ASRock X299 Taichi Motherboard Review

[MAC]

Feature Testing: ASRock RGB LED

Feature Testing: ASRock RGB LED

Click on image to enlarge​

If you're not a fan of RGB LED lighting, the X299 Taichi might just be the motherboard for you. The ASRock RGB LED feature on this motherboard is essentially limited to a few LEDs placed under the chipset cooler. However, if you want to expand that there are two RGB headers on which you can plug 5050 RGB LED strips up 12V/3A (36W), and which can be controlled from within the ASRock RGB LED utility.

These RGB LEDs can be controlled using the aforementioned utility or even a special page in the UEFI BIOS. The lights can be adjusted to a number of different colours and customized to create cool lighting effects. The presets can cause the LEDs to change shades breathe, strobe, cycle through all the colours, fade in and out, flash on and off, just statically display one colour, and more. You can also adjust the speed at which these LEDs turn on and off.

Click on image to enlarge​

With RGB LEDs in only one area you obviously aren't going to be witnessing any magical light shows on this motherboard. However, the LEDs are bright and colourful, and they certainly add a little razzle-dazzle to a system. Those wanting more definitely have that opportunity thanks to the two onboard light strip headers.

Here is a little live action look at RGB LED feature on the ASRock X299 Taichi:

<iframe width="700" height="394" src="https://www.youtube.com/embed/mNoCl5fNg8E?rel=0" frameborder="0" allowfullscreen></iframe>​

 

[MAC]

Feature Testing: Onboard Audio

Feature Testing: Onboard Audio

Since fewer and fewer consumers seem to be buying discrete sound cards, the quality of a motherboard's onboard audio is now more important than ever. As such, we figured that it was worthwhile to take a closer look at just how good the analog signal quality is coming out of the onboard Purity Sound 4 audio subsystem that is implemented on the X299 Taichi. As mentioned earlier, this model features the modern Realtek ALC1220 codec, one Texas Instruments op-amp, Nichicon Fine Gold audio capacitors, and a PCB-level isolation line.

Since isolated results don't really mean much, but we have also included some numbers from the plethora of motherboards that we have previously reviewed. All of the motherboards that we have included are feature onboard audio solutions that are built around the Realtek ALC1150 or ALC1220 codecs, but feature different op-amps, headphone amplifiers, filtering capacitors, secondary components and layouts.

We are going to do this using both quantitative and qualitative analysis, since sound quality isn't really something that can be adequately explained with only numbers. To do the quantitative portion, we have turned to RightMark Audio Analyzer (RMAA), which the standard application for this type of testing.

Since all modern motherboards support very high quality 24-bit, 192kHz audio playback we selected that as the sample mode option. Basically, what this test does is pipe the audio signal from the front-channel output to the line-in input via a 3.5mm male to 3.5mm male mini-plug cable, and then RightMark Audio Analyzer (RMAA) does the audio analysis. Obviously we disabled all software enhancements since they interfere with the pure technical performance that we are trying to benchmark.

​

As you can see, when compared to the ASUS STRIX X299-E, the X299 Taichi achieved numbers that were a little worse in some aspects, about equal in others, and a fair bit better in the areas of total harmonic distortion (THD) and intermodulation distortion (IMD) plus noise. What this means in global terms is that this new ASRock motherboard has achieved the third best audio number that we have ever seen, behind only the aforementioned X299-E and the miniature ASUS STRIX Z270I.

As we have mentioned in the past, we aren't experts when it comes to sound quality, but at this high level we suspect that just about anyone should be satisfied. We listened to a variety of music and spoken word content using a mix of Grado SR225i and Koss PortaPro headphones, Westone UM1 IEMs, and Logitech Z-5500 5.1 speakers, and the playback was clean and loud. Frankly, we have no criticisms whatsoever. We know that most owners will likewise be very happy with this motherboard's onboard audio capabilities, especially those who primarily use headphones since the front output is amplified.

