A pocket-sized 2.8-inch (72 x 72 x 45mm) mini PC based on an Intel Processor N100 SoC with 12GB LPDDR5, and up to 2TB storage has started to show up on Aliexpress for $154 and up under the names ZX01 Plus and Topton M6S.
This is the most compact Alder Lake-N mini PC we’ve seen so far, yet it still comes with two HDMI 2.0 ports, a 3.5mm audio jack, Gigabit Ethernet and WiFi 5 connectivity, and three USB 3.2 ports, plus a non-standard 12V USB-C port that’s now common for this type of mini PC.
ZX01 Plus / Topton M6S specifications:
- SoC – Intel Processor N100 Alder Lake-N quad-core processor @ up to 3.4 GHz (Turbo) with 6MB cache, 24 EU Intel HD graphics @ up to 750 MHz; TDP: 6W
- System Memory – 12GB LPDDR5 @ 4800 MHz
- Storage
- Optional 128GB to 2TB M.2 2242 SATA SSD, but the socket is also said to support NVMe SSDs (PCIe 3.0 x2)
- MicroSD card slot
- Video Output – 2x HDMI 2.0 up to 4Kp60
- Audio – 3.5mm audio jack, digital audio via HDMI
- Networking
- Gigabit Ethernet RJ45 port
- Dual-band WiFi 5 and Bluetooth 4.2 via Intel 7265AC module
- USB – 3x USB 3.2 10Gbps ports
- Misc – Power button, smart silent fan for cooling
- Power Supply – 12V/3A via USB Type-C port (not USB PD compatible)
- Dimensions – 72 x 72 x 44.5mm
- Weight – 150 grams
If you do NOT order the barebone model, the mini PC will ships with Windows 11, and the company specifically says it does support Windows 10, but as we’ve just seen in Beelink EQ12 review, the Intel Processor N100 can run various operating systems such as Ubuntu, Debian, TrueNAS, pfSense, and so on. Ian’s review did however mention that WiFi and Bluetooth did not work. It’s not the same module here, so we’d have to see…
The mini PC ships with a 12V power supply, an HDMI cable, and a user manual. The $154 price tag is for the barebone model with 12GB LPDDR5 RAM and no storage, but sellers provide the mini PC with storage for about $167 with a 128GB SATA SSD and up to $268 with a 2TB SSD. The price is not that much different from the feature-rich Beelink EQ12 mini PC, so I’d say the ZX01 Plus/Topton M6S pocket-sized Intel Processor N100 mini PC is mostly interesting to people who care about the ultra-compact form factor.
Via Liliputing and AndroidPC.es
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
It’s amazing how small these are getting! Intel phone when?
I don’t think they are interested in x86 phones, let alone with a 6W TDP for CPU only. Z8300-8500-8700 were the only ones really suitable for low cost phones.
> a 6W TDP for CPU only
Why do people still talk about these 100% meaningless TDP numbers?
Intel’s TDP ratings are marketing BS and only valid at base frequency (that’s laughable 800 MHz for N100) while the things are supposed to run at burst frequency (2.9/3.4 GHz multi/single threaded). According to Ian’s latest review of Beelink EQ12 N100 consumes +16W fully loaded (CPU only with DDR5) compared to idle.
Intel doesn’t give Base Frequency numbers for those, it’s very likely to be higher than that. Those base frequency numbers are very useless since they will depend too much on what is being run at the moment.
Say, go execute a lot of AVX2 with Prime95, it will probably start at rather high frequency than go down a lot. How much is hard to say as it can depend on the rest of the system too(e.g. how much the GPU, memory controller and stuff are being hit).
They assure it will at least be 800 MHz though, from what I observed the tendency is to go above that for sustained maximum frequency, unless you are thermally throttling(very possible for those small PCs, many of them are even passively cooled!).
Do note that this behavior is basically the same everywhere, that’s why phone reviewers execute many loops of geekbench and observe the behavior.
If you look at the Windows system info screenshot, you’ll see the frequency is 800 MHz.
https://cdn.cnx-software.com/wp-content/uploads/2023/05/EQ12-Intel-N100-Windows-11.png?lossy=0&strip=none&ssl=1
But you’re right that in practice the sustained frequency will be somewhere between 800 MHz and the turbo frequency. It’s never advertised but can be seen in HWiNFO64 while stress-testing the system.
> the sustained frequency
This is another (un)related area. But we were talking about this TDP BS. @linuxium reviewed recently N100 (6W TDP) and N305 (15W TDP) boxes. Consumption difference between idle and ‘full load’ was ~15W with N305 but +16W with N100 but of course N305 outperformed N100 substantially (4 more CPU cores).
While the TDP BS suggests N100 being way more efficient than N305 the exact opposite is true.
