Asus’ X99 Deluxe provided our first taste of the next-gen goodness available in Haswell-E and its associated X99 chipset. The board pairs the platform’s prodigious I/O bandwidth with loads of auxiliary peripherals, wicked-fast Wi-Fi, and a bunch of extra features and accessories. It’s the motherboard equivalent of a burger loaded with multiple patties, bacon, fried onions, sautéed mushrooms, three kinds of cheese, and a fried egg… in addition to all the usual fixin’s.
Although the Deluxe is delicious, it’s a little over the top. The nearly $400 asking price is also tough to swallow for those who won’t fully exploit the board’s excess. For most enthusiasts, the X99-A looks like a more manageable meal. Asus’ most affordable Haswell-E offering distills the Deluxe down to its essential ingredients without losing sight of the platform’s strengths. Asking price: only $274.99 online.
Like its Deluxe daddy, the X99-A follows a monochromatic theme. Asus didn’t just strip a few chips off the higher-end model and slap on the name, though. The X99-A has a different PCB with a reconfigured slot layout. The electrical components seem to be the same, but their deployment is slightly different—and, in some cases, more sparing.
One of the biggest layout changes is the relocation of the M.2 slot. Instead of poking out vertically near the ATX power connector, the X99-A’s mini SSD mount runs parallel to the PCB in the bottom-right corner of the board. That location still provides enough room to host M.2 22110 drives up to 110 mm (4.3″) long.
Like Asus’ other Haswell-E boards, the X99-A has a proprietary “OC Socket” with more pins than the standard LGA2011-v3 design prescribed by Intel. These pins make contact with untapped pads on the processor, and they purportedly prevent unwanted voltage drops in highly overclocked systems. Asus’ promotional literature references a CPU running at 1.8V, an extreme voltage even for chips swimming in liquid nitrogen, so it’s unclear how much benefit there is for conventional overclockers bound by off-the-shelf coolers—and stuck with retail chips rather than cherry-picked samples.
For what it’s worth, Asus also claims the OC Socket enables higher memory speeds and tighter timings. The X99-A supports DDR4-3000 speeds even with all eight of its DIMM slots populated.
DDR4 modules are loaded into matching rows that flank the CPU to the east and west. A VRM heatsink looms to the north, while graphics cards threaten from the south, putting the socket in the center of a very crowded region. We can’t test every hardware combination for compatibility, so we’ve provided some clearance measurements below. We’re now taking measurements from the center of the socket, which should make them easier to compare to the dimensions published by cooler makers. Thanks to TR reader Meadows for the suggestion.
The VRM heatsinks are short enough to stay mostly out of the way. However, memory with taller heat spreaders could conflict with aftermarket coolers that branch out from the socket.
Note that we’ve measured the distances to two sets of memory slots. The pair farther from the socket is occupied by four-DIMM configs, while the closer couple is needed only with six- and eight-DIMM setups.
Six expansion slots match the total on the Deluxe, but only the first, fourth, and sixth connect to the CPU, capping CrossFire and SLI support at three cards. Multi-GPU teaming schemes experiencing diminishing returns after two cards, so the lack of four-way support isn’t a serious limitation. At least the slot spacing leaves enough room for three-way configs with dual-slot coolers and duallie combos with triple-slot coolers.
The distribution of PCI Express lanes to the expansion slots and onboard peripherals depends on multiple factors, including whether the CPU has 28 or 40 PCIe lanes enabled. (The Core i7-5820K is restricted to 28 lanes, while the 5930K and 5960X have 40.) We’ve mapped the possible configurations in the diagram below. The slot order matches the image above, and clicking the buttons adjusts the distribution for different CPUs.
The M.2 and last PCIe x16 slot share Gen3 lanes from the CPU. That arrangement is more about switching than sharing, though. The four lanes available to the M.2 can be re-routed to fuel the expansion slot, but the two can’t be used simultaneously.
On the chipset front, the first x1 and second x16 slots share Gen2 bandwidth with the auxiliary USB controller. Assigning four lanes to the x16 slot via the firmware effectively disables its companions, which otherwise get one lane each. The X99’s remaining lanes are solely devoted to the second x1 slot, the LAN controller, and the SATA Express port.