 

[MAC]

Feature Testing: M.2 PCI-E 3.0 x4

Feature Testing: M.2 PCI-E 3.0 x4

When compared to the previous LGA2011-v3 platform, LGA2066 has significantly improved the availability of high-speed storage interfaces. Not only does the new X299 PCH have an impressive 24 PCI-E 3.0 lanes, which is three times as much as the 8-lane X99 PCH, but it also features a faster link to the processor (8GT/s DMI3 vs. 5GT/s DMI2). The end result of this every X299 motherboard has at least two M.2 slots and they are all full-speed M.2 PCI-E 3.0 x4 implementations. The ASRock X299 Taichi on the other hand has three M.2 slots since the engineers have cleverly decided to use four of the CPU's PCI-E 3.0 lanes to expand high-speed storage capabilities.

​

While few SSDs exist that can reach the 3.5-3.6GB/s real-life limit of this interface, we settled on one that can crack the 2000MB/s barrier: the Samsung SSD 950 PRO 256GB. Despite now being usurped by the SSD 960 PRO, this high performance NVMe PCI-E SSD combines Samsung's powerful UBX controller with its industry-leading 3D V-NAND and is capable of sequential read speeds of up to 2,200MB/second and write speeds of up to 900MB/sec.

One of the ways that we will be evaluating the performance of a motherboard's M.2 interface is by verifying that is capable of matching or exceeding these listed transfer rates. The other is by checking to see whether it performs as well as when we install the SSD 950 PRO onto a ASUS Hyper M.2 x4 expansion card plugged directly into a PCI-E 3.0 x16 slot. The PCI-E lanes that the M.2 slot requires can come from either the processor or more usually the X299 PCH, and we are interested to see how well that lane splitting was implemented and whether it is causing any performance issues. Also, since there are two M.2 slots, we are interested in determining whether there is a performance difference between both of them.

M.2 Top vs. M.2 Middle vs. M.2 Bottom vs PCI-E​

As you can see the performance of the three M.2 slots was excellent, and it's essentially identical to the performance with the PCI-E adapter. The top M.2 slot - the one that gets it's PCI-E lanes directly from the processor - did consistently achieve slightly better results, but the difference is so small that it's well within benchmark variances.

While transfer rates are obviously an important metric, we figured that it was also worthwhile to take a peak at instructions per second (IOPS) to ensure that there wasn't any variance there either:

M.2 Top vs. M.2 Middle vs. M.2 Bottom vs PCI-E​

The results were a little more all over the place in this test in the sense that no one slot or adapter consistently performance the best, but the differences are essentially non-existent and well within the margin of error for this benchmark. Overall, we think that it is fair to say that the M.2 interfaces on the X299 Taichi have been well implemented in so far as they all perform roughly identically.

When we reviewed the ASUS STRIX X299-E we mentioned that the M.2 performance was measurably worse than on the mainstream LGA1151 platform. ASUS has since released a new UEFI BIOS that supposedly addresses this issue, though we haven't had the chance to re-benchmark that model. Either way, we can report that the X299 Taichi is faster than the ASUS is some specific areas, but also a bit slower in others, so M.2 performance on this platform is a bit of a mixed bag at the moment.

 

[MAC]

Auto & Manual Overclocking Results

Auto & Manual Overclocking Results

Though it might feature a new socket, this LGA2066 platform is still fundamentally identical to the LGA2011 series that came before it when it comes to overclocking. Much like Broadwell-E, the new Skylake-X processors are still based on a 14nm process, so you can definitely still theoretically use up to 1.35V CPU core voltage pretty safely. However, that only applies to the i7-7800X and i7-7820X. When it comes to the i9-7900X you can't use that much voltage given the incredible amount of heat it outputs when overclocked and overvolted. Personally, even with an absolutely top-notch dual-fan air cooler or high-end dual-fan AIO we wouldn't recommend more than 1.25V, and even that much voltage will be problematic for most cooling solution. In order to avoid creating extra heat, we also recommend being conservative with the other system voltages, namely the cache/mesh voltage, the system agent voltage (VCCSA) and the I/O voltage (VCCIO). Ideally, you shouldn't need to apply more than 1.15V to the cache/mesh while still being able to reach a ~3200MHz frequency, while near default VCCSA and VCCIO values of 0.95V and 1.05V should allow you to reach memory speeds of up to DDR4-3733.

