Dual CPU motherboard - Any alternative to ASUS?

[Ricky Johnson]

I was wondering if there were any decent alternatives to ASUS motherboards if you're looking at a dual Xeon workstation. Has anyone built anything with an alternative?
Seems a bit thin on the ground for options that I can find.

The reason for asking is that I've recently built a workstation using an Asus z10pe-d8 ws motherboard. It seemed like a good specification and also the only viable option at the time. Anyway, it turns out one of the DIMM slots is faulty (it's the board, not the RAM, I've tested). Along with some faulty temperature sensors on the board itself, which bothers me less so than not having all DIMM slots available.
I've got a working machine for now with 7 memory modules so I'll probably just go with it for the time being but at some point in the future I'd like to get everything working on a new motherboard so it's all functional. I'd prefer not to give any more business to ASUS in the process but I can't seem to find anything else.

As a side note, I'm not just complaining about ASUS after this one fault. I've only ever built 2 workstations and bought a defective ASUS motherboard for the first one as well (refused to boot and, after process of elimination, moved everything to a Gigabyte motherboard where it worked perfectly. RMA'd the ASUS board only to get the same faulty board returned to me 6 weeks later, rattling around in it's box, without explanation). Nothing statistically relevant to anyone other than me but I just seem to have a bad record personally with ASUS so far. It's put me off.

There's a Gigabyte board that seems like it might be okay:
Gigabyte Dual Intel Haswell Extreme MD60-SC0 Server Motherboard - Maybe I'm looking at this wrong but it appears at a glance that a long graphics card might interfere with the cooler on the CPU behind it.
There seem to be quite a few Gigabyte 'b2b' dual CPU boards as well, but none for the v3 xeon cpu's.

[Ricky Johnson]

Sorry, I'm an idiot - I've just noticed there are a ton of Gigabyte workstation boards available for the v3 xeon cpu's on the 'b2b' section of their site.
I don't know why I couldn't find those before but I'm hopeful there will be something there that fits the bill.

Still, would be interested if anyone has any particular recommendations.

[Juraj]

The Asus WS boards are actually pretty uncommon, outlier parts.

Most Xeon builds are done using Supermicro boards, which are highly configurable ( there are almost 20 models per each family ).

I would advice against it though, as the Asus one is far better suited for workstation compared to them (the Supermicro can feel pretty barebone, they're not strictly meant to be used in BigTower although they have some models oriented for that as well).
With Asus WS you also get all the possible features you could ever wish for, while in Supermicro you'll have to choose what is it that you need. You could get that choice wrong, and you won't know until future.

Asus has far more user-friendly bios (uefi) compared to them as well, one has "pro"-consumer in mind, the other tech admin. The support is easier, and more common to be found across community on internet.
While it's mostly hypothetical, the compatibility with consumer-grade parts (like GTX GPU range,etc..) is granted compared to server-range like Supermicro, also bios updates when new such parts become available on market will be swifter.

Regarding Gigabyte, good brand, but they're ranked slightly bellow Asus and are less common-choice for true high-end builds. Asus does reign supreme across whole spectrum of parts. Doesn't matter much though in the end, both will do job perfectly.

There is no track-record that Asus boards would be fault-prone, quite opposite, hence their popularity with enthousiasts,pro-consumers,over-clockers,etc.. which comes at slight price margin above their competition.

I understand shitty feeling when these parts go awry, but I would simply return them, and order another same one. Possibly you'll get newer iteration.

[Brock_Lafond]

I've been using an EVGA Classified SR-2 for the past 3 years now and love it.

If you can get your hands on one, I don't think you'll regret it.

Dual socket LGA 1366
12 Slot, triple channel DDR 3 RAM - up to 48GB; unbuffered or registered ecc
8 SATA ports
7 PCIe 2 slots; 4 at 16x or 6 at 8x and 1 at 16x
2 Duplex gigabit ethernet ports
Overclocking possible through BIOS

[Juraj]

I've been using an EVGA Classified SR-2

SR-2 is 5 years old board for LG1366, 4 generations behind what the OP is asking for (LG 2011-3)

After SR-2, EVGA tried once more with SR-X for first-gen 2011, the board failed as it was built without knowing Intel's final specs (EVGA didn't even know Xeons will be locked). It was basically gaming board for Xeons, not quite what anyone asked for.

So with that effort, they gave up on this and returned back to consumer boards. This is discontinued range.

[Ricky Johnson]

There is no track-record that Asus boards would be fault-prone, quite opposite, hence their popularity with enthousiasts,pro-consumers,over-clockers,etc..

No, I agree. I have no basis to make that accusation and didn't mean to really, it's just that my personal track record with them has been unfortunate!
I drifted into complaining but the initial reason for starting the thread was more that I was curious about the lack of options I could see with these types of boards.
I'd not found those Gigabyte options at the time.

I did see some of the Supermicro boards but they all seemed a bit skeletal, as you say.





 

[Juraj]

It's pretty niche market after all :- )

[Jann]

It really just happens with any hardware and vendor. The Asus Z9PE-D8 WS I had for a while was fine, but a friend who bought one before me, had to RMA the first one, and the second was fine.
Recently we got a bad 5930k cpu. This was a first for me, but after changing motherboard, ram etc. it turned out to be the cpu.

So, if you search on ANY computer part, there most certainly will be some bad experiences. That's why a good warranty is important, especially for recently released products.

