Hot!eVGA SR-X (SR-3)

The ability to overclock is in the chipset because that is where the clock generator is buried, that's how Intel limits certain Sandy Bridge CPUs to overclock in certain boards and how flexible they can be; same logic should be the case with Xeons-- provded EVGA tweaked a version of the C60x chipset that SR-X uses.  Then my guess at the very least most of the Xeon E5's will overclock like non-K parts with a minimum amount of give.  Thus there must be at least one unlocked Xeon this time.

how many are x16 PCIe 3 slots?  

CUDA/OPENCL applications can saturate the x8 bus and ALL build in vram. specially if these are dual gpu cards 

MySchizoBuddy

how many are x16 PCIe 3 slots?  

 
Depends how many cards you run. 3 cards they will all run x16, 4 cards 2 of them will run x8.

EVGA_JacobF

MySchizoBuddy

how many are x16 PCIe 3 slots?  

Depends how many cards you run. 3 cards they will all run x16, 4 cards 2 of them will run x8.

what about running all 6 or 7 are they at x8 or they drop to x4

MySchizoBuddy
how many are x16 PCIe 3 slots? 

If both CPUs are installed then you get a full 80 lanes of 3.0, mathematically it would be best spread among four x16 and two x8, with the 7th slot being x8 2.0 routed through the chipset.  But if one CPU is installed, my guess is that it will be all x8, similar to most X79's. 
 
But EVGA's X79 differs from the rest since their 3-way SLI is configured 16/8/8 instead of 16/16/8, so fingers crossed on what SR-X's PCIe configuration ends up.

EVGA_JacobF

MySchizoBuddy

how many are x16 PCIe 3 slots?  

Depends how many cards you run. 3 cards they will all run x16, 4 cards 2 of them will run x8.

 
With two processors there are 80-lanes. Why can't 4 cards run x16? That would leave 16 additional lanes to route to other slots or use for other purposes.

Here's what he said on OCN:
 

EVGA-JacobF
PCI-E lanes are generated by 1st CPU only.

Also, is a PCI-E 3.0 bridge chip under the chipset cooler. Total it has 48 lanes.

Brocasta
Here's what he said on OCN:

EVGA-JacobF
PCI-E lanes are generated by 1st CPU only.

The ASUS Z9PE-D8-WS does not use the nVidia bridge chip, there for will be a better solution for those running mega cuda cards.  Also they got rid of those silly extra ram slots that any true oc'er will never use since they increase load on the memory controller.
Let me say I'm a huge fan of EVGA, but they totally flubbed the pci-e lane configuration of this board.  It isn't a problem not to have all pci-e slots active with only on processor installed.  people will not be buying this board with the intent of installing a second processor a year or two later, not going to happen.

It's not my preferred design either, but let's just wait and see how the board (actually, both boards) performs after it's released.

The ASUS Z9PE-D8-WS does not use the nVidia bridge chip

 
Neither does the EVGA SR-X.

Also if you populate all PCI-E slots, all will run at 8x with the exception of PCI-E slot 6, which will run at x4.

In that circumstance, all except the 6th slot will run x8 PCIe 3?

Yes.

what is that chip next to the pci-e slots?  when will the block diagram be released?
 
If no bridge chip then why not use all the pci-e lanes off of both processors?

He said it's not an nvidia bridge chip.

Brocasta
He said it's not an nvidia bridge chip.

AND I ASKED WHAT IT WAS!

he is under NDA dont expect much info from him till the powers that be allow it.

linuxrouter
With two processors there are 80-lanes. Why can't 4 cards run x16? That would leave 16 additional lanes to route to other slots or use for other purposes.

We still don't know what the actual layout configuration is, the 16/16/8/8 layout may be if four dual-slot cards were stuck in there. 

rottenmutt
what is that chip next to the pci-e slots?

a variant of the Patsburg chipset, i.e. southbridge.

rottenmutt
when will the block diagram be released?

You may have to wait for a visual guide just like EVGA did with SR-2.  Besides, I answered that back on page 3.

rottenmutt
If no bridge chip then why not use all the pci-e lanes off of both processors?

