Simply NUC Onyx NUC13OXv9 Review: 4×4 Mini PC with Intel Core i9-13900H

The Simply NUC Onyx range of mini PCs feature feature Intel’s 13th generation Raptor Lake “H” series processors and use the familiar 4×4 inch “NUC” form factor. They’re some of the newest systems from Simply NUC, a company that specialises in providing small form factor computing solutions for business and enterprise customers.

The star of the lineup is the NUC13OXv9 model that sells for $999 and up, and features an Intel Core i9-13900H processor, making it one of the first 4×4 “NUC” mini PCs with a Core i9 chip. We wanted to know what kind of performance users can expect from this small computer with such a powerful processor, so Simply NUC provided Liliputing with hardware for review purposes. The unit we tested was provided to Liliputing free of charge, without any requirement to return it, but this review is not sponsored by Simply NUC, and Simply NUC did not modify or approve the content of our review in any way.

Design & Specs

The Simply NUC Onyx has a 117 x 112 x 54 mm (4.61 x 4.41 x 2.13 inches) square plastic case with an internal metal frame for strength and rigidity. It has a detachable plastic top and metal bottom. As an actively cooled mini PC, it is available with a choice of either an Intel Core i5-13500H, Core i7-13700H or Core i9-13900H processor.

The review model’s Intel Core i9-13900H mobile processor has 14 cores in total, comprising of 6 “performance” cores which can boost to 5.40 GHz and 8 “efficient” cores which can boost to 4.10 GHz. Together they provide a total of 20 threads.

The processor also includes an integrated GPU from the Iris Xe Graphics family which has 96 execution units and a maximum dynamic frequency of 1.50 GHz. There is also support for vPro which enables remote access to the mini PC and is typically used by IT departments to manage the device and provide software and security updates.

Looking at the device front-on, there is a pin-hole digital microphone, a pin-hole illuminated disk activity light, a USB Type-A 2.0 port, a USB Type-A 3.2 Gen 2×1 (10 Gbit/s) port, a 3.5mm headphone jack and an illuminated power button.

On the right-side there is a Kensington Lock slot.

There’s a microSD card port on the left side.

The bulk of the ports are on the rear. These include, from left to right:

• 1 x power jack
• 1 x USB4 Type-C port (above)
• 1 x HDMI 2.1 port (below)
• 1 x 2.5 Gb Ethernet port
• 2 x USB Type-A 3.2 Gen 2×1 (10 Gbit/s) ports (stacked)
• 1 x USB4 Type-C port (above)
• 1 x HDMI 2.1 port (below)

There is also a removable plate which can be replaced with one containing additional ports for I/O expansion, although these are currently not available yet.

The plastic top, which has a dull matt and fingerprint-proof finish, can be removed simply by prising it off to expose a metal fan grill and two metal strips either side which form the Wi-Fi aerials.

Additionally, if Windows was purchased as part of the configuration, the Microsoft licence sticker will be found on the underside of the top.

Optionally the top can be replaced with one containing additional ports, again for I/O expansion, including further USB, Ethernet, HDMI, VGA or RS232 ports.

Turning the device upside down, once the four retaining screws with built-in rubber feet are loosened, the metal base can be carefully lifted off, as the underside is connected to the motherboard with a SATA ribbon cable. The base includes a metal heat sink and thermal pad to make contact with an M.2 2280 drive if present.

The base also includes a 2.5″ SATA HDD (7 mm) slot. Attaching a drive is very easy as all that is required is to slide the drive into the bay. It can then be secured in place by replacing the four rubber grommets in the base with the screws included with the mini PC.

Helpful instructions are etched into the underside of the base plate as they are omitted from the Quickstart Guide.

The motherboard has dual SO-DIMM slots that support DDR5 memory.

For the review model they were populated with two Kingston 8 GB DDR5 5600 MHz sticks of memory configured to run at 5200 MHz, which is the highest speed supported by the Core i9-13900H processor.

There is also an M.2 2280 PCIe Gen 4 x4 NVMe SSD slot which was occupied by a Kingston OM8PGP4256Q-A0 256 GB drive in the review model.