[Ricky Johnson]

It really just happens with any hardware and vendor. The Asus Z9PE-D8 WS I had for a while was fine, but a friend who bought one before me, had to RMA the first one, and the second was fine.
Recently we got a bad 5930k cpu. This was a first for me, but after changing motherboard, ram etc. it turned out to be the cpu.

So, if you search on ANY computer part, there most certainly will be some bad experiences. That's why a good warranty is important, especially for recently released products.

You're right. I accept rationally that there must be a chance of defect in all components and that I have no great objective reason not to consider Asus as a choice again.

That said, I didn't think it was a bad idea for a topic to discuss the pro's and con's of any motherboard options for building dual cpu systems
(as it appears a fairly popular consideration for a build type amongst Corona users).

I should've left my own personal annoyance with Asus out of it entirely! It's meaningless to anyone else.

[Juraj]

No, it's good topic :- ).

There really aren't many choices. I think this market (enthousiast bords for Xeon builds) will get even smaller in upcoming years when 10-12core 3+Ghz consumer CPUs will come (like 6960X or whatever that will be called, or if even)

[Ricky Johnson]

Well, if the future of Xeons is going to be CPU's with more cores also then wouldn't the argument between those future dual Xeon builds and a future i7 build stay roughly the same?
In terms of a rendering workstation for enthusiasts - but I think the same argument applies for video encoding. I don't know who else utilises multi-core applications.

[Juraj]

It's just my hypothesis :- ) But here's my reasoning:

Biggest rise in popularity of Dual-Xeon builds for enthousiast happened with arrival of LG-2011, where consumers were stuck for 3 years with 6-core i7 at best.

Intel has been slow with better performance yields per architecture with each new generation (+/- 3perc. Sandy/Ivy/Haswell/Broadwell/etc...) so new Xeons utilize more cores, but at same Watt output (they have a limit of 145W for strongest "WS" Xeon range meant for workstation, which are just re-branded i7), their total performance doesn't rise that much.

On other hand, consumer CPUs are getting similar amount of cores slowly due to public demand, but keep unlocked potential and users can run them at 300+ Watts if they wish, clocking them into potential where they rival dual-xeon builds merely thanks to overclock.
4.5Ghz 8-core, gives same performance as 2x10 cores at 2Ghz, which is mid-level type of Xeons that most users get ( thinking they will get superior performance due to cores alone... ) I can't say how often I facepalm when I see enthousiasts boast about 2x2630 v3 or other weak/wasted money build.

Unless total silicon revolution hits us in next 3 years (highly unlikely), Xeons will be still limited with their output, but consumers will have high-core, unlocked cpus that rival best Xeon builds at lesser fraction of price. Intel won't change their Xeon architecture for enthousiasts (super small fraction of Xeon customers), they will never bring unlocked monsters with extreme heat output because they can hardly produce them and there is no mass market for them (they can't be cooled in server racks anyway).

The enthousiast/pro-consumer niche users who buy them now, will therefore shrink further.

I myself, have given up on them. If I can get high-clocked 10-core i7, I have 80perc. of dual-xeon builds, in regular workstation with super fast single-threading and very responsive test renders. For the rest I use Cloud rendering anyway, the time of building personal farms is long gone by too.

[Ricky Johnson]

That does stack up if the Xeon's specs are going to stagnate. I don't know enough to interpret what that Wattage limitation means for future Xeon's.
Does it mean that they won't be able to add cores to the next generations without lowering the clock-speeds currently available or is this purely a limitation that prevent them pushing the speeds higher than the current range?

If the top i7 chip gains 2 more cores @ 4.5ghz (overclocked) then all dual Xeons need to do to redress the balance (based on current e5-2660-v3) is add 2 more cores to each cpu @ 2.6 ghz. (very roughly speaking!).

However, if the Xeons are stuck at their current speed/core specs, then yes, I can see the appeal of those disappearing.

[Juraj]

Good illustrative example could be 2680, the upper middle horse ( I use these in v2 version ).

2680 v1  /32nm  / 8cores@2.7 Ghz  / 130W   /2012
2680 v2  /22nm  /10cores@2.8Ghz  /115W   /2013

Here you can see small silicon revolution: Due to 32nm -->22nm, they increased both frequency and core amount, at same time even lowering wattage. 32nm has been with us for 9 years !
Yet, this is hardly impressive result but it's what we get, the development of CPUs is moving pretty slowly in past 10 years.

2680 v3 /22nm /12cores@2.5Ghz / 120W /2014

Here you see very modest, almost invisible 'Tock' update. With further increased density across whole range, they had to lower frequency and still ended up with higher wattage to bring any kind of performance improvement. Almost zero upgrade, although at highest densities of the very upper top models, that is still noticeable improvement for massive server builders where density is everything.

True, 14/10nm might shake this again, but nobody expects miracles. 14nm was originally planned by Intel for 2013 :- ) But it will come 2016/2017.

I think Intel is generally moving away from making these CPUs anyway worthy for enthousiast. Their focus is now far more specialized onto massive farms than they have ever been (with E7 4xxx / 8xxx units) with massive density, many sockets and low frenquency. They will simply get less and less usable in workstations as years come by. What good is 4-socket 120 cores at 1.8Ghz for workstation ?

Enthusiasts jumped onto server platform out of necessity, but now when consumer parts finally caught up, I think there is no longer that need. And Intel knows it imho.