We’ve already checked out the specifications and hardware of the Beelink EQi Wildcat Lake mini PC in the first part of the review, with an unboxing and a teardown, before booting the mini PC into the pre-installed Windows 11 Pro for a first quick look.
We’ve now had time to test the Intel Core 3 304 mini PC in more detail on Windows 11 Pro, so we’ll report our experience. The review includes feature testing, benchmark results, 4K and 8K YouTube video playback, 10GbE, 2.5GbE, and WiFi network performance evaluation, cooling performance, and measurement of fan noise and power consumption.
Software Overview and Feature Testing
After fully updating Windows 11 Pro and drivers to the latest version (excluding previews), I went to System > About window again. We still have a 1.50 GHz (base frequency) Intel Core 3 304 processor with 16GB 6400MT/s memory, and 834GB of storage from the UFS chip and the two SSDs I installed. The computer runs the latest Windows 11 Pro 25H2 build 26200.8655.
HWiNFO 64 provides more details about the 15W Intel Core 3 304 1P+4E cores Wildcat Lake-U CPU, the AZW EQi motherboard with 16GB LPDDR5 memory, and the integrated graphics with a single Xe core running at 1000 MHz (max is 2300 MHz). We can also notice that Secure Boot is disabled by default in the BIOS.
I also ran GPU-Z as usual, but the Intel “FD81” GPU in the Core 3 304 is new and not properly recognized by the utility just yet.
Let’s now check the PL1, PL2, and PL4 power limits in HWiNFO 64: 15W (PBP), 35W (MTP), and 80W. The TDP of all Wildcat Lake processors is 15W, so that information looks right.
Beelink made two motherboard variants for the EQi Wildcat Lake mini PC: one with 16GB LPDDR5 soldered on the board, and another with a DDR5 SO-DIMM memory socket for the 24GB and 32GB RAM models. I got a sample of the former for review, so my motherboard comes with four 4GB LPDDR5 SDRAM using Micron chips for a total of 16GB.
Task Manager confirms that and the speed (6400 MT/s). We can also see that 342 MB of RAM is reserved for the hardware. The GPU appears to use 128 MB of RAM, so the rest might be the Intel AI Boost NPU.
Time to check wired and wireless connectivity on the Beelink EQi Wildcat Lake mini PC. It comes with a RealTek RTL8127 10GbE controller,
a MediaTek MT7922 Wi-Fi 6E (and Bluetooth 5.x) wireless module connected at 2999 Mbps,
and a second Ethernet port relying on an Intel I226-V 2.5GbE controller.
I also checked the Bluetooth version in Device Manager->MediaTek Bluetooth Adapter Properties->Advanced tab, where the firmware version is LMP13.xxx, which looks up to Bluetooth 5.4.
I also tested Bluetooth file transfers from and to an Android smartphone, and no issues here.
I installed two PCIe Gen3 x4 NVMe SSDs into the M.2 sockets of the mini PC. A 128GB model in the PCIe Gen4 x1 socket, and a 256GB model in the PCIe Gen4 x2 socket, meaning we should expect up to PCIe Gen3 x1 (~880 MB/s) and PCIe Gen3 x2 (~1600 MB/s) performance, and that’s exactly what we got here, at least for read speeds.
In theory, using a PCIe Gen4 x4 NVMe SSD over PCIe Gen4 x2 should deliver up to 3200 MB/s.
We tested the two Thunderbolt 4 ports and two USB 3.2 Gen 2 ports using an ORICO M234C3-U4 M.2 NVMe SSD enclosure and the USB 2.0 Type-A ports with a Seagate USB 3.0 HDD. HWiNFO 64 and CrystalDiskMark were used to list the reported USB version and speed, and test the actual performance.
For reference, here are the results for the left USB 3.2 Type-A port on the front panel…
… the second (right) USB4/Thunderbolt 4 port on the rear panel…
… and the first USB 2.0 port on the rear panel.

