I've been wanting this for years. =D

Pricing is on the high side, but reasonable-ish I think lol. I was a bit put off by their ultrabook push, where they basically phased out their cheap ULV processors/Netbook solutions. I too can't wait until there's broad support, maybe it needs to be standardized so other manufacturers can jump in. Then they can start their race to the bottom! hah.

My hat is off to Intel for this one. After surviving just fine sans video card for a good while, I realise how much wasted space even a micro-ATX setup is. The future looks bright indeed, if your idea is the same as mine: minute and mute.

Thanks for the review of this, it sounds very tempting. I'm a little surprised by the price - in a good way. It's not cheap, but I was expecting it to be closer to £300/$450 for the base unit, given it's Intel and all that. Hopefully it'll prompt the other mini-PC makers into a price war.

Very interesting and I'm sure it'll cover a huge percentage of computer users realistic needs. I still see it as too restrictive for what I'd want so I'd stick to MicroATX for now.


I hope Intel realise that they can't actually do this for all markets. For the UK, electrical appliances must be supplied with a UK plug. Supplying the wrong plug or no plug is illegal.

I like it in the same sense as a Nexus phone from Google. It's a "Hey! Look what can be done" build.

There's ITX too. Unfortunately lots of cases waste way too much space on drive bays.

"The tickle here is that only-USB 3.0 would mean there's no easy way to install most operating systems, as native USB 3.0 support is found only in Windows 8."

USB 3.0 is backward compatible with USB 2.0, so that shouldn't really matter.

Except that it does. USB3 controllers support USB2 devices, but that doesn't imply USB3 controllers can be 'driven' by USB2 drivers.

The table on page 4 says 45W for CPU+GPU Load, whereas the graph on page 5 says 40W. Which one is correct?

Well, no, of course it doesn't - but I fail to see what that has to do with anything. Why on Earth would a USB 3.0 device have a USB 2.0 driver?

Hmmm... I think it should be 40W. I did get 45W in brief peaks but this was not sustained for more than 10-15 seconds. Perhaps it was at the start of the torture testing when the SSD was working as well to load the program into memory? Not sure. Anyway, take 40W as more representative. Neither figure can be achieved with any normal application anyway. I mean, the most demanding thing you'll probably run on this is... TMPGEnc, and that didn't get much over 20W w/ any type of file.

Not the device, the USB 3.0 controller. While the BIOS might be able to start the boot process from a device connected to a USB 3.0 controller, once the OS starts taking over it will lose access to the boot device as it has no driver for the USB 3.0 controller. Try disconnecting the boot volume from a running OS and see how well that works. That is effectively what happens in this case.

But the USB controller will have been acting as a USB 2.0 device throughout. It will be using the parallel High-Speed bus (the "USB 2.0" part of the USB 3.0 link). Superspeed is not part of the picture here.

In order to comply with USB 3.0, a port (and hence its controller) MUST be capable of operating in full accordance with the USB 2.0 spec - which obviously includes the driver. That's exactly what it would be doing here.

Got a reference for that?

Well, it's mentioned repeatedly throughout the spec (http://www.usb.org/developers/docs/usb_ ... pUniversal Serial Bus 3.0 Specification), most concisely perhaps in clause 11.3:

"USB 3.0 device certification requirements require support for USB 2.0 for all user attached devices."

This, also from the spec, makes things clearer:

"The USB 3.0 connection model accommodates backwards and forward compatibility for connecting USB 3.0 or USB 2.0 devices into a USB 3.0 bus. Similarly, USB 3.0 devices can be attached to a USB 2.0 bus. The mechanical and electrical backward/forwards compatibility for USB 3.0 is accomplished via a composite cable and associated connector assemblies that form the dual-bus architecture. USB 3.0 devices accomplish backward compatibility by including both SuperSpeed and non-SuperSpeed bus interfaces. USB 3.0 hosts also include both SuperSpeed and non-SuperSpeed bus interfaces, which are essentially parallel buses that may be active simultaneously."