> the sustained frequency will be somewhere between 800 MHz and the turbo frequency
And a final one. Based on @linuxium tests with sbc-bench it seems on these Alder Lake N thingies once the CPU temperature hits 70°C the memory controller will be severly throttled while CPU clockspeeds still remain high (strong mismatch between compression/decompression speeds with 7-zip’s internal benchmark).
I’m sorry but you’re comparing apples and oranges here, due to both not having the same number of cores nor frequency. The reality is that one was running at ~30W while the other was running at ~22W but at different workloads. A good comparison would involve running them both at the same frequency and the same number of active cores. There we could see which of the two consumes the least. And it remains possible that under such a workload the N100 would consume slightly less despite probably running on lower quality bins, just due to all the extra circuitry in the N305 dedicated to the extra cores. I wouldn’t like to be a hardware reviewer these days, as it’s never simple to compare two chips on a similar workload because even users’ expectations change depending on what they’re doing (e.g. race-to-idle on laptop to save battery by allowing the user to finish a task faster and close it vs watching a move with the least amount of juice being used while the user is completely passive).
> The reality is that one was running at ~30W while the other was running at ~22W but at different workloads
Nope, I was comparing using Ian’s reviews of Beelink EQ12 (N100) and Weibu N10 (i3-N305) using exact same workloads (Windows Cinebench, Ubuntu stress) and looking at the consumption delta between idle and ‘fully loaded’.
> Intel doesn’t give Base Frequency numbers for those
Not in their slides or on ARK but as ‘CPU metadata’ it’s available (use dmidecode or similar tools). Based on what Geekbench lists as ‘Base frequency’ combined with Intel’s TDP marketing BS we know this about these SKUs:
* N50: 2 cores, 3400 MHz / ? MHz @ 6W
* N95: 4 cores, 3400 MHz / 1700 MHz @ 15W
* N97: 4 cores, 3400 MHz / 2000 MHz @ 12W
* N100: 4 cores, 3400 MHz / 800 MHz @ 6W
* N200: 4 cores, 3700 MHz / 1000 MHz @ 6W
* N300: 8 cores, 3800 MHz / 800 MHz @ 7W
* N305: 8 cores, 3800 MHz / 1800 MHz @ 15W
(no N50 entry appeared in Geekbench browser so far)
He mentioned 6W as being too much for a phone, so 16W would avert him from this idea even more.
While the 6W ‘number’ is pure marketing BS the 16W number is questionable as well. Since so far all the reviewed Alder Lake N boxes show grotesque high idle consumption and running ‘fully loaded’ they need north of 25W!
Who knows whether it’s some silly UEFI bug that will never get fixed for these cheap boxes? Most people block out that UEFI is something entirely different than ‘good old’ BIOS…
Tried and failed I’m afraid back in the early 2010s. X86 can shrink pretty far but struggles in a phone power envelope.
To be fair, they were manufacturing Atom Z2560 (Geeksphone) on a 32nm process. and battery technology is a lot better than a decade ago.
If it happens, it won’t be x86, Intel’s future forays into phones will be fabbing other people’s RISC-V designs.
This has more to do with chip design than ISA advantage.
The advantage of ARM over x86 is low single digit.
These n100 are about the same performance as the i5 6500 in the ex-corporate refurb mini PC I just picked up. It’s no slouch for everyday tasks .. and the n100 does it with 6w. Pretty impressive! 🙂
> and the n100 does it with 6w.
Scary. Marketing BS really works.
Ok fine – idle consumption. Maybe Intel ought to list something besides just that? 🙂
The atom based celerons of the 6500s era were terribly plodding … Z8350 etc. I think it’s a big improvement.
> Ok fine – idle consumption
Huh? Intel’s TDP ratings are not about ‘idle consumption’, they claim to be the thermal challenges of some imaginary workload involving CPU and GPU at some imaginary clockspeeds. And in reality they’re 100% pure marketing BS since these ‘numbers’ have no relationship with anything happening in the real world at all.
But it works. Search for ‘N100 consumption” and you’ll have a hard time not being confronted with the 6W marketing BS (+25W in reality).
I guess you’re right. I usually think about it as somewhere close to idle, and the “prior” TDP something close to full load. I think Intel changed TDP around the time when their tablet chips were losing out to power sipping ARM? And their logic about TDP helping to size the needed cooling solution makes .. some .. sense. But while it might be useful for an OEM figuring out that or battery life, it’s NOT very useful for anyone trying to size a power supply or gaming CPU cooler … imo. Looks like briefly they also disclosed base and turbo power, but I can’t find it for the n100.
I’ve seen similar 15 to 25w draw from people’s testing of the n100 and n305. Anecdotal but seems about right.
Intel used the SDP (Scenario design power) numbers for phone/tablet chips to further muddy the waters.