Here’s a summary for all six USB ports from left to right:
- Front panel
- USB-A – USB 3.2 – USB 3.2 Gen2 (SuperSpeedPlus 10 Gbps) – Read speed: 1060 MB/s; write speed: 1,043 MB/s
- USB-C – USB 3.2 – USB 3.2 Gen2 (SuperSpeedPlus 10 Gbps) – Read speed: 1058 MB/s; write speed: 1,040 MB/s
- Rear panel
- USB-C #1
- ORICO enclosure – Drive not detected, green LED on enclosure
- Beelink Expand M
- First test – Drive not detected, green LED on enclosure after pressing the M “wakeup” button
- Second test after reboot – USB 3.0 – USB 3.1 Gen1 (SuperSpeed 5 Gbps) – Read speed: 428 MB/s; write speed: 413 MB/s
- USB-C #2
- ORICO enclore – Thunderbolt @ 8GT/s – Read speed: 3135 MB/s; write speed: 2478 MB/s
- Beelink Expand M – USB 3.0 – USB 3.1 Gen1 (SuperSpeed 5 Gbps) – Read speed: 423 MB/s; write speed: 413 MB/s
- USB-A #1 – USB 3.0 – USB 2.0 high-speed – Read speed: 44 MB/s; write speed: 43 MB/s
- USB-A #2 – USB 3.0 – USB 2.0 high-speed – Read speed: 44 MB/s; write speed: 44 MB/s
- USB-C #1
The USB 3.2 and USB 2.0 ports work as advertised. However, the Thunderbolt 4 ports were not reliable for me. I managed to connect and test a USB 3.0 NVMe enclosure on both after a few tries and reboots. However, the ORICO Thunderbolt 3 enclosure was even more of a struggle, and I could only make it work on the second USB-C port. The good news is that once it connects, the performance is fine.
Despite the previous issues, I felt adventurous and decided to connect the Khadas Mind Graphics 2 eGPU dock to the Thunderbolt 4. Neither worked with the same error message: “Display connection might be limited”.
If I click on the message, it brings me to more information about the USB4 host router 1.
I went to Device Manager to confirm the NVIDIA GeForce RTX 5060 Ti graphics card was not detected, and indeed, there was nothing there. Obviously, connecting an HDMI cable from the graphics card didn’t result in any video output. So I moved the HDMI cable back to the HDMI output on the mini PC, and played a YouTube video from there to check whether the built-in speakers of the Mind Graphics 2 would work, and they did. The behavior was the same on both Thunderbolt 4 ports on the mini PC. Those two 40 Gbps ports are a big disappointment, especially since I felt they were a key selling point of the mini PC. Hopefully, it’s just a driver/interoperability issue, and it will be fixed soon.
The Beelink EQi Wildcat Lake Core 3 304 mini PC comes with three video outputs: one HDMI 2.0 port, and two USB4 ports with DisplayPort Alt mode. I’m not optimistic, but let’s see if we can get three displays to work.
And it worked (to my astonishment)! I connected the left USB4 port to the 14-inch Crowview portable laptop monitor (1920×1080), the right USB4 port to Khadas Mind xPlay (2880×1920), and the HDMI port to a 32-inch KTC A32Q8 4K smart monitor.
However, I had to somewhat cheat to have all three displays work. I initially connected the Crowview over HDMI, and the two over displays over USB4, the the KTC monitor would not get any signal over USB-C. I eventually switch the cable between the Crowview and KTC displays and it worked.
Beelink EQi Wildcat Lake Core 3 304 benchmarks on Windows 11 Pro
I’ve set the power mode to “Best performance” before starting any benchmarks. Note that the ambient temperature during testing was 28–30°C, and results may vary compared to tests at lower temperatures.
Let’s start with PCMark 10.
The Wildcat Lake mini PC achieved 5294 points in the general-purpose PCMark 10 benchmark.
In the 3DMark Fire Strike 3D graphics benchmark, it got 1940 points.
The Intel Core 3 304 mini PC scored 2,806 points in PassMark PerformanceTest 11.1. The Disk Mark score of 9983.2 points is quite lower than most other mini PCs we’ve reviewed, simply because they come with an NVMe SSD, instead of a UFS 3.1 flash, which offers a middle ground between eMMC and NVMe storage. Let’s run CrystalDiskMark to double the UFS chip performance: 2,013.58 MB/s sequential read speed and 831.01 MB/s sequential write speed. It’s much faster than a typical eMMC flash, but about half of the read performance of a typical NVMe SSD. However, random I/Os results (RND4K Q32T1) are pretty good and slightly faster than with the NVMe SSD found in the GEEKOM A7 2026 Edition mini PC.
