Beelink SEI Review – A Core i3-10110U Mini PC Tested with Windows and Ubuntu

Beelink has launched a new range of mini PCs called the SEi Series. Similar in size and appearance to an Intel ‘NUC’ they are available in various configurations. Beelink sent a Core i3-10110U SEi model for review which is the version that has now replaced their i3-1005G1 model which they had to discontinue due to the lack of processor availability. There is also an i5-8259U model in the series.

Beelink SEi Hardware Overview

The SEi physically consists of a 124 x 113 x 41mm (4.88 x 4.48 x 1.61 inches) rectangular plastic case. It is an actively cooled mini PC and uses Intel’s 14 nm++ Core i3-10110U Comet Lake processor which is a dual-core 4-thread 2.10 GHz processor boosting to 4.10 GHz with Intel’s UHD Graphics for 10th Gen Intel Processors.

The front panel has a power button, a headphone jack, a Type-C USB 3.0 port, and two USB 3.0 ports. The rear panel includes the power jack, two HDMI ports, two more USB 3.0 ports, and a gigabit Ethernet port. The left side panel includes a micro-SD card slot.

The review model included a 256GB M.2 2280 NVMe SSD drive and 8GB DDR4 2666MHz memory:

Additionally there is soldered-on WiFi 6 (or 802.11ax):

as well as the ability to add an additional 2.5” SATA drive.

The marketing specifications include:

Box contents

In the box, you get a power adapter and cord, two different length HDMI cables, a motherboard SATA connection kit, and a mounting bracket together with screws for attaching the device to behind a monitor. Also included are a user manual and a hard disk installation guide:

Review Methodology

When reviewing mini PCs I typically look at their performance under both Windows and Linux (Ubuntu) and compare them against some of the more recently released mini PCs. Starting in 2021 I am now reviewing using Windows 10 version 20H2 and Ubuntu 20.04.1 LTS and test with a selection of commonly used Windows benchmarks and/or equivalents for Linux together with Thomas Kaiser’s ‘sbc-bench’ which is a small set of different CPU performance tests focusing on server performance when run on Ubuntu. I also use ‘Phoronix Test Suite’ and benchmark with the same set of tests on both Windows and Ubuntu for comparison purposes. On Ubuntu, I also compile the v5.4 Linux kernel using the default config as a test of performance using a real-world scenario. Compared to previous reviews there are some further changes for 2021: 3DMark no longer supports Sky Diver so I’ve removed it from the benchmarks and I’ve also removed Novabench but added Passmark PerformanceTest Linux (which is currently in beta). I’m now using CrystalDiskMark 8.0.1 which has updated the default settings from RND 4KiB Q32T16 to RND 4KiB Q32T1.

Prior to benchmarking, I perform all necessary installations and updates to run the latest versions of both OSes. I also capture some basic details of the device for each OS.

Installation Issues

Initially, I upgraded Windows to 20H2 and ran my benchmarking tools. However, when reviewing all the results I found that some were lower than expected especially for Windows CPU-based tests. For example, the CPU Mark for Passmark in Windows was 3738.6 whereas it was 4342.16 in Ubuntu:

Subsequent investigation showed that this was caused by ‘Power Limit’ throttling as indicated by HWiNFO64:

and apparently caused by ‘CPU Power Limit’ being set too low:

The monitoring graph from 3DMark Fire Strike visually shows the CPU throttling:

Having already performed the Ubuntu benchmarking at this point I thought I’d first see the effect of changing the ‘Power Limit’ settings in Windows using Throttlestop. Surprisingly Throttlestop showed different values to those originally reported by HWiNFO64:

and they were also now confirmed when rerunning HWiNFO64:

So I checked the settings in Ubuntu with ‘rapl-info’ which also showed these higher values:

Rerunning the Ubuntu CPU-based tests gave essentially the same results as before as they were within the margin of test variance so it meant the Ubuntu benchmarks were unaffected and that I just had to rerun the Windows benchmarks.

Rerunning Windows CPU based tests now showed the expected performance. For example, CPU Mark for Passmark in Windows was now 4256.7:

and the monitoring graph from 3DMark Fire Strike showed minimal CPU throttling:

Why the values for ‘Power Limit’ changed has not been identified. Now on booting most of the time the ‘Power Limit’ shows as 25W however occasionally a reboot can result in the lower value of 15W regardless of OS:

It is not clear why the 25W value is being set as although it is documented by Intel as the ‘Configurable TDP-up’ for the CPU, Windows shows the ‘Current Configurable TDP Level’ as set to ‘Nominal’ which is defined in the BIOS as:

and the BIOS also shows the current ‘Power Limit’ at 15W:

Obviously, the value for ‘Power Limit’ affects the performance and it is not ideal having to check and reboot if affected by the randomness of this occurrence.

Ubuntu was installed using the latest ISO from Canonical and after booting the ‘dmesg’ showed a couple of PCIe Bus errors:

At one point during testing the Ethernet connection stopped working:

However, after a reboot, no further Ethernet issues were encountered. When the ‘dmesg’ was checked after testing network performance the log was full of the same PCIe Bus errors which appear to be linked to the Ethernet controller:

This spamming of bus errors is well documented on the internet although a specific solution for this particular instance has not been found.

Beelink SEI – Windows 10 Performance

Initially, the SEi comes installed with a licensed copy of Windows 10 Pro version 2004 build 19041.685. After upgrading to version 20H2 build 19042.746 a quick look at the hardware information shows it is aligned to the specification:

A brief check showed working audio, micro-SD, Wi-Fi, Bluetooth, and Ethernet.

