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

 

[EdZ]

I believe the Intel drive restriction still applies to the other RAID levels too.

According to PCWorld "There’s always a catch, and with Intel’s VROC there’s actually multiple catches. The Asus Hyper M.2 card works perfectly fine with, say, a Samsung 960 Pro, and you could see four individual Samsung drives in RAID. If you want to use VROC to create a bootable partition, though, you can use only Intel SSDs"

Optane Memory has been confirmed as a feature, Optane m.2 cache drives have not been confirmed either way yet AFAIK.

And to update myself: one of Intel's slides specifically mentions "Intel Optane Memory and Intel Optane SSDs", so I'd take that as confirmation Optane cache drives will work on X299.

 

[IntoxicatedPuma]

Well...R5 1600 cost with B350 MB : 300$, skylake -X 6 cores + X299 MB : 700-800$ at best...Intel X299 is definitely interested only to have best performance without considering performance/cost ratio.

The 6 core I was referring is the Coffee Lake 115x CPU/boards later this summer, Intel's 6 core is at $389 on X299 so I assume it will be the similar price on the 115x socket. That'd put it at under $500 for a decent board and 6 core from Intel. Seeing as how Ryzen 8 cores are about on par with 3.5-4ghz 6 Core Broadwell ships, I doubt an 8 core Ryzen, much less a 6 core, is going to compete evenly with an Intel chip several generations newer and 500mhz higher than before.

However, if you have a chart that shows performance/price ratio for all of these chips, please remember to include the CPU/motherboard we already have and is paid for, so we can see that actually none of these are really a good bargain by that metric.

 

[jeshikat]

If you want to use VROC to create a bootable partition, though, you can use only Intel SSDs

Sorry, I still meant VROC but levels higher than RAID 0.

 

[BirdofPrey]

I think Intel has kind of done that with the Turbo Boost 3 - because one or two of the cores can ramp up much higher. I think from a desktop perspective big.LITTLE doesn't make a whole lot of sense though because power constraints arent a big problem.

Do remember, though, that x86 encompasses more than just desktops. Power efficiency is important for laptops and tablets to extend battery life. Also even desktops could potentially benefit from a couple extra cores freeing up the main cores. I said I think AMD seems more likely to go down that research avenue since they have always pushed for more cores than Intel, and smaller cores would be a good way to up their core count, and desktops do still tend to run a bunch of low-priority background tasks, so taking them off the main core might also help ease the issues with their lower IPC.

If Xeon Phi gain success, then we could see future chips combining some 'big' x86 cores along with a cluster of smaller more basic x86 cores. If Larrabee (predecessor to Xeon Phi, with some focus on graphics output) is resurrected then this x86 cluster could replace the GPU that takes up a large portion of the die on current Intel consumer CPUs. The advantage of a cluster of x86 cores over a GPU is that unlike with HSA and GPGPU you are using the same ISA for both the 'big' and 'small' cores, so there is the potentail to shuffle workloads in between.

It's possible. I'm actually surprised AMD hasn't pursued a similar project to Larrabee for the GPGPU space since, unlike Intel, they have decent experience in both the high performance GPU (and GPGPU) and x86 space.

It should be noted, though, that while Larrabee is x86 one of the main things of note isn't just the core count, but also the large number of SIMD pipelines in each core.

===
Anyways I am tending to agree X299 looks to be something of a wash. It doesn't look like it has any real advantages.

 

[jeshikat]

I guess Bit-Tech went around Intel and sourced some CPUs themselves so their review of the 7900X is up: https://www.bit-tech.net/hardware/2017/06/16/intel-core-i9-7900x-and-x299-chipset-revie/1

 

[MarcParis]

The 6 core I was referring is the Coffee Lake 115x CPU/boards later this summer, Intel's 6 core is at $389 on X299 so I assume it will be the similar price on the 115x socket. That'd put it at under $500 for a decent board and 6 core from Intel. Seeing as how Ryzen 8 cores are about on par with 3.5-4ghz 6 Core Broadwell ships, I doubt an 8 core Ryzen, much less a 6 core, is going to compete evenly with an Intel chip several generations newer and 500mhz higher than before.

However, if you have a chart that shows performance/price ratio for all of these chips, please remember to include the CPU/motherboard we already have and is paid for, so we can see that actually none of these are really a good bargain by that metric.

You are joining my point of view...x299 is useless as soon as intel will release their 6 cores coffee lake cpu..

 

[MarcParis]

I guess Bit-Tech went around Intel and sourced some CPUs themselves so their review of the 7900X is up: https://www.bit-tech.net/hardware/2017/06/16/intel-core-i9-7900x-and-x299-chipset-revie/1

Megalol....stock core i9 7900x consumes as much as power as 4ghz oc r7 1800x... i'm not looking at power consumption of i9 7900 oc at 4.4ghz...

