News ASRock Unveils the X299E-ITX/ac: Mini ITX + X299 + Quad-channel Memory

Mod edit:

Detailed overview of what we know about the X299E-ITX/ac thus far here: https://smallformfactor.net/news/asrock-x299e-itxac-little-monster-detailed

Original:

ASRock did it! Finally, there's an Intel HEDT platform motherboard with full quad-channel DDR4 memory. The new X299E-ITX/ac is for those who need up to 18 CPU cores and up to 64 GB of quad-channel DDR4 memory in their SFF machines for reasons. The board manages its limited PCB real-estate by going vertical. It features two riser cards, one with a few onboard controllers, and a pair of 32 Gb/s M.2 slots), and the other riser with SATA 6 Gb/s ports, a third M.2 slot, and the headers such as USB 3.1. The board draws power from 24-pin ATX and 8-pin EPS connectors, conditioning it for the LGA2066 CPU using a 7-phase VRM. The lone expansion slot is a PCI-Express 3.0 x16, memory is handled by four DDR4 SO-DIMM slots. Connectivity includes two Intel I219-V driven gigabit Ethernet interfaces, 802.11ac WLAN, and Bluetooth 4.1.

Source

 

[jeshikat]

The smallest I could feasibly see with an external brick, single-slot LP GPU, a CPU heatsink like the L9x65, and no case fans is 85mm x 180mm x 190mm = 2.9L

 

[EdZ]

An interesting bit buried in Anandtech's analysis of Skylake-X and -SP:

[...]that means that the 12-core SKU, the Core i9-7920X, is likely derived from the HCC 18-core silicon. Which also explains why that CPU has been delayed until August.

Basically, there is no MCC silicon. The larger Skylake-X chips coming later this year are actually binned down HCC (18 core) dies.

 

[jeshikat]

Wow, if you'd told me a year ago that Intel would be binning 18-core Xeon dies in a desperate attempt to compete with AMD in the HEDT market I would have laughed.

 

[EdZ]

Intel have been binning high core count dies for decades, that's how the Xeons (and the i3/i5/i7s) are differentiated. Previously the LCC/MCC/HCC/XCC definitions were a bit arbitrary and shifted with each generation, but with the switch to a mesh fabric from the ring bus topology they are mostly constrained to 'set' X*Y core counts (integer counts minus interface blocks, aspect ratio can't be too high or inter-block IO is constrained). The current lineup of LCC/HCC/XCC could just be called MCC/HCC/XCC or LCC/MCC/HCC and the situation would be the same.

 

[Phryq]

The smallest I could feasibly see with an external brick, single-slot LP GPU, a CPU heatsink like the L9x65, and no case fans is 85mm x 180mm x 190mm = 2.9L

Right, but this is only rated for 84W (though maybe that's enough, depending on the chip).

A dark Rock TF would make the case 150mm x 180 x 190, 5.13L going by your specs, and it could cool anything, though maybe it's overkill for my needs.

 

[jeshikat]

I've used the L9x65 on a 140W TDP i7-5930K. With a 25mm thick fan it did pretty well too, though a bit noisy at full load.

 

[Phryq]

Hmmm, watching this vid now,


Shadow rock LP seems good.

 

[darksidecookie]

I'd really like to see clocks on that. Having used a 14-core Xeon for the last few months that never seems to go above 2.5GHz despite supposedly having 3.0GHz max turbo, it's kinda rough unless ALL your workloads are heavily multi-threaded.

I also recently bought a 14 core xeon (e5-2683 v3) for simulations and render purposes and noticed the same thing. So after som digging, i stumbled upon this formum thread. where was explaned that intel nerfed these xeons to only turbo up to 3 ghz on one core, but some overclocker figured out a way to override the micro code and get them to all turbo boost to 3ghtz.

 

[jeshikat]

Neat, I'll have to check into that!

 

[EdZ]

i stumbled upon this formum thread. where was explaned that intel nerfed these xeons to only turbo up to 3 ghz on one core, but some overclocker figured out a way to override the micro code and get them to all turbo boost to 3ghtz.

That's the intended behaviour for Turbo Boost on all CPUs: when one core is loaded, Turbo will hit the maximum frequency, when a small handful of cores are loaded there is a lower Turbo frequency they will hit (this was implemented on later CPUs, very early Turbo Boost implementations only had the single-core turbo), and when many cores are loaded the Turbo frequency will drop even further. AMD use the exact same method on their chips too, using the brand name Turbo Core instead.
The exploit used in that thread is to remove the clockdown for AVX execution. If anyone remembers the issue with certain Prime95 versions that could heavily load AVX and Haswell where the clock-down functionality was missing and CPUs could fry due to overtemp/overcurrent, that's what the AVX clockdown on the larger Xeons is intended to prevent. This failure can occur regardless of cooling because the failure mode is blowing power traces rather than electron tunnelling failures in the silicon. It's like a more extreme version of laptop SoCs where using the GPU at higher clocks causes the CPU to clock down, but instead of just being two areas of the chip a few mm apart drawing power, the components are mixed in with each other and sharing power lanes, so the current draw is local and concentrated.

