To respond to a few comments by iXXeon, shadow001, texinga,  and beaker7, before 3D Fluff ceased updating their cinebench scores site, an Opteron 6174 system with 48 cores and threads running at 2.2 GHz on Win XP 64 bit reported the highest Cinebench 11.5 score of 27.2. 
 
There are probably, at least, two Sandy Bridge Xeons that are strappable and unlocked.  They are the 1P parallels to the i7 3960X and the i7 3930K, i.e., the 1660 and 1650, running at 3.3 -> 3.9 GHz and 3.2 -> 3.8 GHz, respectively.  But we want dual processors.
 
But I still want that SRX to run a pair of 2687W, particularly if EVGA exposes in bios (or sets a switch on the motherboard) to intercept and deflect all of the signals to PCIe devices and any other new slaves of DMI's clockery so that they think that they're being requested to run within a range of 100 to 108 at all times; then my Sandybee E5s will boot and run with a BCLK in excess of 125, giving still vitality to what SR represents.  Jacob, can this be done?  Will signal tweaks come to the rescue?
 

shadow001
Thing is that the latest G34 processors have 16 cores a piece now(not just 12) and run at 3.1 Ghz each(not 2.2 Ghz), so given the system you mentioned scored 27.2, i'm guessing that with 16 extra cores and close to an extra 1 Ghz on every core(all 64 of them), i'm guessing it would probably be north of 40 points on cinebench 11.5, and that's far above the 23~24 points seen on the SR-2 with a pair of X5690's at 4.5 Ghz, and as we all know, the fastest E2600 Xeon's are stuck at 3.1 Ghz despite having 2 extra cores/4 threads over the X5600 lineup. 

They cost 1200$ for the fastest G34 processors with 16 cores at 3.1 Ghz, wich is expensive obviously, but the fastest 2 socket Xeons are usually 1800$ each and about the same clock speeds too.

Your logic is flawless, but I just snagged a pair of 2687Ws at a deep discount from someone who no longer wanted them when he found out that they couldn't be over clocked like CPUs of the past.

shadow001
... .
With a heavy discount it make it better obviously since they're not cheap to begin with, but to be honest i'm not going to ditch the SR-2 i own anytime soon, and might even add a pair of X5690's and overclock them to 4.5 Ghz as they're bound to become a lot cheaper since the E2600's are replacing them and most retailers want to clear any inventory they may still have on X5600's in general.

 
 
Superbiiz currently has naked x5690s for $1509.
 
I think that if EVGA had prior knowledge of the E5 limitations that I just recently became aware of a couple of months ago, that EVGA would not be going down this path to a 2011 tweakable, higher than ASUS and SR2 priced, motherboard, unless they had some tricks up their sleeves. Moreover, if they merely wanted to build an ASUS-like vanilla server board, I suspect that it would already be in the pipeline since afterall EVGA has an SR2 template.  If the reason why an E5 (or for that matter any Sandy Bridge non-E and non-k) chip will not post, if the BCLK is pushed much higher than 5-8 notches, is limited solely to what's happening within the CPU, then I'm SOL,  except in so far as EVGA allows me to tweak everything else and just do minor tweaking of the CPU's BCLK.  However, if the failure to post results from what that DMI bus does to externals, then EVGA may have found a way to intercept, transform and massage those signals to allow the externals to survive the high speed onslaught of a higher BCLK.  My hope does spring eternal.  However and in any event, for about $2.000 less than the total price of my SR2, I'll build an SRX/E5-based WolfPack3 to add to my Cinema4d/Final Cut Pro render farm, and it'll run cooler and quieter than my SR2, while running a tad faster.

It is disappointing to me that you have once again chosen to go with the HPTX form factor instead of something more standard such as EEB (Which is the same size as E-ATX but with a few minor mounting hole differences on the right side of the board)
 
All throughout the SR-2 life cycle people have complained about the HPTX form factor and the lack of cases and all you have responded with was that it was not possible to make the board in a smaller form factor.
 