Globally for a 1000$ chip vs 330$ ryzen, i9 7900 is not a must have cpu. I hope 6 and 8 cores skylake x will be better.

By the way due to its very high power consumption i wonder how it will perfom in mini itx case like dan a4...perma throttle?..

 

[EdZ]

Megalol....stock core i9 7900x consumes as much as power as 4ghz oc r7 1800x... i'm not looking at power consumption of i9 7900 oc at 4.4ghz...

R7 1800x - 4GHz boost, 8 cores, 95W TDP rated
i9 7900x - 4GHz boost (4.3 to 4.5 for lighter threaded workloads), 10 cores, 140W TDP rated
That the i9's tested system power consumption is 10W more than the R7 does not reflect too well on the R7, though it's well known that AMD's TDP figures are generally undershoots whilst Intel's are generally overshoots when it comes to testing the actual consumed power draw.

What I'd most like to see tested is VR performance vs. the 7700k, or at least general gaming testing using frame-rating rather than average framerates so an eye can be kept on maximum frametimes.

 

[MarcParis]

Well simply amd tdp and intel are different : 95w amd tdp = 128w intel tdp. (65w amd=90w intel)
http://www.hardware.fr/articles/956-10/consommation-efficacite-energetique-tdp.html

However power consumption is pretty high on this core i9 7900x, that's clearly an issue for mini itx and sff setup.

Let's see the core i7 7820x also

 

[IntoxicatedPuma]

You are joining my point of view...x299 is useless as soon as intel will release their 6 cores coffee lake cpu..

No, because I never compared Ryzen AM4 vs X299. If you want to compare X299 to Ryzen because you think it's a poor value that's fine, but if buyers only cared about value for performance, then AMD would not be in the situation it is today. If this is your reasoning, then X79 was also pointless if the i7 3820 is compared to the 2500k and FX8100. Threadripper is also a poor "value" because it's 10 core + board is going to cost way more for a system than the 8 core Ryzen for only marginally better performance.

 

[QuantumBraced]

I bet Intel sells the mainstream 6-core for more than the 7800X. The difference between the 7600K and the 7700K is about $100, I just don't see them offering a whole other 2 cores/4 threads (basically an i3) for just $50 more. They will shift pricing for the whole line, but they can't shift it too much, I doubt the 4/4 part would fall $50 or more, then the i3s have to go down to under $100 (and then the Pentiums would have to basically go away, they're i3s now anyway). The current core/thread tiers are too well-established as far as software and gaming requirements, so I just don't see them bumping down prices too much, despite pressure from AMD. If the 6-core is less than $100 more expensive than the 4/8 part, it would be amazing value and they wouldn't sell any 4/8 CPUs -- kind of the situation right now with the 7740X being only $50 less than the 7800X for dropping 2 cores/4 threads and 12 PCIe lanes, they won't sell any of those things. Plus, this is too good an opportunity to make money.

As far as the 7800X being under $400, it's the bottom tier CPU in that class (let's pretend Kaby Lake-X doesn't exist), plus you're paying more for the motherboard and there's no internal competition within the enthusiast line, so that makes sense as a way to promote Intel's 2066 line and compete with AMD on core count on the enthusiast platform front. They would get shit for it though. But that's my bet!

 

[MarcParis]

No, because I never compared Ryzen AM4 vs X299. If you want to compare X299 to Ryzen because you think it's a poor value that's fine, but if buyers only cared about value for performance, then AMD would not be in the situation it is today. If this is your reasoning, then X79 was also pointless if the i7 3820 is compared to the 2500k and FX8100. Threadripper is also a poor "value" because it's 10 core + board is going to cost way more for a system than the 8 core Ryzen for only marginally better performance.

I guess you are misunderstanding me.

For your knowledge, my last amd cpu was a 3800 x2...then I moved to intel qx6700, core i7 920, and then many core i7 mainstream...at that time Intel was proposing both value and performance...even x99 was good for me...because amd cpu were too far behind.
Now with Ryzen, cpu market is completely changed as this cpu is competitive in terms of price and performance. Due to that, Intel cannot simply follow this strategy...x299 is a dead end in terms of marketing for my point of view.

 

[Phuncz]

I guess Bit-Tech went around Intel and sourced some CPUs themselves so their review of the 7900X is up: https://www.bit-tech.net/hardware/2017/06/16/intel-core-i9-7900x-and-x299-chipset-revie/1

The part that is noteworthy for me in the conclusion is this:

So, we've established that the Core i9-7900X is potentially a great overclocker and has superb performance at stock speed too, but what about Intel versus AMD? Well, in games such as Fallout 4, there's still very little difference. The Ryzen 7 1800X was a little slower in Ashes of the Singularity and Deus Ex, even when overclocked. However, the real question is value. The Core i9-7900X is more than twice the price of the Ryzen 7 1800X, yet in Terragen 4, Cinebench, and HandBrake and in games, it's far from being twice as fast. In fact, the deficits are almost exactly what we'd expect from a two-core/four-thread deficit.