 I'm curious of this as well, though it may have something to do with the rest of the layout; whatever was sacrficed for better overclocking.  Either way, it will test users brand loyality whether they want to overclock both Xeon E5's or willing to go for another board that may not overclock as well just to have more lane support.

We are using a PLX chip for PCI-E 3.0.
 

EVGA_JacobF

MySchizoBuddy

how many are x16 PCIe 3 slots?  

Depends how many cards you run. 3 cards they will all run x16, 4 cards 2 of them will run x8.

 
That is what I originally thought ( http://forums.evga.com...67&mpage=1#1379636 ), but EVGA_MatthewH said the "SR-X will support 4-way SLI at full 16x for each slot." I guess he was just misinformed or the final specs were not out then. I would be interested in why only one CPU is being used for the PCIe slots. My only guess is that the SLI specifications don't allow 2 CPUs to power the slots. If that is the case, and if ASUS does not use a bridge chip, don't expect their board to support SLI (unless if they say otherwise). Remember, they showed off their board using Tesla cards, not GeForce cards in SLI. It is a workstation board after all.
 
In retrospect though, this is PCIe 3.0. The difference between x8 and x16 is minimal for now. Although I don't know how much latency the bridge chip adds.

Ah, the tesla thing is a very interesting observation. Good thinking.

EVGA_JacobF
We are using a PLX chip for PCI-E 3.0.

Can I ask why you choose not to use the pci-e functionality in the second processor?
http://www.plxtech.com/pr..ts/expresslane/switches

SLI works off of two north bridges, ask supermicro and asus.

Now it is undetermined if OC'ing CPUs will even be available. So at this point the asus board would has better pcie bandwidth. Not to mention it dont have a silly memory config.

On the Asus board there are seven PCI-Express 3.0 x16 expansion slots, from which four (blue) are x16 capable, and three (black) x8 capable.

Will this be available march 5th for sale? Will actual specs be advertized earlier? What speed is RAM?

A few more details on the SR-X thanks to some fine details on the photos that have been released:
We already know the PCIe setup is as follows:
[x16/x8, x8, x16/x8, x8]=connected via x16 lane using a PLX PEX 8747 PCIe 3.0 switch ( http://www.plxtech.com/pr.ucts/expresslane/pex8747 ), x16/x8, x4, x8
Another x4 lane connects to the South Bridge for SAS support, using all 40 lanes from the primary CPU.
 
Now looking at the pictures, I have a better idea what chips are connected to the South Bridge:
 
Ethernet Controllers: 2x Intel 82574L Gigabit Ethernet Controllers
http://ark.intel.com/prod...it-Ethernet-Controller
Reasoning: We know that the board uses Gigabit Ethernet. Looking at pictures of the board, you will find 2 Intel Gigabit Ethernet controllers, both of which are connected via 64-pins. Looking at Intel's site, there are only three Single Gigabit Ethernet Controllers that interface via PCIe that use a 64-pin design: 82574IT, 82583V, and 82574L. 82574IT is the industrial version of the 82574L, so that can't be it. The 82583V is not a product I normally find in dual Intel Gigabit motherboards, the 82574L is, leading me to think the 82574L is used. In addition, the 82574L supports Jumbo Frames, something a highend motherboard like SR-X would probably support.
 
eSATA controller: 1x Marvell 88SE9220
http://www.marvell.com/st...92xx-Product-Brief.pdf
Reasoning: This one was easy. Looking at the board, you will find a Marvell chip right behind the eSATA ports. The fact that it is a Marvell chip is easily identified by the logo. Looking closely, you will notice that the chip uses only 56 pins. Luckily, only one of Marvell's SATA controllers uses 56-pins, the 88SE9220. One thing to note about this controller is the ability to connect it via x1 or x2 PCIe 2.0 connection. Most likely EVGA used a x2 connection considering that they have an abundance of PCIe 2.0 connections on this motherboard. There simply would be no reason not to. This would guarantee 500MB/s speeds on both those eSATA ports if both are being equally maxed.
 
Back Panel USB 3.0 Controller: 2x ASMedia ASM1042
http://www.asmedia.com.tw...=100&cate_index=98
Reasoning: After examining the photos for a little bit, I spotted the ASMedia logo (first looked like a C-Media logo which would be impossible because the board already uses Realtek for an audio codec). The controller is located in between the 2 Intel Ethernet controllers, just behind the USB 3.0 ports. I couldn't find much information on the controller, but I think it only supports 2 ports. I was able to find the second chip below the secondary processor's DIMMs.
 
