ASRock X99E-ITX/ac LGA 2011-v3 Motherboard Review @ [H]

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
55,634
ASRock X99E-ITX/ac LGA 2011-v3 Motherboard Review - There is a small subset of computer enthusiasts that have been asking for an X99 option in the mini-ITX form factor. Ask and you shall receive, right? ASRock steps up to the plate to provide small form factor enthusiasts with just such a motherboard. We evaluate the X99E-ITX/ac and experience the good, the bad, and the ugly...if any.
 
Now if Intel would get off of their deceased posteriors and release the E5-1691 V3 to the channel, one could build an absolute beast of a system that's about the size of a shoebox...;)
 
I don't understand... I thought a good portion of the performance over the non 2011 i7's was due to the quad channel memory? So they cut it down due to space to dual channel. Due to bandwidth lost, wouldn't it have been wiser to put DDR3 in there instead of DDR4? The 2011 memory controller still can use DDR3, right?

So they built a pretty nice motherboard, and then neutered the platform. I'll save my money.
 
thanks for the review. I was interested in what this little board could do
 
if anyone was wondering about an actual use for these, i've used two at this point to cram 5820K's into extremely portable setups for single camera 1080p video streaming (livestream studio). they work pretty well, no problems so far. i didn't need bells/whistles/quad channel memory, just lots of CPU in a small package.
 
Now if Intel would get off of their deceased posteriors and release the E5-1691 V3 to the channel, one could build an absolute beast of a system that's about the size of a shoebox...;)
You could use an E5-2699 V3 instead? Nets you 18 cores instead of 14.
I don't understand... I thought a good portion of the performance over the non 2011 i7's was due to the quad channel memory? So they cut it down due to space to dual channel. Due to bandwidth lost, wouldn't it have been wiser to put DDR3 in there instead of DDR4? The 2011 memory controller still can use DDR3, right?

So they built a pretty nice motherboard, and then neutered the platform. I'll save my money.
The extra memory channels don't make much of a difference outside of a few things. Why wouldn't you want DDR4 over DDR3 to compensate anyway? It's a DDR4 only affair anyway. 2011-3 technically does support DDR3 on a few CPUs, but none of those are publicly available and neither are the custom made boards for them.

Essentially this is the most powerful ITX system you can build. You can get an 18 core CPU with 64GB of RAM (Kyle presumably doesn't mention this because it requires a Xeon CPU instead of one from the Core i7 line), one of the Intel NVMe SSDs, and a high end GPU. It's what I've done (albeit using a Xeon with fewer cores due to cost).
 
I had been looking at this board a month ago to potentially upgrade my mini-itx system so I'm really happy that you reviewed it. All the same, unless ASRock makes a Rev 2 of the board with at least one significant change I'm skipping it.

The motherboard is already tiny so why was the decision made to go with full sized dimms when DDR4 SODIMMS are starting to come to market?

http://www.corsair.com/en-us/corsai...imm-2133mhz-c15-memory-kit-cmso8gx4m2a2133c15

ASRock could have finagled four SODIMM sockets in the space that two DIMM sockets takes up. I could potentially see the reason being that there wasn't enough real estate to run all the traces for the memory channels, but that could have been worked around by using a PCB with more layers. Maybe I'm oversimplifying it; I'm not an electrical engineer. All I know is that the board (and the platform) isn't cheap. I'd have gladly paid a few bucks more for the memory and the board to get quad channel memory support on this board. :(
 
Good luck overclocking that on an ITX board. There's a reason the variants with more cores are lower clocked. The die is massive and puts out an equally massive amount of heat. The 14 core and up CPUs have a die that's more than double the size of the 8 core ones.
 
Good luck overclocking that on an ITX board. There's a reason the variants with more cores are lower clocked. The die is massive and puts out an equally massive amount of heat. The 14 core and up CPUs have a die that's more than double the size of the 8 core ones.

I don't think its even POSSIBLE to OC the xeon chips.

Do the Haswell E Xeons allow strap overclocking?

I know Linus got a 12% OC on an Ivy-E 12 core, but that was without strapping.
 
It is. You can only do it with the single socket versions though (16xx). Some ES CPUs are unlocked as well.
 
Some ES CPUs are unlocked as well.

I've yet to see any examples of an unlocked high core count (HCC) chip anywhere or any CPU-Z shots showing this. AFAIK, the only unlocked versions are test chips that are internal to Intel (are never lent out).

If there were such a chip in the wild, I'd pay a very pretty penny for it...;)
 
Kyle/Daniel: You guys must not do very much work at all with server hardware. The CPU socket mounts aren't proprietary at all, it's Intel's narrow ILM system used in servers and workstations with width constraints. I did however get a kick out of how you guys ended up mounting your kits.
 
I totally see this as a challenge and your next contest here. "Overclock the Beast" in any matter you can attain. Like to see what a jimmy rigged AIO would do on this board?
 
Last edited:
I totally see this as a challenge and your next contest here. "Overclock the Beast" in any matter you can attain. Like to see what a jimmy rigged AIO would do on this board?

Why "jimmy rig" anything? Asetek has a narrow ILM mounting kit that adapts any of their recent coolers to the narrow ILM form factor. Very cheap too...probably cheaper than buying the parts to "jimmy rig" it would be...;)

http://www.ebay.com/itm/Asetek-Liqu...ultDomain_0&hash=item3f4cf663dc#ht_101wt_1124
 
As an eBay Associate, HardForum may earn from qualifying purchases.
You could use an E5-2699 V3 instead? Nets you 18 cores instead of 14.

Wait, are you saying the mobo works with an E5-2699 V3?

If so, I will have to get one to try it out.

Hell, maybe even 2;-)
 
Last edited:
Kyle/Daniel: You guys must not do very much work at all with server hardware. The CPU socket mounts aren't proprietary at all, it's Intel's narrow ILM system used in servers and workstations with width constraints. I did however get a kick out of how you guys ended up mounting your kits.

I was not going to go out and purchase hardware to do one mount one time, when I could easily do it myself with what I had on the bench.

And no, I do no server hardware work any more. Once our server hardware reaches EOL, we will be fully virtualized.
 
It seems like a waste to build an x99 system with all those PCIe lanes and quad channel RAM, only to be limited to one PCIe slot and two RAM slots :p

I mean, at least to me, the extra RAM channels and PCIe lanes are the only real reason to go with Haswell-E over a regular LGA 1150 haswell.

The extra two cores? Take 'em or leave 'em.
 
I agree Zarathustra[H], that's why there's only one of these boards being made.

I think microATX is a much better platform for small x99 builds. Room for 4 memory slots and 3 x16 slots is the bare minimum for a good time!

For people who want to have overclocked miniITX builds, z97 is still the go-to.
 
Zarathustra[H];1041734741 said:
The extra two cores? Take 'em or leave 'em.

You mean extra 8 cores on i7, and 14 extra cores on Xeon right? :p
 
Subsystem Testing

NOTE: For all Subsystem Testing, an Intel Core i7 5960X (3.0GHz / 3.5GHz Turbo) and 4x 4GB (16GB total) Corsair Vengeance LPX (2800MHz DDR4 16-18-18-36, [email protected]) memory modules running at DDR4 2133MHz speeds (stock testing, up to 2800MHz overclocked) were used. The CPU was cooled with a Koolance Exos 2.5 and CPU-370 water block.

How could you use 4x 4GB (16GB total) ram when it only has 2 slots?
 
Back
Top