USB 2.0 ports work with any backward compatible USB device w/o having to install any driver, as that's built right into the firmware/BIOS of the motherboard.

The problem with installing Windows from a USB flash drive (either USB 2 or 3) via a USB 3.0 port is that the ports don't always function until a driver is loaded for it. That was my experience with a couple of motherboards... which I realize are somewhat older ones, going back to when USB 3.0 was relatively new, which perhaps did not implement backward compatibility correctly?

So....

In the last half hour or so, I've tried my USB 2.0 Windows 7 installation flash drive on USB 3.0 ports on several newer boards... and discovered they all work. So... my original comment in the article was based on limited info; I've retracted the comment. Sorry for any confusion.

Nice to see it works in practice. I'd never tried it myself. Thanks Mike.

Yes. Yes there is.

Any chance to test the audio quality? This might be a tempting option for a media computer, but there's no optical nor analog output. How do you hook it to an amplituner?

Either HDMI or DisplayPort. All modern AV receivers have the former, but not sure whether DP is as widely implemented. From the spec sheet:
Intel® High Definition Audio (Intel HD Audio) in the following configuration:
> 8-channel (7.1) digital audio via one HDMI 1.4a output and/or via one ThunderBolt connector (DisplayPort 1.1a)
And what is Intel HD Audio? http://www.intel.com/content/www/us/en/ ... ation.html

It's a shame... AV receivers don't age as fast as computers and can be a lot pricier, especially the quality ones. Not having the HDMI port doesn't warrant a change for a new model IMO. Oh well, no NUC for me then

Looks like there's a thermal issue with this design.

I had been checking back here at SPCR wondering if anyone was excited about the NUC for the last 6 months or so... Great to see a quality review. Thanks Mike!

hopefully the OEM's can run with it and add some more functionality, or at least bring the price down a little bit.

if i could get one with gig lan, hdmi and either usb 3.0 or esata id buy one today... coax/optical audio, and maybe micro sd for additional affordable storage would just be icing.

Yeah, that's interesting. We didn't see any similar issues, but I doubt we put 5 gigs into the drive in a single go. There's 12gb altogether in the benchmarking and testing folders, but not any single super large files -- sounds like they were trying to load up a handful of big buray rips or such.

I wonder if it afflicts all 6GB/s mSATA SSDs or just Intel's?

Throw the NUC in the PSU hot box and see what happens

I see that Sapphire has announced a Trinity based tiny PC - the Edge VS. From the PR, it looks like it has some decent connectivity. Maybe one of those will find it's way to Vancouver for comparison.

Nice little unit ! thanks for the review Mike !

I'm surpised that no one's compared this to the Raspberry Pi. Sure it doesn't have anywhere close to the performance of this box, but at $35, it can run XBMC & play H.264 movies at 1080P. Max power draw is 700ma @ 5V.

http://www.raspberrypi.org/

Really? That's a fairly rosy summary of what I've read at the XBMC forums about the Pi. My reading is more like: it runs XBMC, struggles with some of the skins, and will play some H.264 movies at 1080p, but generally isn't the first in the queue for recommendation if all you want is something that just works without too much faffing about.

Besides the NUC isn't just about HTPC capabilities, it's a competent desktop machine too, likely sufficient for many/most peoples everyday use. With the Pi you'd have to run Linux, and that isn't everybody's barrel of biscuits (regardless of the merits of Linux v the other OSes).

Exactly. While I can see the appeal of a really cheap computer to tinker with, I think many are viewing the Pi through rose coloured glasses. The Pi is slow, really really, slow. Much slower than a quick glance at the specs would indicate. It is no where near the performance of any 700 mHz x86 chip. On a good day it might perform like a lower end P2. More realistically it's solidly in P1 territory performance wise.

Would a P2 with a finicky H.264 decoder make a good HTPC? Not really, because any video the decoder can't deal with is going to be a slideshow.

The Pi does have some interesting uses, but anything that might need performance it's going to fail at. The NUC might be 9 times more expensive, but it's way way more than 9 times faster. Price/performance the Pi is a bad deal.