First AirJet-Equipped Mini PC Tested (tomshardware.com) 49
An anonymous reader quotes a report from Tom's Hardware: Zotac's ZBox PI430AJ mini PC is the first computer to use Frore System's fanless AirJet cooler, and as tested by HKEPC, it's not a gimmick. Two AirJet coolers were able to keep Intel's N300 CPU below 70 degrees Celsius under load, allowing for an incredibly thin mini PC with impressive performance. AirJet is the only active cooling solution for PCs that doesn't use fans; even so-called liquid coolers still use fans. Instead of using fans to push and pull air, AirJet uses ultrasonic waves, which have a variety of benefits: lower power consumption, near-silent operation, and a much thinner and smaller size. AirJet coolers can also do double duty as both intake and exhaust vents, whereas a fan can only do intake or exhaust, not both.
Equipped with two of the smaller AirJet Mini models, which are rated to cool 5.25 watts of heat each, the ZBox PI430AJ is just 23.7mm thick, or 0.93 inches. The mini PC's processor is Intel's low-end N300 Atom CPU with a TDP of 7 watts, and after HKEPC put the ZBox through a half-hour-long stress test, the N300 only peaked at 67 C. That's all thanks to AirJet being so thin and being able to both intake and exhaust air. For comparison, Beelink's Mini S12 Pro mini PC with the lower-power N100, which has a TDP of 6 watts, is 1.54 inches thick (66% thicker than the ZBox PI430AJ). Traditional fan-equipped coolers just can't match AirJet coolers in size, which is perhaps AirJet's biggest advantage. Last month, engineers from Frore Systems integrated the AirJet into an M2-based Apple MacBook Air. "With proper cooling, the relatively inexpensive laptop matched the performance of a more expensive MacBook Pro based on the same processor," reports Tom's Hardware.
Equipped with two of the smaller AirJet Mini models, which are rated to cool 5.25 watts of heat each, the ZBox PI430AJ is just 23.7mm thick, or 0.93 inches. The mini PC's processor is Intel's low-end N300 Atom CPU with a TDP of 7 watts, and after HKEPC put the ZBox through a half-hour-long stress test, the N300 only peaked at 67 C. That's all thanks to AirJet being so thin and being able to both intake and exhaust air. For comparison, Beelink's Mini S12 Pro mini PC with the lower-power N100, which has a TDP of 6 watts, is 1.54 inches thick (66% thicker than the ZBox PI430AJ). Traditional fan-equipped coolers just can't match AirJet coolers in size, which is perhaps AirJet's biggest advantage. Last month, engineers from Frore Systems integrated the AirJet into an M2-based Apple MacBook Air. "With proper cooling, the relatively inexpensive laptop matched the performance of a more expensive MacBook Pro based on the same processor," reports Tom's Hardware.
70C != good (Score:4, Interesting)
Keeping the CPU temperature below 70C is nothing to write home about. I did burn-in tests on a selection of mini-pcs for a project that needs to deploy hundreds of them. The first one to crash from heat did it at 60C. The target temperature is 50C under load.
Re:70C != good (Score:4, Interesting)
Keeping the CPU temperature below 70C is nothing to write home about. I did burn-in tests on a selection of mini-pcs for a project that needs to deploy hundreds of them. The first one to crash from heat did it at 60C. The target temperature is 50C under load.
YMMV. It clearly depends on the CPU. My Intel i7-10700K is setup with a warning threshold of 90C (It starts throttling at 100C), so 70C is more than good.
Re: (Score:2)
Once again, the benchmark has changed and I think, not for the better. You don't want 70C on your lap (or in your face, for a mini-case), which is why the tradition is 50C even when modern CPUs don't burn until 105C. That heat has to go somewhere and yes, most of it is going into fan-driven air. The heat not going into the air means temperatures higher than 50C, ages the motherboard quickly.
Re: (Score:2)
What I meant is that 70C as a maximum is fine. I'm not crunching numbers on all cores at all times. Only on startup, compilation and a few other things. I don't expect the CPU to be at 70C all day long, only when I need crunching power.
Re: 70C != good (Score:2)
Re: (Score:2)
I ran programs in a loop consuming 100% of the CPU cores under Linux. After about a week of this, one of the machines at 60C crashed reporting an MCE (machine check exception) in the CPU.
I also noticed that one of the machines lied to me about temperatures outside the CPU. It reported ambient 29C and NVME at 39C when booted cold, and the exact same temperatures when the outer case was burning hot to the touch and the CPU was reporting 60C.