The rules for Kaby Lake-X are unsurprisingly similar to mainstream Kaby Lake. Our personal pointers are to increase the vCore up to around 1.35V if you're cooling can handle it, while increasing the VCCIO up to 1.20V, and the System Agent voltage up to 1.25V if you plan on increasing the cache or memory frequency. If you are trying to achieve the highest possible DDR4 memory speeds, increasing the VCCIO to 1.30V and vSA to 1.35V might be worth trying out. These last two are really only needed if you plan on seriously pushing the uncore/cache frequency or the memory frequency.

All of the LGA2066 processors are multiplier and BCLK unlocked, but unless you're trying to extract every last megahertz there's no reason to go crazy increasing the BCLK above 103-105Mhz since you can achieve similar results by just tweaking the various multipliers instead.

Lastly, we highly recommend that you avoid stress testing Skylake-X with the latest build of Prime 95. The simple fact of the matter is that due to its use of AVX its puts an entirely unrealistic load on the CPU, which causes both CPU and VRM temperatures to skyrocket until one or the other will start throttling. If you aim for Prime 95 stability/temperatures you will be robbing yourself of hundreds of megahertz of overclocking potential.

Auto Overclocking​

The X299 Taichi motherboard supports three types of automatic overclocking, one software-based and two found within the UEFI. Within the multi-purpose ASRock F-Stream software suite there is the EZ OC feature, which is a semi-intelligent approach to automatic overclocking. It can accessed by selecting the Performance Mode option and then clicking on the Advanced sub-menu. There aren't really any available options, you just need to click on the Start button and the utility starts off the overclocking process at default clocks and slowly increases to the the frequency until a reasonable sweet spot is found.

Within the UEFI BIOS, there are two different modes - EZ Mode and Advanced Mode - both of which offer distinct automatic overclocking features. In the EZ Mode, there is a CPU EZ OC button that once you click and save will automatically apply an overclock preset. In the Advanced Mode, there is the Optimized CPU OC setting, with four available options, ranging from Turbo 4.2GHz to Turbo 4.8GHz. This is another feature that relies on presets, so it can't customize the overclock to best suit your particular system. On the plus side, it is as quick and easy as selecting the desired option and exiting the UEFI.

With all of that said, let's start off with the software-based EZ OC feature:

Click on image to enlarge​

As mentioned above, in the EZ OC section of the F-Stream app there is option to select presets of up to 4.8GHz when a Core i9-7900X is installed. As always, we tried that most aggressive preset and while we were able to boot into Windows and even pass a 16-thread HyperPI stress test it would consistently crash during the Cinebench multi-threaded test. This preset set a high 1.319Vcore which is more than sufficient to theoretically stabilize our i-7900X at 4.8Ghz, the temperatures were absolutely very problematic with that much voltage. As a result, we ended up choosing the second most aggressive preset - Turbo 4.6Ghz - which proved to be perfectly stable throughout our tests, at least in part due to its more reasonable 1.26Vcore. The memory speed was not touched, nor was the cache frequency, so those voltages were not touched.

When we swapped in a Core i7-7740X, the presets went up to 5.0GHz and our chip has absolutely no issues hitting that, especially with 1.312V. Once again, neither the memory cache or cache frequency were overclocked.

Next up we will take a look at one of the two UEFI-based automatic overclocking options:

Click on image to enlarge​

The first of two types of automatic overclocking found in the UEFI is the CPU EZ OC button. Since it's found in the EZ Mode part of the UEFI, this feature is super simple; you just click on the icon, save & exit, the system reboots and the overclock is applied. In the case of the i9-7900X, lightly threaded workloads pushed the frequency up to a max of 4.3GHz at 1.17Vcore, while heavily multi-threaded workloads saw the cores running at 4.0GHz at 1.08Vcore. When we installed the i7-7740X and enabled the CPU EZ OC feature all the cores just ran at 4.6GHz at 1.260-1.264V no matter the type of workload. Overall, this feature might be super-duper simple and worthwhile for novice users or those will mediocre cooling, but the software-based EZ OC feature is distinctly superior if you're more interested in gaining some additional performance.