We’ll then use Cinebench R23 to test single-core and multi-core performance of the mini PC and the new penta-core Intel Core 3 304 SoC with one performance core and four efficiency cores.
The Wildcat Lake mini PC achieved 1,732 points in the single-core test and 4,863 points in the multi-core benchmark, with an MP ratio of 2.81x, highlighting the performance difference between P and E cores.
Let’s perform additional GPU testing with the Unigine Heaven Benchmark 4.0, where the Beelink EQi computer could render the scene at an average of 19.6 FPS and score 495 points at the standard 1920×1080 resolution.
I then played some YouTube 4K and 8K videos in Firefox at 30 and 60 FPS.
The system handles videos from 4K 30 FPS up to 8K 60 FPS very well. The videos at 60 FPS would freeze from time to time because “Buffer Health” would reach zero. But it’s more of an issue with my Internet connection and YouTube servers than the video decoding capabilities of the mini PC. Since all four tests relied on the AV1 video codec, I tried to find an 8K VP9 video. I could only select 3840×2160@60 with the VP9 video, and it handled it well without any frames dropped.
I switched to Google Chrome, and the results were identical. I also connected a pair of speakers to the 3.5mm audio jack to confirm sound was working properly, which it did.
Contrary to earlier Intel Alder Lake-N and Twin Lake processors, the new Intel Wildcat Lake family supports AI-accelerator workloads. For the Intel Core 3 304, the built-in GPU delivers up to 9 TOPS, and the Intel AI Boost NPU up to 15 TOPS. Let’s check that with Geekbench AI using OpenVino.
See full results for the GPU on the Geekbench website.
See full results for the Intel AI Boot NPU. Results vary, but for the quantized score, the NPU provides a clear performance benefit over the GPU, with over twice the performance.
Beelink EQi Wildcat Lake Windows 11 benchmarks comparison against other mini PCs
Now that we have gathered benchmark results, it’s time to compare the Beelink EQi Wildcat Lake Core 3 304 mini PC against the previous generation Intel Alder Lake-N mini PCs, such as the GEEKOM Mini Air12 (Intel N100) and Weibu N10 (Core i3-N305) mini PCs, as well as a mid-range ($500+) system: GEEKOM A5 Pro 2026 Edition.
Here are the high-level specs of the four systems.
| Beelink EQi Wildcat Lake | GEEKOM Mini Air12 | Weibu N10 | GEEKOM A5 2026 Edition | |
|---|---|---|---|---|
| SoC | Intel Core 3 304 | Intel Processor N100 | Intel Core i3-N305 | AMD Ryzen 5 7530U |
| CPU | 5-core (1P+4E) Widlcat Lake processor up to 4.3 GHz | 4-core Alder Lake-N processor up to 3.4 GHz | 8-core Alder Lake-N processor up to 3.80 GHz | 6-core/12-thread up to 4.5GHz |
| GPU | 1-core Intel Xe3 Graphics @ 2.3 GHz | 24 EU Intel UHD Graphics up to 750 MHz | 32EU Intel HD Graphics @ 1.25 GHz | 7-core AMD Radeon Vega Graphics @ 2.0 GHz |
| Memory | 16GB LPDDR5-6400 | 16GB DDR5-4800 | 8GB DDR4-3200 | 16GB DDR4-4800 |
| Storage | 512GB UFS 3.1 | 512GB M.2 NVMe SSD | 512GB NVMe SSD | 1TB M.2 SSD |
| Default OS | Windows 11 Pro | Windows 11 Pro | Windows 11 Pro | Windows 11 Pro |
| Price (July 12, 2026) | $509 (pre-order), $609 MSRP | $475 on GEEKOM website (with default coupon) | N/A (OEM device) | $499 on Amazon |
The shocker is the current price of the Intel N100 system, though we’ll find some systems for just over $300 with similar specifications.
