three chipset?
 
can you list them out for me so that I can google the info? 

I'd say in order to make a "new" Super Record, please give us a chipset with quad-socket and put in four new 8-core 16-Thread CPUs
 
then 32 slot of ram should probably right...  
4GB each already makes it 128GB of ram 
 
32 Core 64 Thread 128GB here comes the Super Record 

thegreatga

you would setting a record to power it as well as to fitting it into one single case

actyally Mountain mods has two or three cases that will support Quad socket motherboards like the AMD ones you see around the net.

Maybe a stupid/obvious question but is the SR-X gonne be HPTX or something els?

Really, this should be called "SR-EX" or "SR-XXX"
 
 

Looks tasty ! but it better support all version of ocz revo X3 and X2, because iam planing to get one :)  probably max IOPS one.
it was really disappointing that best mobo on planet (SR-2) do not boot up from fastest HDD ..
btw LOM is still slow dual 1 Gbit? or its dual 10 Gbit now?
 
 

robhall86

INFRNL

I think the ram is all shared among the 12 slots but I have no idea. It was also mentioned that they did it this way so you can use your ram from SR-2 for this board and not have to buy new ram.
It is kind of odd depending on how it works exactly, but its nice that they thought about saving us some money and not have to buy new ram. I personally cannot wait for the release

jacob said that the reason was to make is easy on current SR-2 users.

after all it would suck to have 12 sticks and upgrade only to have to put 4 in storage.

Post 63
http://forums.evga.com/tm.aspx?&m=1158148&mpage=3

I thought this system would have quad channel ram instead of tripple...
don't you have to change the ram firmware in order to do that?

Still seems kinda odd that people (such as myself) will spend thousands of dollars on these system and evga wants us to save a hundred dollars on the cheapest thing that goes into the computer... they should implement 16 dimm slots before the release of this bored. But this is not going to happen being that they haven't done so already and this bored will probably come out soon... I will probably just get the sr-2 since it will be cheaper... or nothing being that I already got a good 4way system^^* but buying new parts is an addiction>.< so Iz kantz stopz myself! lol

This new motherboard should come out in this form... but with 7pcie slots^^ maybe call it the HPTX-E lol
It has enough space for 8 dimms per gpu... can also shift it around so that there are 4dimms on each side of each gpu for less latency...

They're not going to create another, bigger form factor for the SR-X. Don't be stupid.

cateno

this tyan for  old opteron socket F not for lga2011
th X79 is capable for 4 cpu

Yeah i know... but the layout is there...
and the srx is not an x79 bored... just like the sr2 is not an x58 bored... sure they use the lga1366 and lga2011 but they both use some kind of dual cpu server chips to work... dunno what is it called... lol

An idea doomed to fail (yet another, and bigger, form factor board that has at least $7000 worth of CPUs on board) is an idea that should not be tested. EVGA doesn't just release products for the sake of releasing them. They have to know if there is a market for the product, and a massive and, by your desire, completely proprietary board that uses E5-4600 CPUs is not exactly going to be purchased by more than 10 people. Not worth the effort. There was enough skepticism that they would try for another dual socket board after the SR-2, let alone a quad socket xeon board that would be out of the price range of nearly everybody except those that buy existing quad socket and octo socket Xeon products and systems.
 
On that note, why shouldn't EVGA do an 8 socket board? A stupid idea is an idea that isn't tested, after all!

I can tell that you didn't read or understand my post.

cateno
the X79 is in four version and one version is for 2 cpu
http://www.comptoir-hardw...1/x79_scenario_vrz.jpg
the SRX have minimum the T or D version for sas port fully fonctional

SR-X uses Patsburg-T while X79 uses Patsburg-X.  Just so you know, Tylersburg used "-S" for X58 and "-D" for i5520 (SR-2's chipset), while Intel's Skulltrail board used a Seaburg chipset.  All "-burg" chipsets are server chipsets, Intel will not call their SNB multi-socket Xeon chipsets "X79".
 
FYI, there are four Xeon E5 series coming and in two sockets (LGA1356 & LGA2011) making five combinations of boards, having five Patsburgs isn't a coincedence; one for each board.
 

takuhari
because you cant fit it into a case, doesnt mean people who uses these in a test environment wont wanna go beyond extremes...
But sure... go ahead and limit yourself because it is not conventional>.<

You're looking at this from the wrong perspective.  Products don't just exist because someone demands, products exist because enough demanded to pay for the venture.  Whatever it is must be worth a company's time and money, otherwise the customer's hopes and dreams of being different don't matter.
 
All computer components and their form factors exist because there was demand for it all throughout the development cycle and that demand continued to spawn not only that initial company's break-even, but supplied enough revenue for other companies in competition.  Non-conventional, i.e. niche, demands are so small that the only way a company can make a buck is if the unit cost is high; but not everyone demanding the product can pay that, so it becomes a casuality argument (i.e. the chicken versus the egg). 
 
Many times people think the reasons certain products don't exist is because of technology or capability-- no, it is usually a Return-On-Investment issue:  The demand doesn't equal the cost to make it.  There needs to be more demand, either more customers or they each need to pay more.  Most customers use their computers for something, they aren't just building it which is a tiny fraction of the time they put it while using the machine.  Something proprietary or unconventional is a risk that can cost money and time, especially in service costs.  To reduce the guesswork, folks they will go for something that falls within an already existing standard.  Hence SR-X reusing SR-2's form factor. 
 
That doesn't mean you can't still get whatever you want that already exists; if you're willing to drop $2000 on an Intel quad-socket board along with four processors cost $4000+ each, go for it.   If you want an overclockable quad-socket board (which means you're willing to pay the 20-50% premium to get it done), you got to understand, this isn't about you.  This is about not having enough of people like you to bother making it in the first place.

A good idea for the next gen bored is optical...
make the buses optical buses... this will make it so that there is no magnetic interference between the bus wires and it will make pciex16 a thing of the past. bus speeds are limited by magnetic interference so this would be a good solution for top SR-quality^^ Only the wires in for emitter and receiver power is needed... this is will not be limited. Standard buses in systems is too slow for future architecture designs.... just sayin

Please post a link to the block diagram.