Click on image to enlarge​

In the Advanced Mode section of the UEFI, there is the Optimized CPU OC setting with four available presets options that vary depending on which processor you have installed. Fundamentally speaking this is exactly the same feature as the aforementioned software-based EZ OC feature. Once again, we were not able to run our i9-7900X at 4.8Ghz due to instability and temperature issues, however the less aggressive 4.6Ghz worked without issues. Likewise, we were once again able to apply the Turbo 5.0GHz to our Core i7-7740X.

Manual Overclocking

Click on image to enlarge​

As you can see, we were able to push the Core i9-7900X to 4.7GHz with our self-imposed 1.25Vcore limit. With an eye towards further managing temperatures, we also set a -1 AVX offset and a -2 AVX-512 offset, which limits the cores to 4.6GHz and 4.5GHz when those particular extensions are used. When it came time to overclock the cache/mesh and memory, we simply set the cache voltage to 1.15V, system agent voltage to 0.95V and the I/O voltage to 1.02V. This allowed us to increase the cache/mesh frequency from 2400MHz to 3200MHz - which provides a VERY nice performance boost - and we overclocked the memory from DDR4-3200 14-14-14 to DDR4-3733 16-16-16. We could have likely pushed the memory much higher, but since we are already tickling the 100GB/s memory read bandwidth mark, more is not really necessary.

Our dainty little Core i7-7740X proved its high frequency capabilities by hitting 5.2GHz with a relatively conservative 1.35V. The temperatures were still very manageable - somewhere in the low 70°C range - so 5.3 or 5.4GHz is certainly not outside the range of possibility for those willing to give it some additional voltage. The uncore/cache hit a wall at modest 4400MHz with 1.35VCCSA and 1.300VCCIO, but on the plus side we were able to push our memory kit all the way up to DDR4-4000, albeit with fairly loose timings.

Overall, overclocking on this motherboard was a problem-free experience. Even despite the monstrous power demands of our overclocked Core i9-7900X we never felt the need to baby the motherboard or give it any active cooling. We are now a full three months after launch day, so there's been ample UEFI updates and all the kinks appear to have been ironed out (Prime95 VRM load aside). Obviously, the automatic overclocking results speak for themselves, they are as high as you could possibly want on either of these processors...and even perhaps higher than you can reasonably cool in the case of the Skylake-X chip.

 

[MAC]

System Benchmarks

System Benchmarks

In the System and Gaming Benchmarks sections, we reveal the results from a number of benchmarks run with the Core i7-7740X and Core i9-7900X on X299 Taichi motherboard. These tests were run at default clocks, with the best automatic overclock, and using our own manual overclock. This will illustrate how much performance can be achieved with this motherboard in stock and overclocked form. For a thorough comparison of the Core i7-7740K versus a number of different CPUs have a look at our "The Intel Kaby Lake-X i7-7740X Review" article.

SuperPi Mod v1.9 WP

When running the SuperPI 32MB benchmark, we are calculating Pi to 32 million digits and timing the process. Obviously more CPU power helps in this intense calculation, but the memory sub-system also plays an important role, as does the operating system. We are running one instance of SuperPi Mod v1.9 WP. This is therefore a single-thread workload.

wPRIME 2.10

wPrime is a leading multithreaded benchmark for x86 processors that tests your processor performance by calculating square roots with a recursive call of Newton's method for estimating functions, with f(x)=x2-k, where k is the number we're sqrting, until Sgn(f(x)/f'(x)) does not equal that of the previous iteration, starting with an estimation of k/2. It then uses an iterative calling of the estimation method a set amount of times to increase the accuracy of the results. It then confirms that n(k)2=k to ensure the calculation was correct. It repeats this for all numbers from 1 to the requested maximum. This is a highly multi-threaded workload.

Cinebench R15

Cinebench R15 64-bit
Test1: CPU Image Render
Comparison: Generated Score

The latest benchmark from MAXON, Cinebench R15 makes use of all your system's processing power to render a photorealistic 3D scene using various different algorithms to stress all available processor cores. The test scene contains approximately 2,000 objects containing more than 300,000 total polygons and uses sharp and blurred reflections, area lights and shadows, procedural shaders, antialiasing, and much more. This particular benchmarking can measure systems with up to 64 processor threads. The result is given in points (pts). The higher the number, the faster your processor.