Time for the benchmark results.
| Beelink EQi Wildcat Lake | GEEKOM Mini Air12 | Weibu N10 | GEEKOM A5 2026 Edition | |
|---|---|---|---|---|
| PCMark 10 | 5,294 | 3,150 | N/A | 5,586 |
| - Essentials | 7,998 | 7,467 | N/A | 10,099 |
| - Productivity | 10,171 | 4,598 | N/A | 9,128 |
| - Digital content creation | 4,951 | 2,471 | N/A | 5,134 |
| 3DMark (Fire Strike) | 1,940 | 1,188 | 1,534 | 3,328 |
| PerformanceTest 11.0 | 2,806 | 1,580 | 2,358 | 3,922 |
| - CPU Mark | 11,371.5 | 6,180 | 9,734 | 16,498 |
| - 2D Graphics Mark | 598 | 248 | 253 | 698 |
| - 3D Graphics Mark | 1,954.8 | 888 | 1,172 | 2,639 |
| - Memory Mark | 2,768.8 | 2,492 | 2,379 | 2,451 |
| - Disk Mark | 9,983.2 | 20,984 | 12,806 | 29,247 |
| Cinebench R23 | ||||
| - Single Core | 1,732 | 918 | 1,030 | 1,344 |
| - Multi Core | 4,863 | 2,927 | 4,505 | 6,157 |
| Unigine Heaven 4.0 | 19.6 FPS | 12.8 FPS | 16.5 FPS | 32.4 FPS |
As one might expect, the Wildcat Lake mini PC is better than Alder Lake-N mini PCs in every way, except for storage performance, which is lower due to the use of a UFS 3.1 device by default (note: you can always install an NVMe SSD for better storage performance). Graphics performance is significantly faster, and single-core performance is about 70% higher based on Cinebench R23. Multi-core is better too, but not quite as impressive due to the 1P+4E design.
Compared to the similarly priced AMD Ryzen 5 mini PC, the memory bandwidth is higher, and single-core performance stands out again, with a result close to what we got with a high-end Intel Core i9-13900H mini PC. PCMark 10 is marginally lower, but the main downsides are 3D graphics and multi-core performance, both of which are significantly higher on the 6-core/12-thread AMD system. Two things not shown on benchmarks to the benefit of the Beelink mini PC are features such as Thunderbolt 4 and 10GbE, which are missing from the GEEKOM A5 2026 Edition.
Network performance evaluation (2.5GbE, 10GbE, and Wi-Fi 6E)
Let’s first test the 2.5GbE port of the Beelink EQi Wildcat Lake mini PC (192.168.1.32) using the iperf3 utility and the UP Xtreme i11 Edge mini PC (192.168.1.36) on the other side.
- Download
1234567891011121314devkit@UPX-i11:~$ iperf3 -t 60 -c 192.168.1.32 -i 10Connecting to host 192.168.1.32, port 5201[ 5] local 192.168.1.36 port 57494 connected to 192.168.1.32 port 5201[ ID] Interval Transfer Bitrate Retr Cwnd[ 5] 0.00-10.01 sec 2.74 GBytes 2.35 Gbits/sec 0 1.50 MBytes[ 5] 10.01-20.01 sec 2.74 GBytes 2.35 Gbits/sec 0 1.50 MBytes[ 5] 20.01-30.01 sec 2.74 GBytes 2.35 Gbits/sec 0 3.42 MBytes[ 5] 30.01-40.00 sec 2.74 GBytes 2.35 Gbits/sec 0 5.17 MBytes[ 5] 40.00-50.01 sec 2.74 GBytes 2.35 Gbits/sec 0 5.17 MBytes[ 5] 50.01-60.01 sec 2.74 GBytes 2.35 Gbits/sec 0 5.17 MBytes- - - - - - - - - - - - - - - - - - - - - - - - -[ ID] Interval Transfer Bitrate Retr[ 5] 0.00-60.01 sec 16.4 GBytes 2.35 Gbits/sec 0 sender[ 5] 0.00-60.01 sec 16.4 GBytes 2.35 Gbits/sec receiver - Upload