Underneath the NVMe drive is a M.2 2230 slot containing a Intel Killer Wi-Fi 6E AX1675x (210NGW) card that also provides Bluetooth Version 5.3 (the LMP Firmware Version shows as LMP 12.13675).

The motherboard also includes several expansion headers including a USB 2.0 header, a USB 3.0 header, a COM header for RS-232/422/485 and most importantly, a 4-pin RGB header (although it is labelled as a second USB 2.0 header on the board).

The box containing the mini PC also includes the power supply and cable with a country specific plug, VESA mounting plate, a packet of screws and an English diagrammatic instruction sheet which is a very basic guide on how to set up the mini PC.

Unlimited performance?

It is well established that high powered processors produce a lot of heat. Mini PC manufacturer’s often employ Occam’s razor in addressing heat issues by simply reducing the processor’s power limits. With the NUC13OXv9, the default value for Power Limit 1 is set to 45 W which is the same as the “Processor Base Power”. The value for Power Limit 2 is set to 54 W even though the “Maximum Turbo Power” is 115 W.

As such, this will compromise the performance as it is leaving power on the table, in part however, because it just cannot use all of it due to limitations of the cooling system.

Simply NUC claims that “the Onyx delivers unmatched computing performance for modern business.” So far only two companies have released 4×4 inch “NUC” mini PCs with an Intel Core i9-13900H processor. Later I will cover the performance as measured by benchmarks and real-world experiences, and then look at verifying this claim through a comparison against the GEEKOM Mini IT13, which is the only other 4×4 “NUC” mini PC available with a Core i9-13900H processor.

But first, a brief discussion about the UEFI (BIOS).

Somewhat disappointingly it is just an Aptio configured version from American Megatrends. Unlike the later Intel’s NUC versions, which are polished and fine-tuned to the individual NUC model, and which are familiar to existing NUC owners, the NUC13OXv9 uses different keys to access and navigate.

Additionally there is nothing in the scant documentation provided with the mini PC that covers which are the most useful options available in it. In particular, from the Advanced menu page, under Power & Performance, there is a setting “PowerLimit Setting” where you can either manually set the Power Limits, or use one of the convenient preset modes of Silent, Balance or Performance.

Balance Mode is the default and will set Power Limit 1 to 45 W plus Power Limit 2 to 54 W, Performance Mode will set Power Limit 1 to 55 W together with Power Limit 2 to 64 W, and Silent Mode will set Power Limit 1 to 35 W and Power Limit 2 to 45 W to minimise the running of the fan. For anyone needing it, as the Simply NUC splash screen does not include any details, the Delete key gets you into the UEFI (BIOS) and the “Boot Menu” key is F7.

How it performs

The review model came with Windows 11 Pro Version 22H2 OS build 22621.900 which was activated once connected to the internet. However, I wanted to test using Version 23H2 and despite all updates being applied, I was not offered the required upgrade option.

So I forced the upgrade by downloading and running the Windows 11 Installation Assistant. Once upgraded and updated I was on OS build 22631.2715. But because Windows was installed with an OEM licence, the upgrade resulted in Windows reporting that no product key was found on the device, resulting in Windows no longer being active. It was necessary to prise off the lid to get to the Microsoft sticker, and then change the product key to the one originally installed as written on the sticker.

Now, with an activated Windows once again, I was able to personalise the layout and configure everything to my liking. I then shrank the Windows partition to create a new 50 GB partition into which I installed Ubuntu 22.04.3.

Some of the usual problem areas for mini PCs include Wi-Fi, Bluetooth, and audio output from the 3.5mm jack. But those all worked without any issues on both Windows and Ubuntu on the NUC13OXv9.

I started performance testing by using Crystal Dew World’s CrystalDiskMark to measure performance on Windows and a script using the “fio” or “flexible I/O tester command” to verify it on Ubuntu.

The drive’s performance was rather disappointing as the write speeds (≈ 1850 MB/s) were half the read speeds (≈ 3950 MB/s) and not as fast as expected from a PCIe Gen 4 drive. These speeds were also reproducible on Ubuntu confirming the drive’s slow writability.