Re: (Score:2)
I also noticed that one of the machines lied to me about temperatures outside the CPU. It reported ambient 29C and NVME at 39C when booted cold, and the exact same temperatures when the outer case was burning hot to the touch and the CPU was reporting 60C.
Why does that have to be a "lie?" Just because the case is hot does not necessarily mean the air inside is--especially when you have fans that are replacing the volume of air inside the case every few seconds.
Re: (Score:2)
It was a passively cooled machine. The case itself was one big heat sink. No fan.
Re: (Score:2)
Also, like I said: the temperature didn't change. It was reported to be exactly the same number, to decimal points, no matter when read. That just doesn't happen in consumer electronics, not unless it's a lie.
Re: (Score:2)
Re: (Score:3)
Dishonest computer parts are the worst.
No.
Dishonest Test Equipment is the Worst!
Then every gamer is doing it wrong. (Score:5, Informative)
I don't know if AMD chips still can't run toaster-hot like Intel chips and maybe that's your reference point, but there's a reason Intel chips don't throttle until upwards of 90C, and then they try to hover just below throttling temps. Apple's custom chips are this way as well. A modern CPU should be able to run continuously at 70C for years without a problem. I don't own a single machine (other than my tablet which I consider a different build philosophy) that can sustain 50C under load, they're always in the high 60s or low 70s. When I had a fanless Chromebook, it would quite regularly get up in the high 80s with a similar little 2-core Atom (N-series Celeron), to the point where it became a kneetop rather than a laptop because I didn't want to have any skin under the hot spot. So I don't know what hardware you were speccing, but your temp targets sound unreasonably low. I'm not saying that running cooler won't make things live longer, it probably will -- but that's not how the products are designed, for the most part. They're designed to run as fast as they can for as long as they can given whatever cooling they're provided, then throttle.
Re: (Score:2)
Even if Intel's CPU can run toaster hot, the chips around them like DRAM and SSDs cannot.
I want to deploy hardware to guest sites where I have no control over environmentals and don't always have continuous contact with the site host. I need it to last a decade, and it's not going to do that at 90C. Or 70C.
Re: (Score:2)
I hope you plan on deploying industrial PCs made for harsh environments.
Re: (Score:2)
Zen4 is designed to run AT 95C (to maximize boost headroom)
Re: (Score:2)
I wasn't sure. I remembered that back in the Athlon days, AMD CPUs didn't even have throttling in the modern sense. They'd catch fire if the heatsink fell off in service, and TJmax was about 70C while Intel's was already around 105C where it has remained. This is in spite of owning a 6-core Thuban for over a decade. I've never seen it spike above 83C, so I thought that (at least for 2011) AMD still had a lower TJmax than Intel.
Spazmania should probably be looking in a market other than off-the-shelf consume
Re:70C != good (Score:5, Informative)
> The first one to crash from heat did it at 60C.
I ran into this problem a month ago on an Intel platform and chased a wild goose on cooling. Machine would hard-reset at about 63C at the on-die sensor. Both Brave and ffmpeg were fingered as "bad software that will overheat your machine".
And "probably linux can't use fans right".
Long-story short: heat is proportional to power draw and the cheap-ass power brick that came with the machine was crapping out over a certain current.
"But the manufacturer supplied it."
$14 to Amazon and two days later the machine is 100% stable under any load.
Also reminded: never give credence to the customer's theory.
Re: (Score:2)
That's fair. And I have not confirmed the voltage delivered by the power brick when the machine is under load. For my purposes the question was moot: if I have to buy another power supply then I miss the target price point.
Re: 70C != good (Score:2)
Shipping junk isn't profitable in the long term.
Re: (Score:2)
Look at the specs.
Modern chips are advertised to run at these temperatures without issues:
https://www.amd.com/en/product... [amd.com]
Take AMD Ryzen 5800X3D, it clearly specifies: "Max. Operating Temperature (Tjmax) 90C". Under 3 years warranty period, you should be able to run it at 90 degrees 24/7 without issues.
The problem is, if the chip has defects, they will be amplified with the heat. So, yours resetting at 63C is a symptom of deeper issues. I hope it lasts a bit more, but over time, it is likely those issues w
Re: (Score:2)
My test results for mini PCs if anyone cares:
https://bill.herrin.us/network... [herrin.us]
Re: (Score:2)
Keeping the CPU temperature below 70C is nothing to write home about.