1234567891011121314151617devkit@UPX-i11:~$ iperf3 -t 60 -c 192.168.1.32 -i 10 -RConnecting to host 192.168.1.32, port 5201Reverse mode, remote host 192.168.1.32 is sending[ 5] local 192.168.1.36 port 35882 connected to 192.168.1.32 port 5201[ ID] Interval Transfer Bitrate[ 5] 0.00-10.00 sec 2.74 GBytes 2.35 Gbits/sec[ 5] 10.00-20.00 sec 2.74 GBytes 2.35 Gbits/sec[ 5] 20.00-30.01 sec 2.74 GBytes 2.35 Gbits/sec[ 5] 30.01-40.00 sec 2.74 GBytes 2.35 Gbits/sec[ 5] 40.00-50.01 sec 2.74 GBytes 2.35 Gbits/sec[ 5] 50.01-60.00 sec 2.74 GBytes 2.35 Gbits/sec- - - - - - - - - - - - - - - - - - - - - - - - -[ ID] Interval Transfer Bitrate[ 5] 0.00-60.01 sec 16.5 GBytes 2.36 Gbits/sec sender[ 5] 0.00-60.00 sec 16.4 GBytes 2.35 Gbits/sec receiveriperf Done. - Full-duplex (bidirectional)
12345678910111213141516171819202122232425devkit@UPX-i11:~$ iperf3 -t 60 -c 192.168.1.32 -i 10 --bidirConnecting to host 192.168.1.32, port 5201[ 5] local 192.168.1.36 port 40790 connected to 192.168.1.32 port 5201[ 7] local 192.168.1.36 port 40794 connected to 192.168.1.32 port 5201[ ID][Role] Interval Transfer Bitrate Retr Cwnd[ 5][TX-C] 0.00-10.00 sec 305 MBytes 256 Mbits/sec 0 1.05 MBytes[ 7][RX-C] 0.00-10.00 sec 2.73 GBytes 2.35 Gbits/sec[ 5][TX-C] 10.00-20.00 sec 253 MBytes 212 Mbits/sec 0 1.05 MBytes[ 7][RX-C] 10.00-20.00 sec 2.73 GBytes 2.35 Gbits/sec[ 5][TX-C] 20.00-30.01 sec 185 MBytes 155 Mbits/sec 0 1.05 MBytes[ 7][RX-C] 20.00-30.01 sec 2.74 GBytes 2.35 Gbits/sec[ 5][TX-C] 30.01-40.01 sec 174 MBytes 146 Mbits/sec 0 1.05 MBytes[ 7][RX-C] 30.01-40.01 sec 2.74 GBytes 2.35 Gbits/sec[ 5][TX-C] 40.01-50.01 sec 176 MBytes 147 Mbits/sec 0 1.05 MBytes[ 7][RX-C] 40.01-50.01 sec 2.74 GBytes 2.35 Gbits/sec[ 5][TX-C] 50.01-60.01 sec 176 MBytes 147 Mbits/sec 0 1.05 MBytes[ 7][RX-C] 50.01-60.01 sec 2.74 GBytes 2.35 Gbits/sec- - - - - - - - - - - - - - - - - - - - - - - - -[ ID][Role] Interval Transfer Bitrate Retr[ 5][TX-C] 0.00-60.01 sec 1.24 GBytes 177 Mbits/sec 0 sender[ 5][TX-C] 0.00-60.01 sec 1.24 GBytes 177 Mbits/sec receiver[ 7][RX-C] 0.00-60.01 sec 16.4 GBytes 2.35 Gbits/sec sender[ 7][RX-C] 0.00-60.01 sec 16.4 GBytes 2.35 Gbits/sec receiveriperf Done.
Perfect results for 2.5GbE networking.
Let’s now test the 10GbE port using the iKOOLCORE R2 max mini PC running OpenWrt fork QWRT.