WinRAR x64

WinRAR x64 5.40
Test: Built-in benchmark, processing 1000MB of data.
Comparison: Time to Finish

One of the most popular file archival and compression utilities, WinRAR's built-in benchmark is a great way of measuring a processor's compression and decompression performance. Since it is a memory bandwidth intensive workload it is also useful in evaluating the efficiency of a system's memory subsystem.

FAHBench

FAHBench 1.2.0
Test: OpenCL on CPU
Comparison: Generated Score

FAHBench is the official FAH benchmark that measures the compute performance of CPUs and GPUs. It can test both OpenCL and CUDA code, using either single or double precision, and implicit or explicit modeling. The single precision implicit model most closely relates to current folding performance.

Since Intel's OpenCL drivers don't yet support Skylake-X processors, we had to use AMD's APP SDK 3.0 tool, which had no such limitations.

HEVC Decode Benchmark v1.61

HEVC Decode Benchmark (Cobra) v1.61
Test: Frame rates at various resolution, focusing on the top quality 25Mbps bitrate results.
Comparison: FPS (Frames per Second)

The HEVC Decode Benchmark measures a system's HEVC video decoding performance at various bitrates and resolutions. HEVC, also known as H.265, is the successor to the H.264/MPEG-4 AVC (Advanced Video Coding) standard and it is very computationally intensive if not hardware accelerated. This decode test is done entirely on the CPU.

LuxMark v3.1

Test: OpenCL CPU Mode benchmark of the LuxBall HDR scene.
Comparison: Generated Score

LuxMark is a OpenCL benchmarking tool that utilizes the LuxRender 3D rendering engine. Since it OpenCL based, this benchmark can be used to test OpenCL rendering performance on both CPUs and GPUs, and it can put a significant load on the system due to its highly parallelized code.

Once again, since Intel's OpenCL drivers don't yet support Skylake-X processors, we had to use AMD's APP SDK 3.0 tool, which had no such limitations.

PCMark 10

PCMark 10 is the latest iteration of Futuremark’s system benchmark franchise. It generates an overall score based upon system performance with all components being stressed in one way or another. The result is posted as a generalized score. In this case, we tested with both the standard Conventional benchmark and the Accelerated benchmark, which automatically chooses the optimal device on which to perform OpenCL acceleration.

AIDA64 Memory Benchmark

AIDA64 Extreme Edition is a diagnostic and benchmarking software suite for home users that provides a wide range of features to assist in overclocking, hardware error diagnosis, stress testing, and sensor monitoring. It has unique capabilities to assess the performance of the processor, system memory, and disk drives.

The benchmarks used in this review are the memory bandwidth and latency benchmarks. Memory bandwidth benchmarks (Memory Read, Memory Write, Memory Copy) measure the maximum achievable memory data transfer bandwidth. The code behind these benchmark methods are written in Assembly and they are extremely optimized for every popular AMD, Intel and VIA processor core variants by utilizing the appropriate x86/x64, x87, MMX, MMX+, 3DNow!, SSE, SSE2, SSE4.1, AVX, and AVX2 instruction set extension.
The Memory Latency benchmark measures the typical delay when the CPU reads data from system memory. Memory latency time means the penalty measured from the issuing of the read command until the data arrives to the integer registers of the CPU.

 

[MAC]

Gaming Benchmarks

Gaming Benchmarks

Futuremark 3DMark (2013)

3DMark v1.1.0
Graphic Settings: Fire Strike Preset
Rendered Resolution: 1920x1080
Test: Specific Physics Score and Full Run 3DMarks
Comparison: Generated Score


3DMark is the brand new cross-platform benchmark from the gurus over at Futuremark. Designed to test a full range of hardware from smartphones to high-end PCs, it includes three tests for DirectX 9, DirectX 10 and DirectX 11 hardware, and allows users to compare 3DMark scores with other Windows, Android and iOS devices. Most important to us is the new Fire Strike preset, a DirectX 11 showcase that tests tessellation, compute shaders and multi-threading. Like every new 3DMark version, this test is extremely GPU-bound, but it does contain a heavy physics test that can show off the potential of modern multi-core processors.