I then set the power mode to ‘High performance’ and ran my (2021) standard set of benchmarking tools to look at performance under Windows. However, due to the ‘Power Limit’ issue, I had to rerun the benchmarking tools having received an inevitable Windows update which took the build to 19042.789:

and the new ‘Power Limit’ values were:

The rerun Windows benchmarking tools results were:

For my specific set of Phoronix Test Suite tests the rerun results were:

All these results can then be compared with other recent mini PCs:

The Core i3-1011U integrated graphics is not particularly powerful, and Beelink SEi not really suitable for ‘AAA’ gaming.

Ubuntu Performance on Beelink SEi Comet Lake mini PC

After shrinking the Windows partition in half and creating a new partition I installed Ubuntu using an Ubuntu 20.04.1 ISO as dual boot. After installation and updates, a brief check showed working micro-SD, Wi-Fi, Bluetooth, and audio including HDMI and headphones.

The key hardware information under Ubuntu 20.04.1 is as follows:

I then set the CPU Scaling Governor to ‘performance’ and ran my Linux benchmarks for which the majority of the results are text-based but the graphical ones included:

As mentioned I also ran Passmark PerformanceTest Linux:

For the same set of Phoronix Test Suite tests the results were:

The complete results together with a comparison against other recent mini PCs are:

again showing that the processor is not really suitable for ‘AAA’ gaming in Ubuntu.

Browsers & Kodi

For real-world testing, I played some videos in Edge, Chrome, and Kodi on Windows and in Firefox, Chrome, and Kodi on Ubuntu. The following tables summarise the tests and results for each:

In browser playback because of higher CPU usage the fan ramps up and becomes audible. The fan RPM is higher in Ubuntu resulting in more noise and when the fan initially ramps it also results in dropped frames. As a result browser playback in Windows is considerably better than in Ubuntu.

I also tested playing an 8K video in Kodi on both OS. In Windows 8K 30 FPS played without issue and 8K 60 FPS played fine with just the occasional skipped frame:

However, in Ubuntu whilst both 8K 30 FPS and 60 FPS played they were juddery with constant frame skipping despite both using hardware decoding:

Windows 10 vs Ubuntu 20.04

Whilst a detailed comparison between the two operating systems is beyond the scope of this review, it is worth noting some of the key findings I observed. First looking at the performance tools common between the two systems. Overall Ubuntu performs slightly better in the benchmarks than Windows and this can be visually shown by comparing the same Phoronix Test Suite benchmarks in each OS:

However, as mentioned above Windows played videos much better than Ubuntu.

Thermals

The Beelink SEi mini PC uses active cooling with dual copper heat pipes and a fan to assist with heat dissipation. During benchmarking the maximum temperature I recorded on the top of the device was around 34°C in an ambient room temperature of 24.5°C. Although the device didn’t become hot to touch, the fan does become audible under load.

When the fan was running ‘quietly’ it registered around 37 dBA on my sound level meter next to the device. When playing a video in browsers on Ubuntu the fan quickly ramps up to around 43 dBA and then can get louder reaching up to 50 dBA as it cools the hot CPU:

Running a stress test on Ubuntu saw the CPU temperature rise quickly to 89/87°C and the fan ran continuously at 51 dBA. The test was interrupted after approximately eight minutes given everything appeared hot but stable and the CPU temperature dropped almost immediately back down to 45°C:

The top of the device reached a maximum of 38°C during the stress test and the ambient room temperature was 26.9°C.

Networking

Network connectivity throughput was measured on Ubuntu using ‘iperf’:

The WiFi results are very good especially for the 5 GHz band with speeds nearly the same as gigabit Ethernet.

Power

Power consumption was measured as follows:

• Initially plugged in – 1.0 Watts
• Powered off (shutdown) – 0.7 Watts (Windows) and 0.7 Watts (Ubuntu)
• BIOS*  – 9.3 Watts
• GRUB boot menu – 7.7 Watts
• Idle – 6.4 Watts (Windows) and 2.6 Watts (Ubuntu)
• CPU stressed – 36.4W (Windows ‘cinebench’) and 29.7W (Ubuntu ‘stress’)
• 4K 30 FPS Video playback** – 14.6 Watts (Windows Edge) and 34.2 Watts (Ubuntu Chrome)

*BIOS (see below)
**The power figures fluctuate notably due to the fan so the value is the average of the median high and median low power readings.

BIOS

The BIOS is one of the most unrestricted I’ve seen and the following video which provides a brief overview is well worth watching to see the settings many BIOS hide:

One setting of interest is the ‘Boot performance mode’ which is set at ‘Max Non-Turbo Performance’ by default:

It determines the CPU performance whilst in the BIOS and gets overridden once the OS boots. It is not worth changing and that probaby goes for all of the settings unless you specifically know what you are doing.

Final Observations

At first glance, the i3-10110U based Beelink SEi  mini PC looks to be similar to the ‘NUC 10 Performance’ but without the Thunderbolt port and having lower USB specs (5Gbps rather than 10 Gbps). But by including a Windows Pro license and offering fully loaded configurations with RAM and storage the SEi is trying to be a price-conscious alternative.

Compared to the previous generation of Gemini Lake mini PCs this is a more powerful device with both improved CPU and GPU performance. This does come at a slight cost as the fan is quite noisy when the processor is under load. It is unfortunate that ‘Power Limit’ throttling can occur seemingly randomly and it is not ideal having to check and reboot if affected. However, the inclusion of WiFi 6 and configurable storage options add to the improvements presented by the SEi.

Highlights	Limitations
WiFi 6	Audible fans
	Possible ‘Power Limit’ throttling

I’d like to thank Beelink for providing the SEi for review. It currently retails at around $489 on Amazon for a 16GB RAM and 512GB M.2 drive configuration.

Support CNX Software! Donate via cryptocurrencies, become a Patron on Patreon, or purchase goods on Amazon or Aliexpress