- Download
12345678910111213141516root@QWRT:~# iperf3 -t 60 -c 192.168.4.100 -i 10Connecting to host 192.168.4.100, port 5201[ 5] local 192.168.4.1 port 52904 connected to 192.168.4.100 port 5201[ ID] Interval Transfer Bitrate Retr Cwnd[ 5] 0.00-10.01 sec 11.0 GBytes 9.41 Gbits/sec 0 3.35 MBytes[ 5] 10.01-20.01 sec 11.0 GBytes 9.42 Gbits/sec 0 3.71 MBytes[ 5] 20.01-30.01 sec 10.9 GBytes 9.39 Gbits/sec 0 6.18 MBytes[ 5] 30.01-40.01 sec 11.0 GBytes 9.41 Gbits/sec 0 6.18 MBytes[ 5] 40.01-50.01 sec 11.0 GBytes 9.41 Gbits/sec 0 6.18 MBytes[ 5] 50.01-60.01 sec 11.0 GBytes 9.42 Gbits/sec 0 6.18 MBytes- - - - - - - - - - - - - - - - - - - - - - - - -[ ID] Interval Transfer Bitrate Retr[ 5] 0.00-60.01 sec 65.7 GBytes 9.41 Gbits/sec 0 sender[ 5] 0.00-60.01 sec 65.7 GBytes 9.41 Gbits/sec receiveriperf Done. - Upload
123456789101112131415root@QWRT:~# iperf3 -t 60 -c 192.168.4.100 -i 10 -RConnecting to host 192.168.4.100, port 5201Reverse mode, remote host 192.168.4.100 is sending[ 5] local 192.168.4.1 port 41282 connected to 192.168.4.100 port 5201[ ID] Interval Transfer Bitrate[ 5] 0.00-10.00 sec 10.9 GBytes 9.39 Gbits/sec[ 5] 10.00-20.00 sec 11.0 GBytes 9.41 Gbits/sec[ 5] 20.00-30.00 sec 11.0 GBytes 9.41 Gbits/sec[ 5] 30.00-40.00 sec 11.0 GBytes 9.41 Gbits/sec[ 5] 40.00-50.00 sec 11.0 GBytes 9.41 Gbits/sec[ 5] 50.00-60.00 sec 11.0 GBytes 9.41 Gbits/sec- - - - - - - - - - - - - - - - - - - - - - - - -[ ID] Interval Transfer Bitrate[ 5] 0.00-60.00 sec 65.7 GBytes 9.41 Gbits/sec sender[ 5] 0.00-60.00 sec 65.7 GBytes 9.41 Gbits/sec receiver - Full-duplex (bidirectional)
1234567891011121314151617181920212223root@QWRT:~# iperf3 -t 60 -c 192.168.4.100 -i 10 --bidirConnecting to host 192.168.4.100, port 5201[ 5] local 192.168.4.1 port 38338 connected to 192.168.4.100 port 5201[ 7] local 192.168.4.1 port 38342 connected to 192.168.4.100 port 5201[ ID][Role] Interval Transfer Bitrate Retr Cwnd[ 5][TX-C] 0.00-10.01 sec 8.24 GBytes 7.07 Gbits/sec 0 4.20 MBytes[ 7][RX-C] 0.00-10.01 sec 10.9 GBytes 9.33 Gbits/sec[ 5][TX-C] 10.01-20.01 sec 8.25 GBytes 7.09 Gbits/sec 0 4.20 MBytes[ 7][RX-C] 10.01-20.01 sec 10.9 GBytes 9.37 Gbits/sec[ 5][TX-C] 20.01-30.01 sec 8.25 GBytes 7.09 Gbits/sec 0 4.20 MBytes[ 7][RX-C] 20.01-30.01 sec 10.9 GBytes 9.37 Gbits/sec[ 5][TX-C] 30.01-40.01 sec 8.25 GBytes 7.08 Gbits/sec 0 4.20 MBytes[ 7][RX-C] 30.01-40.01 sec 10.9 GBytes 9.37 Gbits/sec[ 5][TX-C] 40.01-50.01 sec 8.24 GBytes 7.08 Gbits/sec 0 4.20 MBytes[ 7][RX-C] 40.01-50.01 sec 10.9 GBytes 9.37 Gbits/sec[ 5][TX-C] 50.01-60.01 sec 8.22 GBytes 7.06 Gbits/sec 1 4.20 MBytes[ 7][RX-C] 50.01-60.01 sec 10.9 GBytes 9.36 Gbits/sec- - - - - - - - - - - - - - - - - - - - - - - - -[ ID][Role] Interval Transfer Bitrate Retr[ 5][TX-C] 0.00-60.01 sec 49.5 GBytes 7.08 Gbits/sec 1 sender[ 5][TX-C] 0.00-60.01 sec 49.5 GBytes 7.08 Gbits/sec receiver[ 7][RX-C] 0.00-60.01 sec 65.4 GBytes 9.36 Gbits/sec sender[ 7][RX-C] 0.00-60.01 sec 65.4 GBytes 9.36 Gbits/sec receiver
Pretty good. I repeated the last test with two parallel streams in case it’s CPU-bound (on the iKOOLCORE R2 Max):
|
1 2 3 4 5 6 7 8 9 10 11 12 |
root@QWRT:~# iperf3 -t 60 -c 192.168.4.100 -i 10 -P 2 --bidir Connecting to host 192.168.4.100, port 5201 [ 5] local 192.168.4.1 port 37078 connected to 192.168.4.100 port 5201 [ 7] local 192.168.4.1 port 37084 connected to 192.168.4.100 port 5201 [ 9] local 192.168.4.1 port 37100 connected to 192.168.4.100 port 5201 [ 11] local 192.168.4.1 port 37116 connected to 192.168.4.100 port 5201 ... [ ID][Role] Interval Transfer Bitrate Retr [SUM][TX-C] 0.00-60.00 sec 54.8 GBytes 7.84 Gbits/sec 0 sender [SUM][TX-C] 0.00-60.01 sec 54.8 GBytes 7.84 Gbits/sec receiver [SUM][RX-C] 0.00-60.00 sec 65.3 GBytes 9.35 Gbits/sec sender [SUM][RX-C] 0.00-60.01 sec 65.3 GBytes 9.34 Gbits/sec receiver |
Only slightly better on the Tx part.