Futuremark 3DMark 11

3DMark 11 v1.0.5
Graphic Settings: Extreme Preset
Resolution: 1920x1080
Test: Specific Physics Score and Full Run 3DMarks
Comparison: Generated Score


3DMark 11 is Futuremark's very latest benchmark, designed to tests all of the new features in DirectX 11 including tessellation, compute shaders and multi-threading. At the moment, it is lot more GPU-bound than past versions are now, but it does contain a terrific physics test which really taxes modern multi-core processors.

Futuremark 3DMark Vantage

3DMark Vantage v1.1.2
Graphic Settings: Performance Preset
Resolution: 1280x1024

Test: Specific CPU Score and Full Run 3DMarks
Comparison: Generated Score

3DMark Vantage is the follow-up to the highly successful 3DMark06. It uses DirectX 10 exclusively so if you are running Windows XP, you can forget about this benchmark. Along with being a very capable graphics card testing application, it also has very heavily multi-threaded CPU tests, such Physics Simulation and Artificial Intelligence (AI), which makes it a good all-around gaming benchmark.

Valve Particle Simulation Benchmark

Valve Particle Simulation Benchmark
Resolution: 1920x1080
Anti-Aliasing: 4X
Anisotropic Filtering: 8X
Graphic Settings: High

Comparison: Particle Performance Metric

Originally intended to demonstrate new processing effects added to Half Life 2: Episode 2 and future projects, the particle benchmark condenses what can be found throughout HL2:EP2 and combines it all into one small but deadly package. This test does not symbolize the performance scale for just Episode Two exclusively, but also for many other games and applications that utilize multi-core processing and particle effects. This benchmark might be a little old, but is still very highly-threaded and thus will keep scaling nicely as processors gain more and more threads. As you will see the benchmark does not score in FPS but rather in its own "Particle Performance Metric", which is useful for direct CPU comparisons.

X3: Terran Conflict

X3: Terran Conflict 1.2.0.0
Resolution: 1920x1080
Texture & Shader Quality: High
Antialiasing 4X
Anisotropic Mode: 8X
Glow Enabled

Game Benchmark
Comparison: FPS (Frames per Second)

X3: Terran Conflict (X3TC) is the culmination of the X-series of space trading and combat simulator computer games from German developer Egosoft. With its vast space worlds, intricately detailed ships, and excellent effects, it remains a great test of modern CPU performance. While the X3 Reality engine is single-threaded, it provides us with an interesting look at performance in an old school game environment.

Final Fantasy XIV: Heavensward Benchmark

Final Fantasy XIV: Heavensward
Resolution: 1920x1080
Texture & Shader Quality: Maximum IQ
DirectX 11
Fullscreen

Game Benchmark
Comparison: Generated Score

Square Enix released this benchmarking tool to rate how your system will perform in Heavensward, the expansion to Final Fantasy XIV: A Realm Reborn. This official benchmark software uses actual maps and playable characters to benchmark gaming performance and assign a score to your PC.

Grand Theft Auto V

DirectX Version: DirectX 11
Resolution: 1920x1080
FXAA: On
MSAA: X4
NVIDIA TXAA: Off
Anisotropic Filtering: X16
All advanced graphics settings off.

In GTA V, we utilize the handy in-game benchmarking tool. We do ten full runs of the benchmark and average the results of pass 3 since they are the least erratic. We do additional runs if some of the results are clearly anomalous. The Rockstar Advanced Game Engine (RAGE) is ostensibly multi-threaded, but it definitely places the bulk of the CPU load on only one or two threads.

Middle-earth: Shadow of Mordor

Resolution: 1920x1080
Graphical Quality: Custom
Mesh/Shadow/Texture Filtering/Vegetation Range: Ultra
Lighting/Texture Quality/Ambient Occlusion: High
Depth of Field/Order Independent Transparency/Tessellation: Enabled

With its high resolution textures and several other visual tweaks, Shadow of Mordor’s open world is also one of the most detailed around. This means it puts massive load on graphics cards and should help point towards which GPUs will excel at next generation titles. We do three full runs of the benchmark and average the results.

