I always love to see interesting deviations to the norm when it comes to motherboards, and something such as a mini-ITX based extreme system has been at the top of my list for many years. We never saw a mini-ITX X79 system (the nearest was an extended mini-ITX from Shuttle) but now ASRock has gone head first into the X99 plus mini-ITX arena, showing their first model at CeBIT later this month.

The reason for mini-ITX on the extreme platform is usually for density, though there are a couple of compromises that have to be made. The socket is large, and supporting quad channel memory can be a challenge with SATA ports and 40 PCIe lanes in tow. As a result, ASRock’s X99E-ITX/ac only uses dual channel memory, and we get a single PCIe 3.0 x16 slot for add-in cards.

There is bundled dual-stream 802.11ac wifi, along with dual Intel network controllers and SATA Express. USB 3.1 is also supported through two Type-A ports, presumably using the ASMedia controller we previously tested on other motherboards. The box also mentions Ultra M.2, which means PCIe 3.0 x4 lanes for an M.2 slot and looking at the board it seems to be located between the socket and the SATA Express ports. With all those PCIe lanes to spare, it makes sense to use them in this fashion.

In order to save space, ASRock has used the narrow version of the LGA2011-3 socket (many thanks to liu_d for the spot), which we saw in the our MD60-SC0 review. This narrow socket is incompatible with regular LGA2011-3 coolers, and the number of narrow-ILM CPU coolers on the market is usually limited to servers or OEMs. It would also seem that ASRock is bundling a CPU cooler with the board in order to ensure this is not an issue for the user – this looks like a 2U server cooler, but should be sufficient for 140W CPUs as long as no serious overclocking takes place. These coolers can be loud, but ASRock’s software package comes with fan controller tools both in the BIOS and in software.

Pricing and release dates are not yet announced, but we will get one in for review as soon as we can. The dual channel memory restriction hopefully does not become too severe for performance, but we will run a full range of real world tests to confirm this.

Source: ASRock

Comments Locked


View All Comments

  • Hairs_ - Friday, March 13, 2015 - link

    My comment was as much (actually more) about what boards Anandtech chooses to review, as what Asrock decides to produce. If asrock choose to churn out a million boards in an effort to differentiate themselves, then that's their business model, people will find something relevant in it.

    The last review was for a storage-behemoth oriented board where Anandtech's reviewer *decided not to bother* testing the storage. "Here's an unusual board, if you want to know if it's any use, just make some guesses based on something we reviewed years ago." That is frankly unforgivable, and the antithesis of what Anandtech has stood for since it was founded. No short cuts, no laziness, no leaving the reader to make up guesses about the product being reviewed, that's what made Anand stand out from the crowd. It's a disgrace.

    This board has clearly been chosen for review because of novelty value rather than because there is a viable market for it. Hopefully the reviewer, *this time*, will actually *test* the board, and not just play around with their imagination.

    Meanwhile the vast majority of the boards on the market go untested, because Anandtech has decided that anything under $400 isn't sexy enough to bother looking at.

    An attitude which would make a lot more sense if the boards they did choose actually got tested properly. Clearly there isn't a motherboard review section anymore, there's a "what motherboards would Ian like, which he isn't prepared to buy himself" section.

  • Ian Cutress - Friday, March 13, 2015 - link

    As mentioned in that review, for the previous X79 Extreme11 testing we were able to source drives on a temporary basis to test but not for the X99 version. The underlying chip and implementation is the same, but the socket/chipset support was updated for the next platform which is what we tested. While I would love to sit on a mountain of hardware and test every nook and cranny of every product, it's a delicate balance between time and effort. For example, the 10GBase-T board we reviewed earlier went through appropriate testing because it was the first time we had seen that implementation and we had the capability to test. When new versions of similar products came to market, Anand didn't retest the same implementation in his reviews unless necessary or if he had the time, and you are holding me to higher standards.

    Over my tenure since 2011, I have reviewed ~140 motherboards for AnandTech in both the consumer and server space. At least ~115 of those (from my brief look through) have been under $400, going as low as $40 and spending plenty of time in the sub-$200 segment, especially when mainstream chipsets are launched and over the next several months when they are popular. When there is a relatively new platform catering to the high-end desktop market, such as X99, you are going to see a wave of high-end products being reviewed due to the time of launch and the relative interest of such a platform. Even then, our first four reviews for X99 were set at $230, $310, $324 and $400 and we did a substantial deep dive into the new chipset. Motherboard manufacturers want us to test their high-end equipment because for them it promotes the brand, especially if it gets a good review. We don't go overboard on testing every high-end product from every manufacturer but we take a good stab at the range, especially given that each product takes a good 35-50 working hours to test and write. It's not a testing process you can leave on overnight, either.

    As for what motherboards actually get into my hands to test, it usually comes as a compromise. I ask what they would like me to test, but I also what are their best sellers and more interesting parts. More often than not, what I ask for isn't available, or what they want me to test makes no sense from a coverage perspective (there's no point testing MPower, MPower Max and an XPower as the only three MSI boards in a launch, for example).

    If you have direct requests for specific reviews, specific tests within those reviews, or comments to make on my coverage, I'm always open to email and very easy to talk to. A number of our readers email me already, some on a frequent basis, and where possible I have been accommodating given the schedule I have. If you have specific requests, shoot me a email.
  • extide - Friday, March 13, 2015 - link

    Ian is right, and the LSI 3008 controller is a known quantity at this point, it's nothing new. I think his reasoning was perfectly sound.
  • colsanders - Friday, March 13, 2015 - link

    That troll Hairs shouldn't even get a reply.
  • TomWomack - Saturday, March 14, 2015 - link

    What is the point of working for Anandtech rather than out of your back garden, if you can't buy eighteen 120G SSDs on expenses for a test of a machine with eighteen SATA ports?
  • romrunning - Monday, March 16, 2015 - link

    I"ll repeat this comment: As someone who has built servers that have to be be portable since they are regular used at different locations, I can assure you I like having mini-ITX boards with the power of Xeons. I run virtualized servers, and I now have two host servers where previously I had one large server tower. I've saved myself room & still get the results I want.

    So just because you can't think of the usage scenario, it doesn't mean that no one has a use-case for a nice small Xeon server board (not Atom!)
  • Samus - Tuesday, March 17, 2015 - link

    liquid cooling solves some of those problems.
  • bobbozzo - Friday, March 13, 2015 - link

    The dual Intel NICs onboard would be attractive for a firewall system, but I don't think the price of the board or the CPUs is going to be attractive.
  • wintermute000 - Sunday, March 15, 2015 - link

    You can probably get away with an Avoton for a FW/router unless you're trying to inspect something like >100M @ layer 7 (UTM) in real time at which point your budget should mean you're looking at non-homebrew solutions....

    For traditional routing/stateful FW, anything above a low end haswell is complete overkill
  • madgabz - Saturday, March 14, 2015 - link

    Am I really the only one here thinking this would be a perfect base motherboard for a small, portable highend workstation to bring for some serious Oculus Rift / DataGlove / 3D-mouse showrooming? Think CAD/CAE, or architects needing some oomph for displaying latest projects or even have a small proper workstation that doesnt destroy your ears or sight? (pretty, discreet and silent cases are a no-show in PC world). Plus this could be done much cheaper than buying a Mac Pro workstation.

Log in

Don't have an account? Sign up now