Beelink U59 Pro review – A Jasper Lake mini PC with faster GPU performance

Beelink U59 Pro review

Beelink’s newly launched U59 Pro addresses the weak graphical performance offered by the original Beelink U59, which we also noticed in the Beelink MINI S, and was a direct result of the iGPU limitations. By simply upgrading the CPU from a ‘desktop’ to ‘mobile’ Celeron processor the iGPU received a fifty percent boost in execution units. Beelink kindly sent one for review and I’ve looked at performance running both Windows and Ubuntu.

The Beelink U59 Pro physically consists of a 124 x 113 x 42mm (4.88 x 4.45 x 1.65 inches) square plastic case. As before it is an actively cooled mini PC but now uses Intel’s 10 nm Jasper Lake N5105 processor which is the same quad-core 4-thread 2.00 GHz Celeron processor boosting to 2.90 GHz but with improved Intel’s UHD Graphics as the ‘Graphics Burst Frequency’ increases from 750 MHz to 800 MHz and the number of ‘Execution Units’ goes from 16 to 24.

Whilst the front panel remains unchanged with an illuminated power button, dual USB 3.1 ports, a Type-C USB 3.0 port with Alternate Mode, a 3.5mm headphone jack, and a reset pin-hole ‘CLR CMOS’, the rear panel gets a minor upgrade and now includes dual gigabit Ethernet ports as well as dual USB 3.0 ports, dual HDMI (assumed to be 2.0) ports and the power jack.

The review model included a 512GB M.2 2280 SATA SSD drive which is no longer covered by a metal heatsink but instead a rather thick thermal pad. It also now has Windows 11 Pro pre-installed. Like before there are two sticks of 8GB DDR4 2666 MHz memory:

Additionally, there is the same replaceable M.2 2230 WiFi 5 (or 802.11ac) Intel Wireless-AC 3165 card located under the M.2 2280:

and the ability to add an additional 2.5” SATA drive to the lid which is connected to the motherboard via a short ZIF cable:

The specifications state:

and the Beelink webpage lists all of the USB ports as 3.0 so I tested them together with the Type-C USB port using a Samsung 980 PRO PCle 4.0 NVMe M.2 SSD housed in a ‘USB to M.2 NVMe adapter’ (ORICO M2PAC3-G20 M.2 NVMe SSD Enclosure) which showed that front two USB ports were USB 3.1 (USB 3.2 Gen 2×1 i.e. 10 Gbit/s):

and only the rear and Type-C ports were 3.0 (USB 3.2 Gen 1×1 i.e. 5 Gbit/s):

Additionally, the Type-C port also supports video output through ‘Alternate Mode’:

which together with the dual HDMI ports enables support for triple 4K displays.

Box contents

In the box, you get a power adapter and cord, both a short and a longer HDMI cable, a VESA mounting bracket together with a small packet of miscellaneous screws. Also included is a multilingual user manual:

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. I now review using Windows 11 version 21H2 and Ubuntu 22.04.1 LTS. I 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. On Ubuntu, I also compile the v5.15 Linux kernel using the default config as a test of performance using a real-world scenario.

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

Installation Issues

When booting Ubuntu 22.04.1, there are various BIOS errors being reported in the ‘dmesg’ although the significance of which has not been determined:

The Beelink U59 Pro came installed with a licensed copy of Windows 11 Pro version 21H2 which after applying updates was build 22000.832. A quick look at the hardware information shows it is aligned to the specification:

Similar to when I reviewed the U59 the iGPU showed limited details in HWiNFO and was unknown to GPU-Z:

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

I then set the power mode to ‘High performance’ and ran my standard set of benchmarking tools to look at performance under Windows:

I also tested Cinebench R23:

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

Ubuntu 22.04 Performance

After shrinking the Windows partition in half and creating a new partition I installed Ubuntu as dual boot using an Ubuntu 22.04 ISO as the first point release had been delayed and it was necessary to perform a manual upgrade to 22.04.1. After installation and updates, a brief check showed working audio, Wi-Fi, Bluetooth, and Ethernet.

The key hardware information under Ubuntu 22.04.1 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:

and the latter can be directly compared to when run in Windows using the OpenGL render:

I also ran PassMark PerformanceTest Linux:

which can be directly compared to the results from when it was run on Windows:

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

Video playback in YouTube & Kodi

For real-world testing, I played some YouTube videos in Edge and Chrome on Windows and in Firefox and Chrome on Ubuntu. On Edge, the initial codec for a video is ‘av01’ however as it then typically struggles to play the video it switches to ‘vp09’ whereas Chrome always used ‘vp09’.

The improved iGPU seems to favor Ubuntu where there was an observable overall improvement, unlike Windows which remained very similar to before.