If you have a Xeon E3-xxxx or a Xeon E5-1xxx (both single-socket) then exploiting the bug is unnecessary and you can just use regular old overclocking to do the same thing without the penalty of running on a weird mix of old microcode versions and courting isntability. It's only on Xeon E5-2xxx or higher (dual/quad/octa-socket) where things are really locked down due to inter-chip QPI link stability.

 

[QuantumBraced]

I was reading this review of the 7820X, which I imagine will be the most popular or second most popular X299 CPU.

https://www.overclock3d.net/reviews/cpu_mainboard/intel_core_i7_7820x_skylake_x_review/3

And it can do 5GHz at 1.3v. That's incredible. That is massive improvement over Broadwell-E. The problem is it overheats like crazy even though it's stable. Because of the TIM obviously. I want to find a review where it was tested delidded... I think it's safe to recommend anyone considering Skylake-X pay $50 to SL for a warranty-backed delidding.

 

[tjh1989]

I was reading this review of the 7820X, which I imagine will be the most popular or second most popular X299 CPU.

https://www.overclock3d.net/reviews/cpu_mainboard/intel_core_i7_7820x_skylake_x_review/3

And it can do 5GHz at 1.3v. That's incredible. That is massive improvement over Broadwell-E. The problem is it overheats like crazy even though it's stable. Because of the TIM obviously. I want to find a review where it was tested delidded... I think it's safe to recommend anyone considering Skylake-X pay $50 to SL for a warranty-backed delidding.

Yep that's exactly what I want to know. What kind of water-cooling do you think a 7820x will require?

 

[jeshikat]

Looks like people were cooling the FX-9370 and FX-9590 space heaters with the Hyper 212 EVO. So maybe not with the stock TIM, but with delidding it should be reasonable to get away with something much more compact than a 360mm rad for these Skylake-X chips.

 

[Boil]

I was reading this review of the 7820X, which I imagine will be the most popular or second most popular X299 CPU.

https://www.overclock3d.net/reviews/cpu_mainboard/intel_core_i7_7820x_skylake_x_review/3

And it can do 5GHz at 1.3v. That's incredible. That is massive improvement over Broadwell-E. The problem is it overheats like crazy even though it's stable. Because of the TIM obviously. I want to find a review where it was tested delidded... I think it's safe to recommend anyone considering Skylake-X pay $50 to SL for a warranty-backed delidding.

Looks like people were cooling the FX-9370 and FX-9590 space heaters with the Hyper 212 EVO. So maybe not with the stock TIM, but with delidding it should be reasonable to get away with something much more compact than a 360mm rad for these Skylake-X chips.

From the Silicon Lottery website, regarding de-lidding...

"Not compatible with Intel Skylake-X CPUs, including the 7800X, 7820X, and 7900X."

 

[Davila]

From the Silicon Lottery website, regarding de-lidding...

"Not compatible with Intel Skylake-X CPUs, including the 7800X, 7820X, and 7900X."

You CAN delid Skylake-X CPU's. In fact Silicon Lottery does sell these CPU's delidded. I'd imagine that they don't allow you to send in those processors for a delid due to safety issues. If one thing goes wrong for them it could cost them $600. On the other hand, if they buy the processors themselves and delid them then they are working in safer conditions. They wouldn't want to ruin your CPU.

 

[Boil]

You CAN delid Skylake-X CPU's. In fact Silicon Lottery does sell these CPU's delidded. I'd imagine that they don't allow you to send in those processors for a delid due to safety issues. If one thing goes wrong for them it could cost them $600. On the other hand, if they buy the processors themselves and delid them then they are working in safer conditions. They wouldn't want to ruin your CPU.

You are, indeed, correct; Silicon Lottery DOES sell delidded Skylake-X CPUs, I totally missed that on the product page...!

 

[chx]

I wonder what a Scythe Fuma would do with these.

 

[jØrd]

Cool them down?

sorry, couldn't resist

 

[chx]

Yeah but would it be adequate at 1.3V instead of watercooling.

 

[3lfk1ng]

After researching a lot of compact cooling possibilities, I honestly think that the Ncase M1, with its support for dual 240mm radiators might be the perfect SFF case for the 10+ core ITX builds on the 2066 platform. Sure, it would require a custom loop but 2x240mm radiators shouldn't have a problem keeping a top of line i9-7900x (or higher) in check with some additional headroom for overclocking.

@chx Yes! The Scythe Fuma is awesome. At just 149mm high, it trades blows with the much larger 160/165mm HSFs and 240MM AIO solutions (source). Outside of the stock sleeve bearing fans, my only concern is ram clearance, so provided you have enough headroom to accommodate for the fan being slightly offset, I think that the Fuma will score quite well among the SFF community.