Avoton & Xeon E3, move away. Xeon D is here.

[jgreco]

If they are getting this kind of performance out of Broadwell D will that likely scale up to Skylake?

I'm still running my veeery long in tooth overclocked C2D E8400 and if this is a sign of the performance to come in Skylake I'll hold out for that.

It has gotten difficult to predict Intel's improvements because they have been willing to compromise some of their traditional tiering in order to combat the ARM datacenter boogeyman; the evolution of Atom from a low speed low power low memory solution into the E3-challenging 64GB Avoton monster is a great example. In the long run they need that to still work out for them so I am guessing that - like the Xeon D - the next year or two will be transformative.

 

[Zserver]

Has anyone bought one of these babies yet? Googling reveals a pair of the Supermicro mobos at Acme Microsystems (?!)
X10SDV-TLN4F is $893
X10SDV-F is $820

The picture for the TLN4F version is the same one as the -F version, but I think it is supposed to come with a CPU fan, tho. Not going to need 10G ethernet so the -F will be enough for me.

 

[Ericloewe]

Has anyone bought one of these babies yet? Googling reveals a pair of the Supermicro mobos at Acme Microsystems (?!)
X10SDV-TLN4F is $893
X10SDV-F is $820

The picture for the TLN4F version is the same one as the -F version, but I think it is supposed to come with a CPU fan, tho. Not going to need 10G ethernet so the -F will be enough for me.

ACME? Do I get a free anvil or a free Road Runner-seeking Rocket*?

*May explode prior to launch or upon launch. May track coyotes instead.

The X10SDV-F sounds like a crummy deal. Xeon E5 is probably cheaper. The X10SDV-TLN4F sounds better, since it has dual 10GbE ports.

 

[Zserver]

Oops. The X10SDV-F page says "preorder" in teeny tiny print. However, the X10SDV-TLN4F doesn't say that so naturally I must assume it is in stock and ready to ship. ACME all the way!

 

[marbus90]

ixsystems will get one of these X10SDV-TLN4F for the project codenamed "FreeNAS mini on steroids" :)

 

[jgreco]

Sooner or later, iX is going to be running into a problem where the lower end of their target market is firmly crossing into FreeNAS-level territory. Up until the last year, it's been unusual to see people contemplating E5 based platforms, or large scale deployments, for example. The Xeon D stuff is just going to make the problem more interesting.

 

[marbus90]

There's still a model number left below the TrueNAS Z20. Also the Z20 only uses 64GB RAM, so that might be a replacement candidate already.

 

[Bastian Büttner]

For those who haven't seen it yet:
Last week, ServeTheHome posted a review of the SuperMicro X10SDV-TLN4F based on a Xeon D-1540.
128 GB RDIMM DDR4 verified working flawlessly.

In his forum post, the administrator of STH states on the availability of Xeon D platform:

Also - early May we will see the X10SDV-F and early June we will see the X10SDV-TLN4F out in general availability.

In another forum there are some benchmarks and power consumption results for the Xeon D-1540.

 

[jgreco]

That power consumption is firmly in-or-below the range we've been seeing for Sandy and later platforms. It's amazing.

 

[Zserver]

For those who haven't seen it yet:
Last week, ServeTheHome posted a review of the SuperMicro X10SDV-TLN4F based on a Xeon D-1540.
128 GB RDIMM DDR4 verified working flawlessly.

In his forum post, the administrator of STH states on the availability of Xeon D platform:


In another forum there are some benchmarks and power consumption results for the Xeon D-1540.

Thanks for posting those links. I will check the STH blog rather than repeatedly asking here and risking the wrath of the grinch.

 

[jgreco]

Wait, what am I expected to be doing wrathy here? I'm at least as interested in this as anyone.

 

[Lix]

I will have the 10Gb board in my hands this weekend. I can load FreeNAS on it and report back if it works out of the box.

 

[marbus90]

That board has arrived at ixsystems, jpaetzel is just waiting for the chassis to finally test it.

 

[Lix]

It installs and runs 9.3. No divers for the 10gb nics, as expected.

 

[jgreco]

Well that's a bit of a bummer. I'm guessing that shouldn't be a long term problem though, as Intel's usually pretty good about driver support.

 

[cynod]

Reviving this thread - there's a full review of the Supermicro BEAST on Tom's Hardware. Wow.. in 45W that is crazy. Plus full virtualization, dual 10Gbps and tonnes of RAM and storage support.

http://www.anandtech.com/show/9185/intel-xeon-d-review-performance-per-watt-server-soc-champion

 

[jb510]

I read that Anandtech post and had similar thoughts. Xeon D looks awesome for a NAS that also does transcoding or virtualization... The Atom 2750 seems sufficient for my current wants (low power, decent performance, 1080p transcoding one stream), but who wouldn't want a little more performance for free?

That board only having 6 SATA connect bugs me, I'm dreaming of a board made for NAS like the ASrock C2750D4I but with a Xeon D on it. Also I don't follow it, but maybe a higher performance Atom with USB3 is in the works as well. I think a new NAS is still a few months away, but this should be interesting to watch.

 

[Ericloewe]

There's said to be a storage-oriented Xeon-D SKU in the works, but demand is far outstripping supply for now as it is.

 

[Zserver]

Actually, I'd be happy with a board like the ASrock with just 2 SATA ports on it for a DOM and a small SSD. Four RAM slots and my M1015 and I'd be good.

The prices just aren't budging downwards.

 

[Ericloewe]

I expect ASRock or Supermicro to make a Xeon-D board sized in the [miniITX ; microATX] range with an onboard LSI SAS 3008 controller, once demand from ludicrous-size customers cools down a bit.

Actually, I'd be happy with a board like the ASrock with just 2 SATA ports

Since the SoC has 6 SATA ports, might as well expose two as DOM SATA ports and the other four in a SFF-8087 port (or whatever the SAS3 connector is called - did the name change to U.2?). It's what they did in their voodoo-shrunk X10SL7-F.