreme6. We even have awards for the best-in-class and best-value categories. Best of Tom’s Hardware is an exclusive award, though. Nevertheless, we're happy to put both boards on even footing with Tom's Hardware Approved recognition. Between the two, the platform you choose will depend on the system you're trying to build.




