Onboard USB 3.0 Controller: 1x ASMedia ASM1042
http://www.asmedia.com.tw...=100&cate_index=98
Reasoning: There are no good photos of the onboard USB 3.0 controller chip on the SR-X, but the logo looked similar to the back panel chip and there is no reason to use different controllers.
 
Onboard Firewire Controller: 1x TI XIO2213B
http://www.ti.com/product/xio2213b
Reasoning: Again, the photos of the bottom of the board are not that great, but this is what I have. The logo on the controller appears to be a TI logo. I cannot spot a PCI bridge on the motherboard, indicating that the Firewire controller should be able to connect via a PCIe interface. TI makes 2 Firewire Controllers that connect via PCIe, the XIO2213B and the XIO2200A. According to the X79 Classified manual, 1394b is optional, indicating that the controller used on that motherboard is one that supports 1394b. The XIO2200A only supports 1394a, leaving only the XIO2213B to be used on the X79 Classified. I assume the chip used on the X79 Classified will be the same one used on the SR-X.
 
Audio Codec: Realtek ALC898
Reasoning: The audio codec is definitely Realtek, supports 8-channel audio, and uses a 48-pin connection. That really doesn't limit the options. I say it probably is the ALC898 since that is what the X79 Classified uses.
 
In short, I currently count the use of all 8 out of the 8 available PCIe 2.0 links from the South Bridge. (x2 for the dual Gigabit, x2 for the eSATA, x2 for the 4 external USB 3.0 ports, x1 for the 2 internal USB 3.0 ports, and x1 for the internal Firewire.) Of course, I could be wrong on a few of the chips. Whoever gets this board should be able to verify which chips are used. As a note of reference, I used the Photosynth photos of the board located here:
http://photosynth.net/vie...4f50-ae10-a25d35d33d73
 
I also used Jacob's leaked photo of the SR-X on Twitter located here:
http://twitpic.com/73bm7z

Interesting detective work overall, but why do you think the SR-X has Firewire support?
I'm a bit surprised, since Firewire is pretty much a legacy technology at this point, even with 1394b. I wouldn't include it on a high end mobo, as it is just something else that could mess things up.

If you look at the onboard ports at the bottom of the board, you will see 1 blue port/pin layout, 3 red ports/pin layouts, and 1 white port/pin layout. The blue one is for USB 3.0. The red ones are for USB 2.0. The white one is for Firewire. It is labeled "1394B HDR" aka Firewire 1394B Header. The same port/pin layout is on the Classified X79.
 
As far as the usefulness of Firewire, it was one of those technologies that never broke through since it was costly to implement it. Nevertheless, there are applications where it is superior to USB, for instance, with the use of external audio DACs (normally very expensive ones). This is a good read on how Firewire is superior to USB for those types of applications:
http://www.weiss-highend....chnicalBackground.html

Yeah, now I see it. You know what's funny? When the SR-2 came out, people were posting, "Where's the Firewire header?", because the SR-2 didn't have Firewire. I think these were people with older video cameras, etc. Now that hardly anybody needs Firewire, the SR-X has it.
 
What will be interesting to see, is if the SR-X has Realtek audio digital I/O ports on the back panel or if you will still need to run you own cable to the digital audio I/O header on the mobo,like with the SR-2.

There is an optical S/PDIF port on the back with the other audio ports.

A block diagram for the SR-X would be sweet. Supermicro and Tyan have released some of theirs, probably accidentally, but still. Without going back and looking, I know the SuperMicro boards use all 80 lanes.
 
Fun fact: one of the SM boards has 11 PCIe slots.

Aggressor Prime

There is an optical S/PDIF port on the back with the other audio ports.

Cool. Thanks.

Jacob, 
 
As the E5's are starting to show up in online ordering, can you tell us if we'll need to wait for some future variant of unlocked E5 to get full overclock capabilities?  Would like to get started ordering but don't want to jump the gun.
 