SATA Express brings us nicely to the X99-A’s storage payload, which we’ll explore on the next page.
Storage, ports, and the little things
Because the X99 chipset has 10 SATA 6Gbps ports, the X99-A can host a stack of drives without resorting to third-party controllers. There are some associated limitations, though. Only six ports are managed by Intel’s Rapid Storage Technology software, which has built-in RAID support. The remaining four (the ones on the left in the picture below) are separate from the RST framework. These second-class citizens work just fine with individual drives, but they can only participate in RAID arrays via third-party software.
Two of the full-fat SATA ports reside inside a slim SATA Express connector. This shared physical interface is tied to a “flex I/O” link in the chipset that can hook into dual SATA drives or one SATAe device. The latter has access to 1GB/s of bandwidth from dual PCIe Gen2 lanes, much like similar flex implementations on Z97 boards.
The X99-A’s M.2 slot is part of the storage picture, too, but it bypasses the chipset and hooks into the CPU. Mini SSDs have access to a staggering 4GB/s of bandwidth—enough headroom for not only today’s fastest M.2 drives, but also tomorrow’s next-gen hotness. The only catch is that the CPU has no clue what to do with SATA-based SSDs, ruling out compatibility with most of the M.2 drives on the market right now. And that’s just fine, because PCIe SSDs have a lot more potential. Besides, the X99-A has enough SATA connectivity already.
Both of the X99-A’s internal USB 3.0 headers are visible in the image above. Each one is connected to dual ports in the X99 chipset, leaving only two native SuperSpeed ports for the rear cluster pictured below. Asus bolsters those connections with a two-port controller and one-to-four hub, bringing the number of rear-mounted USB 3.0 ports up to six.
|1 Gigabit Ethernet via Intel I218-V|
|8-channel audio via Realtek ALC1150 and amplifier|
|4 USB 2.0 via X99|
|2 USB 3.0 via ASMedia ASM1042AE controller|
|3 USB 3.0 via X99 and ASMedia ASM1074 hub|
|1 USB 3.0 via X99|
The table above illustrates how the various ports are connected. The ports supplied by the ASMedia controller are slower than the other USB 3.0 options, so we wouldn’t recommend them for truly high-speed devices. That said, the ASMedia chip still hit 250-260MB/s in our sequential tests, so it’s certainly fast enough for mechanical storage.
Intel Gigabit Ethernet? Check.
CMOS reset button? Nope. What looks like a reset switch is actually the trigger for USB Flashback, which can update the motherboard firmware with only a thumb drive and power supply connected. Good guess, though.
On the audio front, Realtek’s ALC1150 codec collaborates with Texas Instruments’ R4580 amplifier. The amp can be switched between the stereo and front-panel outputs on the fly, and it’s joined by the usual assortment of enhancements, including audio-specific capacitors and isolated analog traces. More importantly, the drivers support DTS UltraPC II surround virtualization and DTS Connect multi-channel encoding.
The onboard audio sounds decent to my ears, and the X99-A scores well in RightMark Audio Analyzer, which measures analog signal quality. If you’re really serious about audio quality, though, you’re better off using the digital output or installing a discrete sound card or USB DAC.
Although they’re not terribly exciting, builder-friendly features like the X99-A’s cushioned I/O shield and front-panel wiring blocks make system assembly much easier. The shield won’t slice your fingers or get caught up in the rear ports, and the blocks simplify front-panel wiring immensely. Both should be standard equipment on all enthusiast-oriented motherboards.
Just behind the port blocks is a series of switches to control features like XMP profiles, EPU power saving, and TPU auto-overclocking. These features can be activated in the firmware, of course, but some folks apparently prefer onboard switches. The X99-A also has physical power and reset buttons, a POST code display, and a DirectKey header that can be used to boot directly into the firmware. Unless you want to short the DirectKey header with a screwdriver, you’ll need a momentary switch mechanism (like those used for chassis power and reset buttons) to trigger the boot-to-firmware shortcut.
So ends our tour of the X99-A’s hardware, but there’s a lot more to this board than what’s on the PCB. Next, we’ll look at the firmware, the software, and how well this thing overclocks.
