GMKtec EVO-X2 Power Consumption and Noise
The system came with a 19.5V 230W power adapter by HuntKey.
At idle, the system was great in the 8-14W range. That might be one of the big benefits of the integrated GPU. We also only measured 37-38dba at idle in our 34dba noise floor studio.
Running Llama 3.3 70B Q6_K inference, we saw 147-160W and would get into that 41-43dba range. Gaming, we were in the 170-180W range at 43-44dba.
The system was far from what we hear with loud systems. At the same time, I wish it was a bit quieter. Also, on the power consumption 170-180W from a single package with LPDDR5X memory is quite a bit.
Key Lessons Learned
To me, the big lesson learned was really on the AI side. We knew that the gaming performance was going to be very good just because it is a 40 CU RDNA 3.5 GPU. To me though, the gaming performance would not sell me on this machine at $1,999. There are too many other options in the market to hit this level of raw GPU performance.
What sells this system to me is very simple: 128GB of LPDDR5X, that 40 CU RDNA 3.5 GPU, and AMD’s ROCm support.
Partitioning the system with 64GB for the CPU and 64GB for the GPU or 32GB for the CPU and 96GB for the GPU completely changes the usefulness of the system. Running local 70b parameter models we saw the impact immediatley when we were filming of having larger higher quality models that provided useful results versus smaller faster models that ultimately gave us unusable output.
The hard truth about today’s generative AI is this: bigger models may be slower, but they are better. Or put another way, if you truly want to offload tasks to AI agents, you want to use the highest quality models possible. Speed is important too. We saw almost double the performance by switching from a partial CPU offload to a fully GPU accelerated model.
If you are doing AI, and are willing to give up some speed, then this is the lower-cost solution that can run models that are too big to fit in the memory footprint of a NVIDIA L40S. To be clear, the NVIDIA GB10 systems are going to be very neat in this space with their ConnectX networking built-in for clustering. Although we expect consumer/ Windows versions of the GB10 at some point, the 128GB AI version is going to be $3000+. AMD’s offering is available today, and at a lower cost. Perhaps AMD’s biggest competition right now for boxes like this is the Apple Mac Studio. The M3 Ultra 512GB version we have in the studio is simply awesome for AI, but it is also a $10,000 machine. Running large models on a $2000 device with up to 96GB dedicated for the GPU is actually quite a bargain if you view it form the angle of AI system and component pricing.
GMKtec had the wisdom to put this into a small package that can be powered on the desktop. I wish that at this price it also had things like 10GbE and the serviceability was a notch better. Still, the big value here is in the APU.
Final Words
From a system perspective the GMKtec EVO-X2 is in-line with what we have seen from other AMD Ryzen AI Max+ 395 systems, and it actually worked well in our testing. $2000 for a mini PC feels like a lot if you are looking for basic desktop functions or gaming. From an AI perspective, this is a really strong system.
Here is perhaps the crazy thought. If you have a regular task that can be automated by a 70b parameter model, or you want to run larger image or video generation AI tasks, the $2000 price might actually be a bargain. There is enough here that I do not think that this system is perfect, but also I really like this little box because it has some excellent capabilities.
It is a great system, but it’s been a while since I’ve seen such an ugly enclosure on a modern computer.
Pls inform me cost of evo x2
Make it 10GbE, quieter, and not an ugly AF box and it’ll sell like sno cones in Hawaii
what makes it ugly compared to a square black box?
I’d agree with others this is a horror show on the eyes. At least a square black box is easy to hide. This doesn’t have clean lines and 2 tone with RGB
In my opinion the Ryzen processor is very nice. At the same time, I’m not interested in a weird looking mini PC that is difficult to service. On the other hand, if there were a mini ITX motherboard with this processor, I could see placing four in a suitable 2U rack-mount case.
I like it aesthetic much more than a box, give me interesting asymmetrical every time over “it’s a box.
I differ on the claim “it’s meant to be vertical” , the feet were thought after naming ports, if you want to read it the way it’s meant to be read, this is a horizontal box.
They probably decided it’d have smaller footprint on desk or against a wall while vertical aaaand, “lets add feet on edge”
Where did the get the designer for this thing from? Such a great machine in such a poor housing! Looks like a 1987 clone pc!
Ugly?!? It’s a dam computer it’s a new platform . Why is there always something to bitch about. Couldn’t complain about it performance so this is you complaint???
WOW
Great review, thanks a lot! But what about image and video gen? Is that possible with 96GB RAM dedicated for the GPU?
(a) Typo on this page (bargin, rather than bargain)
(b) You mention Windows on GB10. Is that actually confirmed or conjecture?
So much hate on the design of the case. I mean, yeah, it’s a little weird looking, but I’m fairly certain I’ve seen worse. I’m indifferent to case aesthetics, myself. Form follows function, as I see it. If the thing works as the manufacturers intended, that’s all that matters. RGB, funky stepped housing design, all that’s secondary to the actual functioning of the unit. Yeah, there’s a few places where they could have done better, but if it works and does what its designers set out to accomplish, is the visual aesthetic really THAT big an issue?
The box is meant to stay vertical. If you let it stay horizontal, you are going to clog the cpu ventilation holes. I have been working on the evo-x2 for about two weeks and I am quite happy with it.
The FA-EX9 looks a bit more promising as a workstation, no 10gbe but 128GB 8000mhz RAM like the X2 and it has oculink (seems it’s shared with the second nvme port?). It’s starting to popup on aliexpress but I haven’t found a full review yet.
Oculink seems like a great choice until you realize how much an additional gpu enclosure and gpu costs. You might end up with another 2k.
Frame.work has an mITX board with 128Gb of RAM for 2000$, shipping end of the year if ordered now. Jeff Geerling has a video where he put them in a miniRack. An miniPC with triple fans isn’t really what I want on my desk. As an enthusiast I’d rather put it in a jonsbo N5 with lots of storage as an all-in-one NAS/compute server.