 

[MAC]

Voltage Regulation / Power Consumption

Voltage Regulation

Despite the fact that this is a high-end desktop platform, there are almost no LGA2066 motherboards that have any onboard voltage measurement points, and that is the case with the X299 Taichi. It is regrettable since that is our preferred method of accurately measuring the various system voltages. As a result, in this abbreviated overview, we utilized the AIDA64 System Stability Test to put a very substantial load on the system while also monitoring the stability of the all-important CPU Vcore line. This was achieved with a 60+ minute stress test, and in order to increase the strain on the motherboard's voltage regulation components we overclocked our Core i7-7700K to 4.6Ghz at 1.25V (in the BIOS).

​

As you can see, there are two rails listed above in AIDA64. The CPU core is the CPU input voltage (in this case 1.80V) while the CPU VID is the Vcore, which we had set to 1.25V. While the input voltage would occasionally drop down to 1.798V the Vcore itself never wavered from 1.250V. We did not touch the Load-Line Calibration (LLC) settings, everything aside from the CPU multiplier and the Vcore were at default settings. Overall, this motherboard does a fantastic job of maintaining a perfectly stable Vcore, even when under the strain of a heavily overclocked Core i9-7900X.

Power Consumption

For this section, every energy saving feature was enabled in the BIOS and the Windows power plan was changed from High Performance to Balanced. For our idle test, we let the system idle for 15 minutes and measured the peak wattage through our UPM EM100 power meter. For our CPU load test, we ran Prime 95 In-place large FFTs on all available threads, measuring the peak wattage via the UPM EM100 power meter. For our overall system load test, we ran Prime 95 on all available threads while simultaneously loading the GPU with 3DMark Vantage - Test 6 Perlin Noise.

If we compare the X299 Taichi's power consumption numbers to that of ASUS STRIX X299-E, the ASRock wins by quite a large margin. Having said that, it's not just a matter of having a more efficient VRM, there's definitely some behind the scenes BIOS trickery occuring. For example, if you look at the stock i9-7900X numbers, the ASUS board's CPU load numbers are almost 200W higher, which tells us that ASUS are allowing the processor to operate unrestricted of power concerns under AVX workloads, whereas the ASRock motherboard is clearly limited power consumption. The manual overclock numbers are also significantly lower on the ASRock, despite both motherboards being setup essentially identically. When looking at the Core i7-7740X, there's once again a fairly sizeable advantage for the X299 Taichi over the STRIX X299-E.

 

[MAC]

Conclusion - A Good, But Lacking X299 Solution

Conclusion - A Good, But Lacking X299 Solution

If you're short on time and just jumped straight to the conclusion, here is a rundown of the ASRock X299 Taichi's specifications: a 13-phase CPU power design, high-quality 12K capacitors, four steel-reinforced physical PCI-E 3.0 x16 slots with support for 3-way SLI and CrossFireX, one PCI-E 3.0 x1 slot, ten SATA 6Gb/s ports, and three full-speed PCI-E 3.0 x4 M.2 slots. There are also two high-speed USB 3.1 Gen2 ports, up to eight USB 3.0 ports, and up to six USB 2.0 headers. Rounding things out are two Intel-powered gigabit LAN ports, an onboard Intel 802.11ac Wi-Fi module plugged into a M.2 E-Key slot, two physical BIOS chips, and a debug LED display.

While connectivity is clearly one of the X299 Taichi's strong suits, there are a few caveats along the way. For starters, the onboard Wi-Fi - while nice to have - is a paltry 433Mbps 1x1 solution, which is a little slow at this point. At a minimum, a higher-end 867Mbps 2x2 solution should be mandatory at this price range. Next up of the caveats list, only two of three M.2 slots support RAID, since those two are connected to the X299 chipset while the third is linked to the processor (more on this later). On a similar note, only eight of the ten SATA ports support RAID, since the remaining two ports are powered by a separate ASMedia SATA controller.