I also played variously encoded videos in Kodi all of which played up to 8K @ 30 FPS without issue and used hardware for decoding:

However, like before, whilst hardware decoding was used when trying to play 8K @ 60 FPS videos it resulted in frame skipping and juddery playback on the U59 Pro:

Taking a detailed look at comparing the CPU, GPU and memory performance of the U59 Pro and U59 confirms that both the CPU and memory performance is similar and that the iGPU performance shows a respectable performance improvement in graphical benchmarks:

However, this graphical improvement only appears to be reflected in browser performance on Ubuntu which resulted in smoother playback:

Thermals

The Beelink U59 Pro, as mentioned, uses active cooling. Running a stress test on Ubuntu saw the CPU temperature climb to an average of around 73°C for the duration of the test with a peak of 76°C:

During the stress test, the maximum temperature I recorded on the top of the device was around 42.7°C in an ambient room temperature of 15.6°C and the fan was hardly audible reaching 34 dBA on my sound meter next to the device during the test. If the CPU frequency is monitored during the stress test it can be seen that it flat-lined at 2800 MHz for the duration of the test:

Networking (Ethernet & WiFi) throughput

Network connectivity throughput was measured on Ubuntu using ‘iperf3’:

Power consumption was measured as follows:

  • Powered off (shutdown) – 0.3 Watts
  • BIOS – 14.0 Watts
  • GRUB boot menu – 13.2 Watts
  • Idle – 11.9 Watts (Windows) and 5.4 Watts (Ubuntu)
  • CPU stressed* – 22.7 Watts (Windows ‘Cinebench’) and 18.3 Watts (Ubuntu ‘stress’)

*The power figures fluctuate due to the fan so the value is the average of the median high and median low power readings.

BIOS

Powering up the mini PC and hitting the F7 key results in a boot menu that includes access to the BIOS which is relatively unrestricted.

Final Observations

Whilst the Beelink U59 Pro has improved graphical performance compared to the original U59, casual Windows users may not necessarily benefit from this unless they are using applications that specifically make use of the additional execution units like gaming but then both the CPU and storage become bottlenecks. Having said this, however, the overall graphical improvement has addressed the main issue of the earlier U59. Another advantage of the U59 Pro over the U59 is the additional Ethernet port which may be required in commercial/business usage. Other advantages are that it runs relatively silently and is similarly priced (and can even be found cheaper) to the U59.

HighlightsLimitations
Better Linux iGPU performance than U59Only SATA storage
Dual Ethernet ports
No SD card slot
Additional SATA drive expandability
Not suitable for ‘AAA’ gaming

I’d like to thank Beelink for providing the Beelink U59 Pro for review. It retails at around $229 for the tested configuration of 16GB/500GB on Beelink store, and can also be purchased on Amazon.

Share this:

Support CNX Software! Donate via PayPal or cryptocurrencies, become a Patron on Patreon, or buy review samples

23 Replies to “Beelink U59 Pro review – A Jasper Lake mini PC with faster GPU performance”

    1. That’s odd that the N5105 mobile processor looks identical to the N5095 desktop processor, except for the former having a more powerful GPU, and yet a lower TDP. I suppose the desktop processor may burst for a longer period of time?

      Reported power consumption numbers for the N5095 are also higher in most cases. I reproduce them here to ease the comparison:
      N5095 (15W TDP):

      • BIOS* – 10.8 Watts
      • GRUB boot menu – 10.5 Watts
      • Idle – 9.6 Watts (Windows) and 5.0 Watts (Ubuntu)
      • CPU stressed – 20.4 Watts (Windows ‘cinebench’) and 22.2 Watts (Ubuntu ‘stress’)

      N5105 (10W TDP):

      • BIOS – 14.0 Watts
      • GRUB boot menu – 13.2 Watts
      • Idle – 11.9 Watts (Windows) and 5.4 Watts (Ubuntu)
      • CPU stressed – 22.7 Watts (Windows ‘Cinebench’) and 18.3 Watts (Ubuntu ‘stress’)
      1. I mentioned in the opening paragraph that the processor had swapped from being a ‘desktop’ to ‘mobile’ one implying the lower TDP gimping that Intel defines, however the ‘Power Limits’ between the two devices are probably responsible for the increase power consumption as PL/2 goes from 25W to 30W in the lower TDP N5105 U59 Pro device.

        1. > lower TDP

          And as we learned the last years: TDP with Intel is a rather useless metric anyway since TDP is all about ‘base frequency’ while with benchmarks or any real-world workload we’re at ‘burst frequency’ that is subject to thermal/power limits.

          N5100 in my PN41 has a 6W TDP rating (at 1.1 GHz) compared to N5105’s 10W TDP rating (at almost twice the clockspeed: 2.0 GHz).

      2. > N5095: 20.4 Watts (Windows ‘cinebench’) and 22.2 Watts (Ubuntu ‘stress’)
        > N5105: 22.7 Watts (Windows ‘Cinebench’) and 18.3 Watts (Ubuntu ‘stress’)

        Makes no sense at all 🙂

        IMO using stress/stress-ng without additional monitoring is rather useless. Better use cpuminer to generate load since the generated scores provide insights at the same time (like: CPU cores throttled due to different UEFI/ACPI settings resulting in different thermal trip points).