Firmware and software tweaking options
The X99-A’s firmware is pretty much identical to that of the X99 Deluxe. Apart from a few platform-specific options, the UEFI is also the same as what’s available on Asus’ Z97 boards. Since our X99 Deluxe and Z97-A reviews already provide detailed tours, we won’t spend too much time rehashing them here. There are, however, a few finer points that are worth reiterating.
The first is the slick usability of Asus’ motherboard firmware. The options are laid out intuitively, the graphical fan controls are excellent, and the tuning wizard steps uninitiated users through the auto-overclocker and RAID setup. Using the UEFI feels more like navigating Windows software than poking around in a firmware interface.
There’s no shortage of fine-tuning options beneath the user-friendly facade. Advanced overclocking and memory tuning options are easily accessible, and the fan section is loaded with extra goodness. For example, users can define whether the temperature used to drive each fan’s rotational speed is drawn from one of multiple onboard sensors or from a two-pin header that accepts external probes. The fan control logic works with both three- and four-pin spinners, and independent controls are available for five of the six onboard headers. (The CPU and CPU_OPT headers share the same profile.) There’s even a built-in fan speed calibrator.
Asus’ Dual Intelligent Processors 5 tweaking software brings much of the firmware’s functionality to Windows. The utility even adds a few unique elements, like the ability to define spin-up and spin-down times for temperature-controlled fans. Hardware monitoring is built into the software, as well, complete with configurable warning thresholds for voltages, temperatures, and fan speeds.
One of the best parts of DIP5 is the configurable auto tuner, which lets users adjust a wide range of variables governing how the automated overclocking mechanism cranks up frequencies—and how it tests for stability at each step along the way. This optimization wizard can also adjust power settings and fan profiles automatically, and it can even push GPU frequencies on compatible graphics cards.
We put DIP5’s auto-overclocking intelligence to the test with a Core i7-5960X processor strapped to a mammoth Cooler Master Nepton 280L water cooler. The configuration options were left at their stock settings except for the duration of the stress test, which we increased from 15 seconds to 90 seconds.
After a few reboots, the tuning mechanism settled on a 4.5GHz clock speed driven by a 45X all-core multiplier and a 1.308V CPU voltage. The system was stable under a combined CPU and GPU load, and temperatures stayed under about 70°C.
4.5GHz matches the highest speed we’ve achieved with this particular CPU-and-cooler combo. The fact that we hit that frequency using the software auto-tuner is especially impressive—and good news for newbies who don’t want to get their hands dirty with manual tweaking.
Speaking of which, we spent a little time fiddling with various overclocking knobs to see if our CPU would hit 4.6GHz. We made it to the Windows desktop at that speed, but we couldn’t come up with a mix of voltage and power settings that kept the rig stable under load. The time we spent tweaking did, however, confirm that Asus’ firmware and software are great for making manual adjustments.
The only issue we encountered was with the software’s hardware monitor, which reported a CPU voltage 0.02V higher than what was set in the utility (and reported by CPU-Z). That’s a pretty minor problem in the grand scheme of things, especially since an earlier version of CPU-Z made a similar error.
When equipped with the same components running at the same speeds, motherboards typically have little impact on system performance. There are rare exceptions, so we still run a range of application and peripheral tests to look for outliers, but we didn’t find any with the X99-A. In most cases, the performance differences between this board and the X99-Deluxe are smaller than the run-to-run variance.
Gigabyte’s X99-UD4 also appears in the graphs below. That board’s DRAM multipliers only go up to 26.66X, so the UD4 can’t run our DDR4-2800 modules at full tilt without also overclocking the CPU. We tested the UD4 at DDR4-2666 speeds to ensure stock CPU clocks, and that handicap was barely evident outside our memory bandwidth benchmark.
Nothing to see here, folks.
We also measured cold boot times, and the X99-A made it to the desktop slightly faster than the other Haswell-E boards we’ve tested:
Shaving a few seconds off the boot time is nice, but it’s hard to get excited about a task that’s performed so infrequently. If you really want quick boot times, you’re better off with Z97 boards, which typically boot in 13-16 seconds with similar hardware.