Yes it is. It is below thermal throttling limits for virtually all SKUs from Intel and AMD.
The first one to crash from heat did it at 60C.
No it didn't. Your heat measurement was incorrect. There's not a chip an the market that would crash from such a pathetically low temperature.
Re: (Score:2)
No it didn't. Your heat measurement was incorrect.
It was an Intel N3160 and the temperature was as reported by the N3160's internal sensors.
Re: (Score:2)
https://www.intel.com/content/... [intel.com]
TJUNCTION 90 C
Re: (Score:3)
Depends on silicon binning.
Datacenters use Xeon CPUs, which are rated to run over 90c for extended periods of time:
https://forums.anandtech.com/t... [anandtech.com]
Recent "gaming" phones also run very hot, some even come with external (or even internal) fans.
https://arstechnica.com/gadget... [arstechnica.com]
There is nothing that stops a very good CPU to run at these temperatures non-stop. However cheaper, lower quality ones will obviously fail early.
(Lower quality in terms of physical silicon defects. Usually you'd not want to throw away d
Re: (Score:2)
7 watts is the "average" power dissipation, not the maximum. The maximum is 25 watts, and it was drawing at or near that in the test where it ran at 70C.
Passively cooling 25 watts to a 30 degree differential with ambient takes a much larger heat sink than fits in the physical design they talked about in the article. Much larger.
How does this cope with dust ? (Score:3)
One problem with fans is that they get clogged with dust which makes them not work as well. Once a year I use an air blaster to blow dust out of the inside of my PC; one problem with that is that it can push dust into fan bearings - which is not good. I assume that this will attract just as much dust but, not having rotating bearings, will not suffer when I use an air blaster.
Can anyone enlighten me ?
Re: (Score:2)
I watched a video a while back where the CEO said you could either blow the unit out with air or a computer system utilizing one could momentarily reverse the flow at startup which will loosen any dust to be blown out when the normal flow resumes. Either way, I still think dust will be a major issue with these.
Re: (Score:2)
On this page, they specifically mention that the unit's backpressure is sufficient to let it work with dust filters.
https://www.froresystems.com/a... [froresystems.com]
It's $600 (Score:4, Informative)
Fourteen clicks later:
https://www.zotacstore.com/us/... [zotacstore.com]
Re: (Score:1)
The tech market is waiting for them to go bankrupt so they can clone it.
Thought I recognized this - From 1983! (Score:5, Interesting)
Hardly a new concept. I'm looking at a Swedish electronics magazine from 1983 here (yes, they used to come already printed, on paper!)
There's an ad for miniature piezoelectric fans, no moving parts. The smaller version is 71 x 17 x 71 mm, so perhaps a bit larger than what's available today.
Remember that it was already a commercial product by then, and back in 1983 in Sweden the market was always far behind what was available in the USA, seeing how there wasn't really a huge free international market etc.
"a fan can only do intake or exhaust, not both" (Score:2)
Funny, I reverse the polarity on my case fans and I can do intake or exhaust. Or just physically rotate them 180 degrees on the Z-axis.
Re: (Score:1)
Re: (Score:2)
Well, duh. You're only supposed to reverse the polarity of your shields!
Re: (Score:2)
Case fans are DC and run in either polarity. The stupid LEDs on them, not so much.
Re: (Score:2)
When I reverse the polarity, it emits a tachyon pulse...
Re: (Score:2)
Near silent (Score:3)
Instead of using fans to push and pull air, AirJet uses ultrasonic waves, which have a variety of benefits: lower power consumption, near-silent operation
I'm going to have to get my dog's opinion on this.
Re: (Score:2)
I was thinking the same thing. If the ultra-sonic waves are audible to a dog or other household pet, that could be a form of abuse every time you open you computer.
Re: (Score:2)
How do you know that the existing CPU/case fans don't produce a bunch of ultrasonic harmonics already? It seems like this new device would only be a problem if it were excessively loud in the ultrasonic range.
AirJet uses ultrasonic waves, (Score:2)
Prepare to drive your pets beserk...
Cat deflection field (Score:2)
Cats can hear up to roughly 65KHz, so this thing should send them shrieking and running to the other side of the house. I think we need to do a test...
Re: (Score:1)
Interesting Product, Unimpressive Name (Score:2)
skinny pc bad. really really skinny pc good. (Score:2)
Literally the only thing I look for when shopping for a laptop is how skinny it is. Nothing else matters.