For the micro SD card reader, I tested using both UHS-I and UHS-II cards. Unfortunately it didn’t really matter as the performance was even more disappointing than with the NVMe drive. Internally the NUC13OXv9 uses a Realtek USB 2.0 Card Reader which gave speeds of just under 30 MB/s for both read and write regardless of which card was used, and this despite “the mini PC with its powerful processor will save you hours when processing your photographs, only if you come back tomorrow after they’ve finished loading” said no one ever. If you want faster speeds it is better to purchase a USB 3.0 (USB 3.2 Gen 1×1) card reader as I was able to get ten times the speed compared to the included reader.

The USB Type-A ports all performed at expected speeds, although I did experience some difficulties in getting a good connection on the rear slots with cables whose metal sleeve was fractionally shorter than those that worked without issue. All the “problem” cables were longer than 12 inches (30 cm) and Simply NUC have advised that this might be the cause of the problem, which is a shame as they work with other devices.

The two USB4 ports are marked as 20 Gb/s however when tested on Windows they ran at 40 Gb/s.

Ubuntu was interesting as although they showed up as 40 Gb/s, with 2 lanes at 20 Gb/s, when a USB4 device was connected, it received data at 40 Gb/s but could only transmit data at 20 Gb/s.

The main inconvenience on Ubuntu was that whilst USB 3.x and Thunderbolt 3 devices were plug and play, USB4 devices only worked if they were plugged in prior to booting. This is currently an OS issue rather than a device/mini PC issue.

The following benchmarks were all run using the default power setting of Balance Mode. For reference, the average CPU mark for PerformanceTest as listed on PassMark’s website is 29781.

On Windows I ran:

• PassMark Software’s PerformanceTest (general performance)
• UL’s 3DMark (CPU and graphics) and Procyon (office productivity)
• Maxon’s Cinebench (CPU)
• Primate Labs’s Geekbench (CPU and graphics)
• Unigine’s Heaven (graphics)

On Ubuntu I only ran the following benchmarks:

• PassMark Software’s PerformanceTest (CPU and memory)
• Primate Labs’s Geekbench (CPU)
• Unigine’s Heaven (graphics)
• Thomas Kaiser’s “sbc-bench” (server performance)

The “sbc-bench” results can be viewed online at http://ix.io/4Men.

Whilst the Ubuntu results for the relevant benchmarks are similar to those on Windows, it must be noted that by default, Windows render uses Direct3D 11 graphics (Direct3D11) in the Heaven benchmark whereas Ubuntu uses Open Graphics Library (OpenGL) so a direct comparison cannot be made.

Interestingly, even just using Balance Mode, the NUC13OXv9 CPU Mark of 29862.6 is right on the average. The overall PassMark Rating is low however as it has been impacted by the performance of the NVMe drive whose Disk Mark was only 22823.1 due to the slow write speed.

For real-world testing of the integrated graphics, I played various YouTube videos in Edge on Windows and there were no issues encountered when playing videos up to 1080p 60FPS. However, from there up to 4K 60FPS, the videos kept dropping frames at a rate of between 1% and 2%. In contrast, Chrome had no issues. On Ubuntu, both Firefox and Chrome played YouTube videos flawlessly although I didn’t try any resolution greater than 4K 60FPS.

Networking is provided by either Ethernet or Wi-Fi. The 2.5 Gb Ethernet port when tested on Ubuntu averaged 2.22 Gbits/sec upload and 2.35 Gbits/sec download.

The Intel Killer Wi-Fi 6E AX1675x (210NGW) card which supports dual-stream Wi-Fi in the 2.4GHz, 5GHz and 6GHz bands, was also tested on Ubuntu by connecting to a Wi-Fi 6 router. The upload speed for a 2.4 GHz connection was 213 MBits/sec and the download was 191 MBits/sec. However, using the 5 GHz band the upload speed was a very impressive 1.40 GBits/sec with download at 1.46 GBits/sec.

A common criticism of some mini PCs is that the fan is loud. However loudness can be perceived differently by individuals for a number of reasons, the most obvious one being age. The frequency of a sound can alter both the perception of loudness and whether a sound is subjectively interpreted as being annoying. All of which is a long-winded way of saying what I hear may not be what you will hear.