Now Asus releases an EEB Dual Xeon LGA 2011 board with Quad-SLI support and overclocking with more SAS 6Gb/ps ports than the SR-X and at a lower price whilst still supporting the same overclocking features as the SR-X does. I think this proves it can be done but you didn't want to do it.
 
I really do question the lop-sided memory configuration of the SR-X it looks quite ridiculous. The board has a lot of empty space that is devoid of chips, surely the board could have been compacted in to a smaller EEB form factor if the empty space between the IC's was made smaller. Similar to how the Asus board looks.
 
And to top it all off as if things weren't bad already, it now turns out the new XEON's can only receive a BCLK overclock of about 5MHz - I really question why anyone would want to buy this SR-X after all we know now, of course the XEON overclocking situation isn't your fault EVGA but seriously what's the point in this board now? The only customers who will be interested now are workstation users and they will definetly want a standard E-ATX, CEB or EEB form factor, not HPTX.
 
Board without a market? I think so.

After the disappointment at the lack of unlocked XEONS there is some good news. A pair of stock E5-2687W XEONS will beat a pair of overclocked X5690 XEONS (see the Inquirer).
 

Imagine how badly the X5690's would be beaten if you could overclock the E5-2687W's?....That's the whole point of the SR-X.

moviemanxs

I know and I agree but the issue is and I'm guessing a bit here, that EVGA had to start many months ago on this project in order to be ready even close to the March7 release date on the cpu's and I'm betting way back then that no one knew the chips would have a locked BCLK strap. So here they are, tons of money invested and everyone( almost) is acting like EVGA did something wrong in some way.
They gambled and the dice don't always come up what you want. Ok, won't do the 34x150 strap that people were dreaming of but it will
do some incredible numbers and in terms of computational power is capable of doing more than a X5690 on the SR2 board
Lets give it a shot and also give poor Jacob a break.
Now I've bee saying over at XS "CALL INTEL and tell them that for $3800.00/pair you expect all the functions to be enabled and not be buying crippled cpu's. Write them, call them, post on this. Get everyone you know in on this.
Friends,wives,grandmothers,everyone!
Trust me, they get a couple hundred thousands emails,complaints and phone calls someone somewhere will notice.

 
EVGA didn't do anything wrong on purpose, but even before these Xeons were released, all they had to see is how the clock generator setup on regular X79 chipset boards are designed, where there's basically no independent BCLK adjustments anymore, and were dependent on unlocked multipliers on the socket 2011 processors, of wich only the 3930K and 3960k allow it. 
 
Those 3930k and 3960k processors are the same as these Xeon's that have just been launched, except they have 2 cores disabled and only 1 working QPI link, so the doubts were already there since last year if there were going to be Xeon's with unlocked multipliers available of not, but EVGA continued on the development of the SR-X anyhow and hoped for the best...
 
And it's a small technicality in the end, since intel can release extreme editions of the above processors, without calling them Xeon's at all, where all 8 cores are enabled and both QPI links working and the SR-X is back in business.....Call them 3930k EE and 3960k EE, charge 1000$ and 1500$ respectively, intel still makes a lot of profit margin and we're golden on the overclocking front.
 

already asked once and I did not answer ... when the MB is available for sale?

EVGA_JacobF

SR-X is in production and will be available the 2nd week of April.

 
Can't wait to get my hands on one

Tutor
Thing is that the latest G34 processors have 16 cores a piece now(not just 12) and run at 3.1 Ghz each(not 2.2 Ghz), so given the system you mentioned scored 27.2, i'm guessing that with 16 extra cores and close to an extra 1 Ghz on every core(all 64 of them), i'm guessing it would probably be north of 40 points on cinebench 11.5, and that's far above the 23~24 points seen on the SR-2 with a pair of X5690's at 4.5 Ghz, and as we all know, the fastest E2600 Xeon's are stuck at 3.1 Ghz despite having 2 extra cores/4 threads over the X5600 lineup. 

They cost 1200$ for the fastest G34 processors with 16 cores at 3.1 Ghz, wich is expensive obviously, but the fastest 2 socket Xeons are usually 1800$ each and about the same clock speeds too.