One of the more unique aspects of this motherboard is that in order to support three full-speed M.2 slots, ASRock needed to divert four of the CPU's PCI-E lanes towards the topmost M.2 slot when it is occupied with a PCI-E SSD. While doing this with a 44-lane or even 28-lane Skylake-X processor doesn't come with any significant penalties - they still support up to x8/x8/x16/x8 and x8/x0/8/x8 respectively - the situation is very different if you install a 16-lane Kaby Lake-X chip. In that scenario, the primary PCI-E x16 slot will be limited to x8 and the secondary PCI-E x16 slot will no longer just be limited to x4, it will be disabled entirely.

This inability to run two-way SLI (CrossFire is still works) with a Kaby Lake-based processor is obviously much worse than the x8/x8 breakdown of any mainstream LGA1151 motherboard. Is this a serious issue though? Not really, because in our opinion this is an HEDT platform that should be primarily focused on utilitizing the capabilities of the higher-end Skylake-X processors, and not on kowtowing to the limitations of two lower-end processors that ideally shouldn't even exist. Lastly, if you're one of the literally dozens of people planning on building a Kaby Lake-X system with two graphics cards and three PCI-E M.2 SSDs, just know that you're going to have to reign in your ambitions.

When it came time to test the Purity Sound 4 onboard audio, we were excited because designs based on the Realtek ALC1220 codec have been getting better and better. The audio numbers were frankly fantastic, good enough to place in the Top 3 best results that we have ever seen from an onboard audio solution. We appreciate the fact that they have added a headphone amplifier for the front-panel audio output since that is where people are most likely to be plugging into their headset/headphones/ear buds. Overall, this is a high-end audio experience.

When it comes to the RGB LED lighting feature, we like the fact that ASRock used really beefy RGB LEDs – much larger than we have seen before – that can really put out a ton of light. Both the software utility and the UEFI sub-menu gave us ample control over the lighting effects and colours. We do think that some LEDs should have been added to more than one location, but at least there are two LED light strip headers for those who truly care about aesthetics.

When it came time to overclock, the X299 Taichi proved itself to be perfectly capable. The three automatic overclocking features worked very well, capable of achieving results ranging from 4.2GHz to all the way up to 4.8GHz with a Core i9-7900X. We were very surprised to see that 4.8GHz option and elected not to use it since it set an elevated 1.30-1.31Vcore. In our opinion, that is too high for a day-to-day system since there's no way to keep the temperatures under control. With that in mind, we settled on the 4.6Ghz option that set the Vcore to 1.25V, which could still easily overload most cooling solutions when running any AVX workload.

Nevertheless, it absolutely blazed throughout our benchmark suite and it was perfectly stable. Like our manual overclock, we used an identical Vcore but pushes the CPU to 4.7Ghz was stable as rock as well. On the memory front we pushed our 32GB G.Skill Trident Z DDR4-3200 kit up to DDR4-3733, which was good enough for over 100GB/s of memory read bandwidth. Pretty incredible! The single 8-pin CPU power connector did not prove to be a limitation in our case, but we were undoubtedly getting close to the limit, a secondary connector would be needed for those pushing even higher than we did. Another mild complaint is the given the advanced VRM components that ASRock are using, we wish there was a way to monitor MOSFET temperatures from within the UEFI or A-Tuning utility. Instead we had to resort to measuring the MOSFET heatsink temperature...and it was pretty damn hot. This motherboard would be benefit from a heatsink with significantly more mass and surface area. Thankfully, as you'll read below, ASRock clearly agrees with us.

Overall, we really liked the X299 Taichi, especially at a retail price of $270 USD. It looks good, sounds good, performs well, and it's got top-notch connectivity. Having said that, we found the MOSFET heatsink to be a little too puny for our liking and the single 8-pin CPU power connector is barely sufficient when it comes to handling the power demands of very heavily overclocked 10-core+ Skylake-X processors. Thankfully, both of those shortcomings have already been fixed by the refreshed X299 Taichi XE, which has a much beefier VRM cooling solution and a second 8-pin EPS connector. We expect the original Taichi to disappear completely once supplies are exhausted, leaving only the superior refresh model.

Overall, while the X299 Taichi is great motherboard with a few downsides, the X299 Taichi XE fixes the issues and is otherwise identical. As a result, while we aren't awarding our Dam Good Award to the X299 Taichi, we see absolutely no reason why the X299 Taichi XE wouldn't win that acclaim.