Although motherboards have little effect on performance, they can influence power consumption. We measured total system power draw (sans monitor and speakers) at the wall socket over a five-minute idle period and under a load comprising Unigine’s Valley graphics benchmark and Cinebench’s multi-core CPU rendering test. The Asus boards were tested with and without their EPU power-saving measures enabled.
The X99-A is particularly power-hungry at idle, where it consumes 3W more than its Deluxe sibling and 16W more than the Gigabyte board. The contest is closer under load, but even with its EPU mojo turned on, the X99-A still draws 11W more than our low-power leader.
Asus’ LGA2011 boards have a history of higher power draw, so the results aren’t terribly surprising, even if they are a little disappointing. At least the deltas are small enough that cooling requirements shouldn’t be affected.
Here’s the X99-A’s full spec sheet in case we missed anything.
|Platform||Intel X99, socket LGA2011-v3|
|DIMM slots||8 DDR4, 64GB max|
|Expansion slots||3 PCIe 3.0 x16 via CPU (x16/x16/x8*)
1 M.2 22110 PCIe 3.0 x4 via CPU (*shared with PCIe x16)
1 PCIe 2.0 x16 via X99
2 PCIe 2.0 x1 via X99
|Storage I/O||1 SATA Express via X99 (shared with 2 SATA RAID 6Gbps)
6 SATA RAID 6Gbps via X99
4 SATA 6Gbps via X99
|Audio||8-channel HD via Realtek ALC1150 with amplifier
Real-time multichannel S/PDIF encoding via DTS Connect
Surround virtualization via DTS Ultra PC II
|Ports||1 USB 3.0 via X99
3 USB 3.0 via X99 and ASMedia ASM1074 hub
2 USB 3.0 via ASMedia ASM1042
4 USB 3.0 via internal headers and X99
4 USB 2.0 via X99
4 USB 2.0 via internal headers and X99
1 Gigabit Ethernet via Intel I218-V
1 analog front/headphone out (amplified)
4 configurable analog ports
1 digital S/PDIF out
|Overclocking||All/per-core Turbo multiplier: 31-80X
Min/max CPU cache ratio: 12-80X
CPU strap: 100, 125, 167, 250MHz
Base clock: 80-300MHz
Base:DRAM ratio: 100:100, 100:133
CPU voltage: 1.0-2.0V
CPU cache voltage: 1.0-2.0V
System agent voltage: 0.8-2.0V
CPU input voltage: 0.8-2.7V
DRAM A/B voltage: 0.8-1.9V
DRAM C/D voltage: 0.8-1.9V
PCH core voltage: 0.7-1.8V
PCH I/O voltage: 1.2-2.2V
VCCIO CPU 1.05 voltage: 0.7-1.8V
VCCIO PCH 1.05 voltage: 0.7-1.8V
VTTDDR A/B voltage: 0.2-1.0V
VTTDDR C/D voltage: 0.2-1.0V
PLL termination voltage: 0.2-2.1938V
|Fan control||2 x CPU (combined), 4 x SYS:
Standard, silent, full-speed profiles
Manual profile with source temp, three temp/speed points per fan
DC and PWM fan support
The PCIe configurations are for 40-lane CPUs. See the first page of the review for the lane assignments for 28-lane chips.
Our testing methods
If you’ve made it this far, you may be curious about what our test system looks like. Here it is:
We used the following configurations for testing.
|Processor||Intel Core i7-5960X|
|Motherboard||Asus X99-A||Asus X99 Deluxe||Gigabyte X99-UD4|
|Platform hub||Intel X99||Intel X99||Intel X99|
|Chipset drivers||Chipset: 10.0
|Audio||Realtek ALC1150||Realtek ALC1150||Realtek ALC1150|
|Memory size||16GB (4 DIMMs)|
|Memory type||Corsair Vengeance LPX DDR4-2800 SDRAM|
|Memory config||2800MHz @ 16-18-18-36-2T||2800MHz @ 16-18-18-36-2T||2666MHz @ 16-18-18-36-2T|
|Graphics||Asus GeForce GTX 680 DirectCU II with 340.52 drivers|
|Storage||Corsair Force Series GT 120GB
Samsung 840 Pro 256GB
|Power supply||Corsair AX850 850W|
|OS||Microsoft Windows 8.1 Pro x64|
Thanks to Asus, Cooler Master, Corsair, Intel, and Samsung for providing the hardware used in our test systems. And thanks to the motherboard makers for providing those.
