Now all you have to do is update the picture: https://linuxmint.com/tmp/blog/3528/thumb_mini2_2.png
(I'd love if you could append this to the comments on that page)
I read the comments on this blog, and - quite frankly - I really wish I could reach out and SMACK
some of the people posting there!
A lot of the negative comments were about the choice of processor - Intel vs AMD - and some of them were pretty vicious. What really
burns my biscuits about stuff like that is that none, repeat ZERO
, of the people posting appear to have ANY
experience with a hardware design project of any
I've done that. Maybe not so complex as a fanless PC, but I've done embedded controller design projects back before "embedded controller" and "i2C" were buzz-words. And it's not easy. There's always trade-offs that have to be made. Sure, Compulab would LOVE
to put a full-spec i7 processor in their boxes, (or an AMD Ryzen unlocked monster), but - at it's most basic level, they're BIG
- and PC board real-estate is not unlimited.
The power budget for a machine that size is not something that can be taken for granted. In fact, it can become incredibly complex. The Raspberry Pi Foundation discovered that quickly with the Pi-3. All of a sudden power supplies that would power the Pi-2 comfortably, were now woefully inadequate.
The system's heat budget isn't as simple as it sounds either. All of those big "Bad Boy" processors generate heat - lots of heat - and cooling them isn't a trivial exercise. It doesn't make sense to build a fanless system if the only way to cool it is to drop it in liquid nitrogen.
The choice of processor? OK, maybe at the time these postings were made, AMD had a low-power Ryzen-based SOC on the horizon, but "on the horizon" doesn't sell boxes. "On the horizon" doesn't cut it with FCC / UL / C-mark / European qualification testing either. And all that testing takes time. LOTS of time. And LOTS of money. They needed to find something available THEN
, not "on the horizon" - with enough lead time to test, fail, re-test, re-fail, and then finally pass all the qualification testing for all the markets they want to sell it in. Even if you get lucky and pass all the tests first time, all of that testing can take almost a year to complete. And they still haven't sold a single box yet to offset the boatloads of money they're pouring into it.
OK, it pukes when tested with the Cinebench benchmark test? Jeez! What'd you expect?! If you want to buy a pocket-sized system like that, and expect to run Blender in full render mode, you've got wires loose.
What's it good for then?
- I can see kiosk-based systems scrambling to get a box like that.
- It fits the back of any reasonable - and some unreasonable - flat-screen monitors, so it would make a great office workstation, Point-Of-Sale endpoint, public system, etc. etc. etc. It has more I/O ports than a cat has hair, so you can plug in stuff like displays, bar-code scanners, payment card readers, etc.
- It sips power and can handle the full MIL-STD temperature range - and it's built like a brick latrine - so you could use it in virtually any industrial application. What's exceptional about that system in this context is that - normally - an industrial-grade machine that can handle that kind of temperature range would cost thousands of dollars - and that for the stripped-down version.
- It (supposedly) runs Mint like it was designed for it. And it was! If I had a place for it, (and my wife would let me spend the money), my wallet would come flying outta my jeans so fast it would leave scorch marks!
Great work finding and fixing this, and a top-notch job by Compulabs!