Typically after booting, the fan starts and whilst idling, it cannot be measured by my sound meter next to the device, although a faint whirring sound can be heard. When under load, such as a stress test on Ubuntu, the fan noise increases and can reach up to around 40.9 dBA in Balance Mode.

Running a stress test in Performance Mode obviously generates more heat and the fan noise increases reaching up to 49.7 dBA. Silent Mode is just that. If you put your ear to the device you can just make out the fan operating very quietly.

Running a stress test is impressive because for the most part, the NUC13OXv9 stays silent. Although this mode relies on thermal throttling, occasionally it has to run the fan slightly faster for short bursts just to ensure the CPU temperature remains in the high 80s and below 90°C. These short couple of second bursts can be measured at around 31.4 dBA and are hardly noticeable.

To visually see the effectiveness of the cooling in Balance Mode, I monitored running Cinebench R23 Multi-Core on Windows.

I also ran a stress test of all cores on Ubuntu and recorded the key CPU numbers for Utilization, Frequency, Temperature and Power.

Once the temperature goes over 70°C the initial frequency drops from 3800 MHz to an average of 3500 MHz. Although the temperature continues to climb, it peaks occasionally at 80°C but stays at around 79°C.

Upping the Power Limits

Whilst the NUC13OXv9’s PassMark CPU Mark is already similar to scores reported for a Core i9 13900H processor, the Cinebench R23 Multi Core score of 15063 is low compared to the average reported on several websites: 17959 (Notebookcheck), 17864 (NanoReview), 18760 (CPU-Monkey).

So I set the PowerLimit Setting to Performance Mode and on running Cinebench R23 I got a Multi Core score of 17237, which is still slightly lower. I then manually set both Power Limit 1 and 2 to 80 W, to see the effect on the score. It improved to 17815 which is still just shy of the reported averages. I then reran the benchmark with monitoring on, which showed that the maximum temperature hit 93°C.

During both runs the fan noise reached 53 dBA and subjectively it was loud but in an acceptable way. Realistically, the Performance Mode setting is perfectly adequate to provide the balance of power versus noise, whilst still leaving some headroom for anyone wanting to push performance a little bit further using manual settings.

Whilst running in Performance Mode might be good for generating high CPU benchmark scores, I really wanted to see how it would affect office productivity activities. So for this I ran the Procyon benchmark which, as a reminder, measures various tasks using Microsoft’s Excel, Work, Powerpoint and Outlook. Interestingly the overall score was 7746 which is a drop of 2.5% from the Balanced Mode score of 7947. I thought this might be attributable to a benchmark margin of error, however after repeating the benchmarks several times it definitely gives a slightly lower score in Performance Mode.

The CPU improvements were also seen on Ubuntu. Running the PerformanceTest benchmark showed an increased CPU Mark from 30351 to 32474, and after running Geekbench, the Multi-Core improved from 12340 to 13002.

Adding an eGPU

Given the NUC13OXv9 has two USB4 ports, I was interested to see whether an eGPU would work given the confusion over the speed of the ports, and also what the performance might be like. I connected my self-made eGPU which consists of a NVIDIA GeForce RTX 3050 housed in a ADT-Link K43SG and connects via a USB4 M.2 NVMe Enclosure and a USB4 cable.

The first game I tried was Shadow of the Tomb Raider selecting “Quality” for DLSS and the “High” preset for Graphics. Running the built-in benchmark resulted in an average FPS of 101 and the game was 100% GPU bound. Not surprising since I was only using an RTX 3050. Next I ran the Horizon Zero Dawn benchmark using the Preset “Favor Quality” and got an average FPS of 70. Finally I ran the benchmark in Call of Duty: Modern Warfare II. Using just the default settings I got an average of 55 FPS with the GPU causing 99% of the bottlenecks.

Obviously if you can afford the NUC13OXv9 you would probably be pairing it up with a better graphics card, which, based on the results seen here, would make for very good gaming performance.

How does it stack up?