        Cinebench R23 uses AVX on Intel and as such real clockspeeds (or different Turbo behaviour) are also important.

        1. The power consumption figures are only meant to be indicative rather than metrics to give people an idea of the power usage profile when the CPU is idle vs when busy. I was already thinking about dropping the BIOS/GRUB readings simply because they were relatively meaningless – I originally only included them as they were easy to measure!

          Another point that doesn’t get much coverage in reviews is that power consumption is affected by the fan, including the number of fans in a device and type/efficiency of fan(s) etc. However I don’t plan on starting this discussion as it is a level of detail beyond just a simple product review.

          1. > I was already thinking about dropping the BIOS/GRUB readings

            I always enjoyed them since they give a clue what happens if all the power management mechanisms aren’t (properly) at work.

            A PC idling at ~11W in UEFI vs. 5W in Linux just shows how important the software side of things is not only for performance but also for consumption.

            BTW: IMHO calling UEFI BIOS is rather misleading…

          2. Based on your interest then I’ll keep showing the BIOS/GRUB readings.

            Whilst the definition of ‘UEFI’ and ‘BIOS’ are clear, when the terms are used in mini PCS they are often confused and used somewhat interchangeably. For example, after power-on and pressing <DEL> to enter ‘Setup’ as instructed on the screen, AMI’s Aptio Setup screen is displayed and the first column is headed ‘BIOS Information’ with the first entry has ‘BIOS Vendor’ set to ‘American Megatrends’ but further down there is an entry for ‘Compliancy’ set to ‘UEFI 2.7; PI 1.6’. Even AMI’s website refers to it as ‘BIOS’ in places, e.g. ‘To find out how to enter your system’s BIOS setup’ (ref: www ami com/top-three-keys-used-to-enter-uefi-bios-setup) as well as ‘BIOS/UEFI’ and ‘UEFI BIOS’ in various places on their website. Often mini PC vendors talk of ‘BIOS’ updates when offering fixes or solutions to user issues so again there may be confusion when people do not realise they are referring to UEFI updates.

            If I called it ‘UEFI (BIOS)’ would that be more acceptable without being too confusing?

          3. > ‘UEFI (BIOS)’

            Would prefer that since while there are many consumers who never heard of UEFI (while this exploit-friendly OS nowadays runs on every modern PC in the background) it might raise awareness that this is something different than ‘good old’ BIOS handing over control to some OS after booting from four decades ago.

  1. Fun thing: on “chineese 4 port firewall” from Ali with same N5105 processor seller claim 64Gb RAM supported. I installed 32 and they works fine…

    1. Beelink’s U59 N5105 page (id=334) also says 64 GB is supported.

      Intel’s N5105 page says 16 GB maximum.

      Intel should fix its nuisance bad info. Maybe I should buy one of these just so I can get in on a class action lawsuit.

      1. > Intel’s N5105 page says 16 GB maximum.

        As with every other Jasper Lake SKU. With Gemini Lake (Refresh) it was exactly the same: Intel claiming 8GB max while in reality those users who bought slow DIMMs could use up to 32GB. Those who bought more expensive ones with low latency / high clocks had no luck and the modules didn’t work reliable or at all (even with upgraded UEFI).

        Most probably with Jasper Lake it’s the same: Intel playing safe to avoid consumers ‘getting in on a class action lawsuit’. 😉

    2. Care to provide output from ‘dmidecode –type 17’ to get a clue which timings have been negotiated?

      1. You point to very interesting info, thanks.

        I see no timing info in the dmidecode output:

        decode-dimms give more info:

        1. Ah, just realized that it’s ‘decode-dimms’ that does a better job (if supported by your system – you need most probably package ‘i2c-tools’ and should do a ‘modprobe eeprom’ before calling ‘decode-dimms’)

          See for example starting at ‘Decoding EEPROM’ here with a Jasper Lake thingy: http://ix.io/41aB

          1. I don’t think ‘decode-dimms’ shows XMP timings if you have then enabled in the UEFI (BIOS) and I’d be interested in learning what command would show them if anyone knows.

          2. At least CPU-Z on Windows should list profiles and timings. But no clue about other OS at all…

          3. i2c-tools installed, after “modprobe eeprom” nothing changed, only second module detected. What exactly we search for?

          4. Sorry, I overlooked decode-dimm output form your 1st answer (been on cellphone). I guess the information we’re searching for is simply not there 🙁

  2. Not to start a terminology flame war but in my opinion bios and uefi are used interchangeably. It hasn’t even been possible to buy a new computer with an actual bios in it since 2019. Everything new comes with a uefi but as a sys admin I pretty much only hear people refer to it as a bios. Kind of similar to people using the windows add/remove programs terminology even though it was renamed years ago. That is all 🙂

Leave a Reply

Your email address will not be published.

Advertisement
Advertisement