Time to test WiFi connectivity, but adding a Xiaomi Mi Router AX6000 to our test bed. Here are the results for 5 GHz WiFi 6.
- Download
1234567891011121314devkit@UPX-i11:~$ iperf3 -t 60 -c 192.168.31.26 -i 10Connecting to host 192.168.31.26, port 5201[ 5] local 192.168.31.12 port 58422 connected to 192.168.31.26 port 5201[ ID] Interval Transfer Bitrate Retr Cwnd[ 5] 0.00-10.01 sec 2.02 GBytes 1.73 Gbits/sec 0 4.04 MBytes[ 5] 10.01-20.01 sec 2.08 GBytes 1.78 Gbits/sec 0 4.04 MBytes[ 5] 20.01-30.00 sec 2.05 GBytes 1.77 Gbits/sec 0 4.04 MBytes[ 5] 30.00-40.01 sec 2.06 GBytes 1.77 Gbits/sec 0 4.04 MBytes[ 5] 40.01-50.01 sec 2.08 GBytes 1.78 Gbits/sec 0 4.04 MBytes[ 5] 50.01-60.01 sec 2.07 GBytes 1.78 Gbits/sec 0 4.04 MBytes- - - - - - - - - - - - - - - - - - - - - - - - -[ ID] Interval Transfer Bitrate Retr[ 5] 0.00-60.01 sec 12.4 GBytes 1.77 Gbits/sec 0 sender[ 5] 0.00-60.02 sec 12.4 GBytes 1.77 Gbits/sec receiver - Upload
123456789101112131415devkit@UPX-i11:~$ iperf3 -t 60 -c 192.168.31.26 -i 10 -RConnecting to host 192.168.31.26, port 5201Reverse mode, remote host 192.168.31.26 is sending[ 5] local 192.168.31.12 port 49970 connected to 192.168.31.26 port 5201[ ID] Interval Transfer Bitrate[ 5] 0.00-10.01 sec 1.43 GBytes 1.22 Gbits/sec[ 5] 10.01-20.01 sec 1.78 GBytes 1.53 Gbits/sec[ 5] 20.01-30.01 sec 2.12 GBytes 1.82 Gbits/sec[ 5] 30.01-40.01 sec 2.12 GBytes 1.82 Gbits/sec[ 5] 40.01-50.01 sec 2.05 GBytes 1.76 Gbits/sec[ 5] 50.01-60.01 sec 2.01 GBytes 1.73 Gbits/sec- - - - - - - - - - - - - - - - - - - - - - - - -[ ID] Interval Transfer Bitrate[ 5] 0.00-60.02 sec 11.5 GBytes 1.65 Gbits/sec sender[ 5] 0.00-60.01 sec 11.5 GBytes 1.65 Gbits/sec receiver
Outstanding wireless performance at 1.77 Gbps download speed and 1.65 Gbps upload speed. So we have a winner here when it comes to wired and wireless networking performance.
Stress test and thermal performance
We tested the thermal performance of the Beelink EQi Wildcat Lake mini PC by running the 3DMark Fire Strike benchmark while monitoring the maximum CPU temperature under full CPU+GPU load with HWiNFO 64. The highest CPU temperature recorded was 86°C on the P-core (and overall package), and the utility didn’t report any CPU thermal throttling, although it reported power throttling. Interestingly, the E cores are always shown to have reached power limits, even after a reboot without running anything.