We used the following versions of our test applications:
- 7-Zip 9.20 64-bit
- TrueCrypt 7.1a
- Chrome 34.0.1847.131
- x264 r2431
- DiRT Showdown demo
- CrystalDiskMark 3.0.3b
- FRAPS 3.5.99
- RightMark Audio Analyzer 6.2.5
- Cinebench R15
- Unigine Valley 1.0
Some further notes on our test methods:
- All testing was conducted with motherboard power-saving options enabled. These features can sometimes lead to slightly slower performance, particularly in peripheral tests that don’t cause the CPU to kick into high gear. We’d rather get a sense of motherboard performance with real-world configurations, though; we’re not as interested in comparing contrived setups with popular features disabled.
- DiRT Showdown was tested with ultra detail settings, 4X MSAA, and a 1920×1200 display resolution. We used Fraps to log a 60-second snippet of gameplay from the demo’s first race. To offset the fact that our gameplay sequence can’t be repeated exactly, we ran this test five times on each system.
- Power consumption was measured at the wall socket for the complete system, sans monitor and speakers, using a Watts Up Pro power meter. The full-load test combined Cinebench’s multithreaded CPU rendering test with the Unigine Valley DirectX 11 demo running in a 1280×720 window. We reported the peak power consumption during the Cinebench run. Our idle measurement represents the low over a five-minute period sitting at the Windows desktop.
- The Force GT 120GB SSD was used as the system drive for all tests. The Samsung 850 Series 512GB was connected as secondary storage to test Serial ATA and USB performance, the latter through a USAP-compatible Thermaltake BlacX 5G docking station. The Samsung SSD was secure-erased before each test that involved it. The Corsair drive was also wiped before we loaded our system image.
- Ethernet performance was tested using a remote rig based on an Asus P8P67 Deluxe motherboard with an Intel 82579 Gigabit Ethernet controller. A single Cat 6 Ethernet cable connected that system to each motherboard.
- Analog audio signal quality was tested using RMAA’s “loopback” test, which pipes front-channel output through the board’s line input. We tested while the system was loaded with Cinebench’s multithreaded rendering test, the Unigine Valley benchmark, and a CrystalDiskMark 4KB random I/O test running on the Samsung SSD attached via USB 3.0.
The tests and methods we employed are publicly available and reproducible. All tests except power consumption were run at least three times. Unless otherwise indicated, we reported the median result for each test. If you have questions about our methods, hit our forums to talk with us about them.
Asus’ Z97-A is my favorite board of that generation, so I had high hopes for its X99 counterpart. The X99-A doesn’t disappoint. It has everything we loved in the Z97-A—a smart spec, builder-friendly features, robust audio, great overclocking options, and excellent fan speed controls—grafted to an even more potent platform. The X99-A delivers just about everything Haswell-E and its X99 sidekick have to offer, too. The only thing missing is four-way CrossFire and SLI support, a feature of questionable value outside competitive overclocking circles.
Rather than piling on toppings to add flavor, the X99-A focuses on the main ingredients. If this were a burger, the patty would be a blend of grass-fed sirloin, chuck, and short rib. There would be only one kind of cheese, aged cheddar perhaps, and the lettuce and tomato would be garden fresh. The bun would be right out of the oven, and instead of being slathered in special sauce, it would be deftly spread with a hand-whipped aioli. And the bacon, because you’ve always gotta have bacon, would be double-smoked at least.
The X99-A isn’t so much a case of less being more as it is an example of the right things being done really well. Seasoned enthusiasts should appreciate the subtleties, and even though some of the finer points will be lost on newbies, the sum of all those little touches results in a smoother building, setup, and overclocking experience for all.
Only two elements of the X99-A give me pause. The first is the relatively high idle power consumption, which probably won’t be a concern for most folks building Haswell-E rigs. The other is the $274.99 asking price, which is a little higher than the going rate for X99 boards with comparable specs.
For me, the premium experience is worth the mark-up for experienced veterans and first-time builders alike. The X99-A is an excellent board—and a TR Editor’s Choice.