 
The fastest 62xx series Opterons are 2.6 Ghz, not 3.1.  Top end is the 6282SE which will run you about $1200 per.
 
I have a 4x Opteron 6174 system that runs benchwell in about 2:10 and Cinebench in 26 something.
 
My dual E5-2687 at 105 BCLK runs becnhwell in 1:58 and Cinebench in 27.xx, and is faster in single threaded apps.
 
The scaling decreases as the core count increases generally. But yes, a 64 core 2.6 Ghz Opteron would probably be faster than dual E5-2687 in Cinebench.  It's also slower in poorly threaded tasks, leaves you even fewer case and board choices, and requires the use of a server OS or linux - which, on Windows, can become expensive if you want more than 32GB RAM as their server versions are tiered by socket count and memory.
 
So in my experience, for a workstation the 2P angle is the best balance.  If for some reason you could only have 1 box and just cared about render speed, the Opteron might be a good choice.  You'll also find the heatsink choices on G34 very limited, and most if the Dynatron stuff is outrageously loud, so you'll almost certainly have to water cool it if you want to  use it deskside.

I've been sitting on the fence for a sr-2 for some time now. I have however been staging my current setup to enable a quick adoption of an sr board. That said I can not wait for the sr-x to become available. I already have the Lian-Li V2120 case, and would just need the cpus, and extra 8 sticks of ram, one extra h70, and that's it. Seriously can't wait.
 
Of course I'll only be doing dual card sli (2 580s), but I'll be happy knowing I can upgrade for some time to come.
 
Thanks EVGA for giving us computer enthusiasts something to drool over!
 
 

Late april is what i heard....It's already in manufacturing.
 
Anyhow, as much as i'd love the get the SR-X i need to have confirmation that there will be Xeons with unlocked multipliers(or that EVGA found a solution around that issue bios wise), as in practical terms, there's nothing slowing down my current SR-2 and i see that ever since the socket 2011 Xeons hace been released, the prices on the earlier generation X5600 series has been droping quite a bit....I can even find X5680's at 1100~1200$(they usually cost 1700$), wich would be more than enough to hit a 4.5 Ghz overclock, water cooled.

Indeed, I heard that the may be overlockable by around 200mhz but i'm not sure i'm even going to bother, I was after stability and a high core count on the LGA2011 socket. Just picked up one of these:
 

 
Second one coming in a week or two...
 

psyq321

At the moment there are no Xeons with unlocked multipliers, and I doubt very  much that EVGA managed to get around the missing ICC path limitations in the current Xeon E5 SKUs.

So, mild BCLK bump is your only option - expect to push it to 105-106 (and if you are really lucky few MHz above) which will translate into ~200-ish MHz overclock.  You could still probably do memory overclocking (if this makes sense for you) so the board itself still has some good applications and advantages compared to the plain vanilla server boards which are completely locked.   There is also undervolting - not something most "enthusiasts" do (they do it other way 'round :-) ) but it could be interesting for thermally limited applications if people are willing to experiment and find the lowest possible stable voltage.

This way you could probably get the TDP of 2687W down to ~100W (I could do this even with the B1 samples - and the retail C2 silicon should be even better in this regard)

Does it warrant the difference in price compared to plain vanilla server boards like Supermicro X9DAi?   This is entirely depending on your use and need for EVGA-specific features.

But in any case, at least for the time being - I think we can forget about huge overlclocking gains for E5 Xeons.

 
I suppose for applications that are very memory bandwith sensitive and not dependent on CPU clock speed, the option to overclock the memory makes sense, but it is a pretty specific scenario.
 
On the good news front, i've seen X5680's going for under 500$ a piece, and i can only guess why they're so low(company going out of business perhaps ???), as in most other suppliers, they're still at 1200$ so that has my scam meter going into overdrive....It is tempting to max out this SR-2 with the processors running at 4.5~4.6 Ghz and water cooled though. 

Excellent :D This was what I had hoped would happen <3 Thanks for this bit of news!