 

10-core Ivy Bridge-EP Sample Tested
 
http://www.techpowerup.co...-EP-Sample-Tested.html
 


 
The second image seems to be a 8 core-16 thread x 20 cpus system.
 
The multipliers seem to be unlocked for the first xeons, and 8.8 seconds for the 1M pi seems to be great at 2.8GHz, my e5645 at 3.8GHz does it in 10 seconds.

Methinks next week is going to be very illuminating. Intel is supposed to unveil and probably release the 2600s next week at cebit, and once that happens, Jacob will be more able to answer questions about how OCing is going to work.

massdestructor

10-core Ivy Bridge-EP Sample Tested

http://www.techpowerup.co...-EP-Sample-Tested.html




The second image seems to be a 8 core-16 thread x 20 cpus system.

The multipliers seem to be unlocked for the first xeons, and 8.8 seconds for the 1M pi seems to be great at 2.8GHz, my e5645 at 3.8GHz does it in 10 seconds.

 
 
ETA LGA 2011 Ivy Xeons?
 
Could go for 20 physical cores on SR-X @ 4.x ghz.

Probably end of the year or first part of next year.

hi
the cpu yes , the SRX , hope midle 2012
 
i have two 8C 3ghz for the SRX

Unless EVGA is having problems with the BIOS or something on the SR-X, I'd say its release is imminent.

Wow... I'm glad this thread is doing so well^^*
Haven't made a good thread before>.<
 
I heard that the x79 platform has some issues...
I don't think this should be released until all those issues are worked out...
Not saying this is using a x79 chipset...
Like the SR2, it may use a server version of it, however, I think the x79 problem will be similar.
 
Just saying...
Correct me if I'm wrong^^

Who knows. Jacob is in Taiwan with vince and tin, and per this thread:
 
http://kingpincooling.com.../showthread.php?t=1640
 
It sounds like they're in the presence of either a functional SR-X system or Kepler. I would put more faith in it being the SR-X because I can't see them wasting untold amounts of LN2 on midrange Nvidia hardware. Unless GK104 is not midrange...

So Im at a loss here.. Im not sure which information is correct since its all coming from the same person. EVGA_JacobF  stated that 4 way SLI will run at all 16x ..   a few pages ago he stated that now it will run 2 16x and 2 8x ...   their shocase at CES 2012 advertises SR-X to run 4 way sLI at 16x ...  so which is it..  (The video below at 0:38 you can clearly see 4 way sli @ 16x   :   ) 
So which is it can a mod please clerify ...   On several threads you guys have stated both versions..  perhaps it will be two versions of the board...    just someone please clerify the actual set up we will get.
 
PS: I love the waterblocks also shown in the CES 2012 showcase.

sorry the address was removed /watch?v=J3_Crzey0as at youtube for the CES footage ..    which was also posted by jacob..

Just chill. We'll know for sure very soon.

Brocasta

Just chill. We'll know for sure very soon.

 
  
      When's that?
         Pretty soon. Don't worry.
     Pretty soon?..  How long is that? 
         In a little while, man! 
      ...A little while? What'll we do 'til then?
  
 

I really hope EVGA is launching their new PSUs with the SR-X.

lollygag

Brocasta

Just chill. We'll know for sure very soon.

 
 
     When's that?
         Pretty soon. Don't worry.
     Pretty soon?..  How long is that? 
         In a little while, man! 
      ...A little while? What'll we do 'til then?
 

love ren and stimpy! I used to have an original full size poster of that pic:)

Aggressor Prime

I really hope EVGA is launching their new PSUs with the SR-X.

Me too. That 1500w unit is just perfect for the SR-X.

So it begins:
http://www.newegg.com/Pro...arch=1&srchInDesc=

wow, a couple srx boards and some intel hardware amd the us military will be contacting us for computer might... hahaha

Still don't get why the 2690 is more expensive than the 2687w.
 