So basically no IPC gains yet again. This means ThreadRipper, with all things being equal, will face off with Skylake-X core per core and clock per clock.

Another noteworthy piece:

However, temperatures were definitely a concern with Cinebench and Terragen pushing 100°C with our 240mm AIO liquid cooler [Fractal Design Celcius 24]. As a result, while stable and potentially tameable under custom water-cooling, we decided to go for 4.6GHz for benchmarking, which required a super-low 1.22V. Interestingly our Core i7-6950X ran much cooler despite using a significantly higher voltage, albeit at 4.4GHz. This could well be due to thermal paste having been used between the heatspreader and CPU core with the new Skylake-X CPUs, in which case delidding could potentially yield significant benefits given the high heat density.

Even though the Skylake-X seem prime examples for serious overclocking with delidding, it does give the people a choice: limited overclocking on AMD for 300-500$ or extended overclocking with delidding and oversized cooling on Intel at 600-1000$.

 

[MarcParis]

Is intel serious about delidding a 1000$ + cpu?..

 

[Phuncz]

I don't understand why they changed the HEDT CPUs to "basic" TIM, the only logic seems to be maximise profit as the soldered IHS was prefered for overclocking. Purely speculation from my point of view, but maybe Intel was gearing towards limiting overclocking on HEDT as a continuation of their monopoly on that market.

 

[jØrd]

Well simply amd tdp and intel are different : 95w amd tdp = 128w intel tdp. (65w amd=90w intel)

From what I understand these numbers are more guestimates than gospel. IIRC the way AMD measure TDP is the maximum heat that can be expected to be dissipated by the heatspreader in a set amount of time (something small like a microsecond or whatever but I forget the exact time) whereas Intel's TDP figure is the maximum amount of heat you can expect to be pulling off the heatspredder in totality. I could have this a little off, its been a while since I looked into it, but more-or-less this is how the two are getting their numbers. There isn't really any maths you can do to convert one teams TDP into another (well there likely is maths you can do but it would probably require some pretty expensive and accurate test gear in a temperature & humidity controlled environment so you can perform multiple measurements using the "AMD method" on Intel chips and vice versa and probably quite a large pool of sample chips from both vendors to get good averages and such to then base that conversion maths on).

 

[MarcParis]

http://www.hardware.fr/articles/956-10/consommation-efficacite-energetique-tdp.html
As I've linked several times, AMD is not counting TDP like Intel.
You are right AMD TDP is maximum of heat to be dissipated, and Intel TDP is max consumption.

Based on their measures, R7 1800X is using 128W of power so 128W Intel TDP, whereas you just have 95W heat to dissipate..

 

[jØrd]

Based on their measures, R7 1800X is using 128W of power so 128W Intel TDP, whereas you just have 95W heat to dissipate..

For that given chip under that given workload on that given day. My French is super rusty (the only French I can remember is how to order beer & I can't read or write in French at all) so if there is more to it than that then I may well be mistaken but a single chip and a single test / single test set from a single source isn't enough for me to consider that number as gospel and neither should anyone else imo. The only way to get a number that could be considered authoritative and applicable to any given chip would be to test a number of chips in a controlled environment a number of times under a number of workloads and then working backwards from those results and publishing the actual mathematical equation you get from that work so others can replicate your testing to confirm your maths are accurate. All that being said though my maths is only marginally better than my French so its entirely possible that I am missing something obvious here.

 

[QuantumBraced]

Wow, I didn't make much of the lack of soldered-on IHS, but now I see just how big an impact it makes -- from the article, they were struggling with thermals @ 4.7GHz, 1.28 VCore with a 240mm AIO whereas der8auer got 5GHz stable at 1.32 VCore with a 280mm AIO. Usually, Intel chips are not limited so much by thermals as they are by VCore, so even a moderately good heatsink could handle a 140W chip (if you were okay with more noise). Now, because they skimped out on the solder, we're in a situation where we can't take advantage of the full potential of the chip, unless we compensate for poor thermal transfer below the IHS with some insane 360mm AIO or a fully custom loop.

I mean 1.32 VCore is pretty insane anyway for this chip, but still, Intel essentially negated 200MHz of performance on $1000+ chips, because they were too cheap to install the IHS properly. Seriously, why aren't all of their chips soldered, is it really THAT much more expensive?

 

[Kmpkt]

My theory is that the motivation behind the TIM swap on the i9 chips is that it forces overclockers to void their warranty by de-lidding or alternatively forces them to overclock much less aggressively. Since it is really easy for Intel to prove you've de-lidded your CPU, it also becomes really easy to deny what is likely a significant number of warranty claims from customers who have damaged their chip overclocking. Less RMA = more $$$ - sounds like something Intel would do!