The claim was “the Onyx delivers unmatched computing performance for modern business” so I feel it is justified to compare the NUC13OXv9 with the other Intel Core i9 mini PC on the market: the GEEKOM Mini IT13. Tabulating the results I got from running the same set of benchmarks under similar conditions on each device with default settings, shows the following:

Some may (and will) argue it is an unfair comparison as the Power Limits are different. The counter argument is: this is how the devices are shipped by the manufacturer and given there is no advice provided with either device on changing the Power Limits, it must be assumed that this is how the manufacturer’s see the performance of the device for the majority of usage cases.

Certainly the CPU and GPU benchmarks seem to favour the NUC13OXv9 whereas the office related tasks as measured by Procyon are better on the Mini IT13. However I believe there is a reason for this.

Simply NUC have basically shot themselves in the foot with their choice of NVMe drive. If just these office tests were repeated having switched the drives around, it would be interesting to see the result. Unfortunately it is not that easy to do so without spending time with reinstallations and driver updates on both machines. So whether it stacks up or not, I’ll let you decide in the comments.

Power Usage

Power consumption, rounded to the nearest Watt, was measured as follows:

• Powered off (shutdown) – 1 Watts
• UEFI (BIOS) – 20 Watts
• GRUB menu – 39 Watts
• Idle – 13 Watts (Windows) and 9 Watts (Ubuntu)
• CPU stressed* – 82 Watts (Windows “cinebench”) and 77 Watts (Ubuntu “stress”)
• Video playback** – 30 Watts (Windows Edge 4K60fps) and 30 Watts (Ubuntu Firefox*** 4K60fps)

• Maximum power reading observed.
  ** The power figures fluctuate so the value is the average of the median high and median low power readings.
  *** Chrome browser power usage on Ubuntu is much higher at 44 Watts.

Verdict

The Simply NUC Onyx NUC13OXv9 has a lot to offer, especially for the business community. Importantly it supports vPro and comes with a three year warranty which are two criteria that typically rule out most “Chinese” mini PC alternatives. Using DDR5 makes it future proof memory-wise, although that did impact the relevant benchmarks but this could just be related to how those scores are calculated.

The faster than Gigabit Wi-Fi was impressive and shows that the design and layout of the aerials is spot on. But above all, it was the effectiveness of the cooling coupled with the quietness of the fan that makes this mini PC so good.

But, there is room for some improvements. The terrible M.2 NVMe drive should be replaced with one whose write speeds at least match its read speeds and that both are faster than PCIe Gen 3.

The UEFI (BIOS) is very different both visually and with the short-cut keys for anyone used to an Intel NUC. This may be a compatibility issue for any company that has written IT operations manuals or user documentation, and who would be reluctant to either rewrite it or modify it to cater for multiple devices. Not only that, but the existing support documentation on Simply NUC’s website aligns with Intel NUCs and needs to be updated to support the NUC13OXv9. Finally, having the speed of the USB4 ports labelled as 20 Gb/s when they function at 40 Gb/s in Windows is somewhat confusing.

It is also unclear who is responsible for support when connecting USB4 devices to them on Ubuntu given there are a couple of issues. Simply NUC has clarified that the labelling is due to certification purposes and that whilst the processor does support Thunderbolt capabilities, the ports do not have Thunderbolt certification. However the labelling is the “USB 20Gbps Port Logo” which is one of the USB Performance Logo trademarks owned by the USB Implementers Forum (USB-IF). The USB Performance Logo Usage Guidelines state that using the 40Gbps logo must have been “submitted to and passed the USB 40Gbps Test Procedure, and has been posted on the USB-IF Integrators List”. So one can only assume that the USB4 ports have just a “20Gbps” certification at this point.

The high price for the NUC13OXv9 will likely be the barrier that prevents potential consumer sales. People may prefer to wait to see what ASUS offers, or hold out for a Core i9-14900H or even a “Meteor Lake” processor which Intel is rumoured to be close to launching.

I’d like to thank Simply NUC for providing the review unit. It’s available for purchase from Simply NUC’s website, and prices for a Simply NUC Onyx system with a Core i9-13900H processor start at $999 for a model with 8GB of RAM and 256GB of storage. The configuration featured in this review (91R-WH28-001) costs $1239.00.