Now all we need is a motherboard.  Newegg only has a supermicro
 
http://www.newegg.com/Pro...x?Item=N82E16813182348

Ahhhh the anticipation is killing me!

sokyoku

So Im at a loss here.. Im not sure which information is correct since its all coming from the same person. EVGA_JacobF  stated that 4 way SLI will run at all 16x ..   a few pages ago he stated that now it will run 2 16x and 2 8x ...   their shocase at CES 2012 advertises SR-X to run 4 way sLI at 16x ...  so which is it..  (The video below at 0:38 you can clearly see 4 way sli @ 16x   :   ) 
So which is it can a mod please clerify ...   On several threads you guys have stated both versions..  perhaps it will be two versions of the board...    just someone please clerify the actual set up we will get.

PS: I love the waterblocks also shown in the CES 2012 showcase.

 
Its 2x16 and 2x8

Any info on release date now that Xeon 2600-series has launched?

Asus board is up a serversdirect.
 
http://www.serversdirect....oduct.asp?pf_id=MB1375

EVGA_JacobF

sokyoku

So Im at a loss here.. Im not sure which information is correct since its all coming from the same person. EVGA_JacobF  stated that 4 way SLI will run at all 16x ..   a few pages ago he stated that now it will run 2 16x and 2 8x ...   their shocase at CES 2012 advertises SR-X to run 4 way sLI at 16x ...  so which is it..  (The video below at 0:38 you can clearly see 4 way sli @ 16x   :   ) 
So which is it can a mod please clerify ...   On several threads you guys have stated both versions..  perhaps it will be two versions of the board...    just someone please clerify the actual set up we will get.

PS: I love the waterblocks also shown in the CES 2012 showcase.

Its 2x16 and 2x8

 
What the, since when did this happen? Why can't it do 4 way X16?
 
EDIT: Okay so I read the last few pages, why aren't you using lanes from the secondary CPU?

 

beaker7

Still don't get why the 2690 is more expensive than the 2687w.

Now all we need is a motherboard.  Newegg only has a supermicro

http://www.newegg.com/Pro...x?Item=N82E16813182348

 
Intel Xeon E5-2690 Thermal Design Power 135W
  Intel Xeon E5-2687W Thermal Design Power 150W 
 
the 2690 is less power hungry and in servers that counts

bugzyme

 

beaker7

Still don't get why the 2690 is more expensive than the 2687w.

Now all we need is a motherboard.  Newegg only has a supermicro

http://www.newegg.com/Pro...x?Item=N82E16813182348

Intel Xeon E5-2690 Thermal Design Power 135W
Intel Xeon E5-2687W Thermal Design Power Thermal Design Power 150W

the 2690 is less power hungry and in servers that counts

 
I get that.

But why wouldn't you just buy the cheaper one and downclock it?

beaker7
Still don't get why the 2690 is more expensive than the 2687w.

TDP rating. Unlike enthusiasts, the multi-CPU community has a segment willing to pay more for a CPU that can operate at a lower TDP with the same performance as a higher wattage part; resulting in lower power bill and lower cooling bill, considering the typical Xeon customer purchases dozens to hundreds of these CPUs.
 
Just so you know, Asus has several variations of dual socket boards because they've been making multi-socket boards for both AMD and Intel for years.  There isn't just one just because that's all enthusiasts care about regarding SR-X. 
 

Halfdead14

EVGA_JacobF
Its 2x16 and 2x8

What the, since when did this happen? Why can't it do 4 way X16?
EDIT: Okay so I read the last few pages, why aren't you using lanes from the secondary CPU?

They are sort of, 40 lanes per CPU means that 2x16 + 2x 8 = 48 lanes, the other eight has to come from somewhere.  I'm sure the other lanes are going in other slots since we still don't know the configuration, JacobF is just stating that  four dual-slot cards will see dual x16 3.0 and dual x8 3.0.  I'd like to think there are single-slot configurations with more lanes, but I don't know.  We need the configuation to determine what is or isn't possible with one or two CPUs.
 
I'm sure EVGA made sacrifices just to enable Xeon overclocking in pairs; just as well if lanes matter more than overclocking, then I suggest get a regular dual 2011 by either Supermicro or these by Tyan or wait for Asus and Gigabyte to chime in. 

Yep, the other 2S Romley boards use both sets of 40 lanes. I'm sure EVGA bucked this trend for a reason.

So at least in my browser, when looking at the forums, today there is a big blank spot in the upper right where the X79 banner was.  Is that a hint of things to come shortly?