After a reboot, we repeated the test with the Cinebench R23 multi-core benchmark. The CPU temperature reached up to 86°C, and again, no thermal throttling was detected, but power limits were exceeded.
Overall, the cooling solution looks to be perfectly adequate.
Fan noise
The Beelink EQi Wildcat Lake is probably the quietest actively-cooled mini PC we’ve reviewed so far. The fan is barely audible even under heavy loads. We measured the noise with a sound level meter placed 5 cm from the top of the mini PC:
- Idle and light loads – 38.2 – 38.5 dBA
- Cinebench R23 multicore – 39.8 – 40.4 dBA
The sound level meter measures 36.4 – 37.0 dBA in the room when everything is quiet.
Beelink EQi Wildcat Lake power consumption
We measured the power consumption with a wall power meter as follows:
- Power off – 1.1 Watts
- Sleep – 3.2 – 3.3 Watts
- Idle
- WiFi 6 only – 5.7 – 6.2 Watts
- With 2.5GbE – 6.0 – 6.4 Watts
- With 10GbE – 6.8 – 7.2 Watts
- Video playback – 13.4 – 15.3 Watts (4Kp60 YouTube video in Firefox)
- Cinebench R23 multi-core
- First few seconds – 30 – 32.2 Watts
- Longer runs – 19.5 – 19.5 Watts
Remark: Unless otherwise stated, the mini PC was connected to WiFi 6, an RF dongle for a wireless keyboard and mouse combo, and a Crowview portable monitor over HDMI 2.0 during the measurements.

As noted in the unboxing and teardown part of the review, the Beelink EQi motherboard features a USB power jumper. By default, it’s set to provide USB power even when the mini PC is powered on or in sleep mode, and switching to “1-2 SO POWER” disables USB power. So I opened the case, moved the jumper to 1-2, and measured power consumption again:
- Power off – 1.7 Watts
- Sleep – 3.2 – 3.3 Watts
- Idle (WiFi 6 only) – 5.4 – 6.0 Watts
Not much difference, and I was still able to wake up the mini PC while in sleep mode. The USB RF dongle was connected to the front USB port.
Conclusion
The Beelink EQi Wildcat Lake Core 3 304 mini PC delivers excellent single-core performance, close to high-end Intel processors, which should be great for workloads such as web browsing, outstanding 10GbE and WiFi 6 throughput, and supports up to three 4K displays. 4K and 8K YouTube video playback works great up to 60 FPS and at low power, something I can not claim for many other recently reviewed mini PCs. The system is also the quietest actively-cooled mini PC we’ve reviewed, and the thermal solution is adequate without any thermal throttling detected under load.
Compared to the Alder Lake-N and Twin Lake families, the new Wildcat Lake family delivers higher performance in every way, but when we pitted the Beelink EQi 304 to a similarly priced AMD Ryzen 5 system, multi-core and 3D graphics performance were on the low side. UFS 3.1 storage performance is pretty good, but doesn’t match the throughput of NVMe SSDs. However, the main downside of this Wildcat Lake mini PC should be one of its key selling points: its two Thunderbolt 4 connectors. There appears to be a serious problem with PCIe link negotiation, so I had trouble testing NVMe enclosures (unreliable connection, but performance was great once connected), and an NVIDIA eGPU was not recognized at all. For this reason, I cannot recommend the computer at this stage, but I’m hopeful this is just a driver issue that will be fixed in time once the mini PC starts shipping to customers early next month.
I’d like to thank Beelink for sending the “Wildcat Lake Core 3 304” mini PC for review. The model reviewed here, with 16GB LPDDR5 and 512GB UFS 3.1 storage, is available for pre-order for $509, but as just mentioned, better to wait for clarification or a fix regarding the issue related to the USB4/Thunderbolt 4 ports.
Continue reading “Beelink EQi 304 Review – Part 3: Ubuntu 26.04 on a Wildcat Lake mini PC“.

Jean-Luc started CNX Software in 2010 as a part-time endeavor, before quitting his job as a software engineering manager, and starting to write daily news, and reviews full time later in 2011.
Support CNX Software! Donate via cryptocurrencies, become a Patron on Patreon, or purchase goods on Amazon or Aliexpress. We also use affiliate links in articles to earn commissions if you make a purchase after clicking on those links.



