To do a bit of justice to CPU vendors, it’s becoming increasingly difficult to predict power usage nowadays: the top frequency isn’t always held for long depending on the instructions used due to thermal concerns, so even the power drawn under such conditions isn’t useful for thermal cooling. In addition nowadays CPUs can be configured for different average power usage levels. And these ones are just made to limit heating and preserve batteries while being fast for short periods (typical usage: browser fetching a complex page quickly and idling while the user reads it). Mainboard makers will design their power regulators based on the max current in the datasheet. For example if you see max 250A at 1.1V it means 275W peak power. But cooling is another story due to the many variations above. For server CPUs I’ve seen both “TDP” and “max turbo power” columns, with the last one being roughly 60% higher. It makes sense since server CPUs tend to maintain highest power levels longer or even forever. But for entry level CPUs I guess it’s even more complex than for regular desktop CPUs due to the wide frequency and voltage variations, and the ability of the chip to regulate its higher frequency bins based on the die temperature, which can vary a lot in such a situation :-/
> the “prior” TDP something close to full load
At certain clockspeeds (“Base frequency”) that have become irrelevant over the years since they’re so much off from those clockspeeds the CPUs will really run: with N100 it’s 800 MHz vs. 2900/3400 burst clock (multi/single).
And wrt idle: at least based on Ian’s reviews of Alder Lake N boxes here there’s something seriously wrong since those PCs all idle above 10W.
maybe I’m wrong for general consideration, but i think i read that (at least for one device) a M.2 SSD on high performance quality level for high performance demand disk I/O (production line including U.2/U.3 E1.x/E3.S devices also) could have idle power at ~5W (with probably higher connection bandwidth ~PCIe3-4 and some pretty high capacity ~1-3xTB).
Seems to me, with today’s variety on consumer hardware CPU’s (and peripherals variety) it’s very often a comparison, that ‘should’ be called summary with possibility for, more or less momentarily (including progress on firmware/software, technology support, clearances on cost), variances?
Getting this more standardized, it could be asked “What’s idling?”, “What’s CPU’s (types of cores?) idling power?” and “What are most demanding devices on idling or performance for several tasks?”. Not every platform is prepared for such detailed insights to users(?)
(simplest approach would be cpu and memory only idling power (with knowledge about software influence)?)
Just got an i5-6600T. I assume the main advantage of an N100 would be substantially faster integrated graphics, along with the video decode capability.
I ordered a 6500T box but it came with higher power 6500.
Yes, you are right, imo, re advantage. Also heat.
Never buy anything with a 12V USB Type-C. Just plain stupid.
I agree, and it seems to become more and more common 🙁 At work we bought two MeLE Quieter3 or something like this. they’re really great machines for a meeting room, except that they have that shitty 12V USB-C adapter. I had to put some color tape on it with a big warning sign, hoping nobody will ever have the idea of plugging it into their phone or such a thing! I’m extremely angry at them just for this. Otherwise the device is cool and perfectly matches our needs.
In fact I’m wondering if one solution couldn’t be to make some barrel-to-USB-C adapters to plug and glue into such devices so that they’re powered by real 12V supplies coming with a barrel connector. Or just cut the adapter’s cable in the middle and attach a pair of male and female barrel connectors. This will repurpose the PSU and avoid any risk.
> Never buy anything with a 12V USB Type-C
In addition: never buy anything with a fan online if the return procedure is not simple (since that fan might be annoying as hell) and never buy any PC thingy from vendors known to not provide UEFI updates (applies to other firmwares too, e.g. for Wi-Fi chips).
So is their anything left to buy?
Seriously even “big” vendors often stop shipping uefi/*ware updates very quickly (depending on the purchase date and how long a device was already on the market one can be already happy for 1-2 years updates).
The short lifecycle of products and the (to some extend) included planned obsolescence doesn’t look to get better in the near future.
My problem with this computer is that it comes from China. It could be preloaded with malware embedded in the hardware with no way to remove it or disable it without destroying the computer. China could use it as a backdoor
> It could be preloaded with malware embedded in the hardware with no way to remove it or disable it
True for every CPU designed in the US.
Now we know why you are so actively commenting.
At least you had some reasonable reasons when talking about TDP for every comment..
TDP ist meaningless. You need to measure power consumption while you running standard tasks: OS loading, idling, and then stress tests (CPU/GPU/network). Also you need to add that RAM works in single channel.
In Vietnam and got one of these little bad boys last week. Using it for retro gaming and as my sonarr/radarr/plex media center.
Have 2 x 1TB drives (enclosure with SSD M.2 NVMe ) attached to the front USBs. And a mouse dongle at the back.
Just to report a few things.
1. It’s fricking awesome
2. USB 3.0 only
3. DDR4 ram not 5
If it’s a media center or a retro gaming PC your after (ps2 max), or just something to browse the web/do non taxing work on this is a beast for the size of it.
Blown away. Immediately started looking for SSD M.2 NVME NAS devices, cause 2TB isn’t going to last me long!