Ivy Bridge-EP (Xeon E5) specifications leaked

[CougTek]

Those of you curious about the details of the upcoming Ivy Bridge-EP chips that are supposed to be launched this September can view most of their specification on this page on ASRock's website. Only thing missing is the core count, but you can figure it out by the amount of cache (20MB=8 cores, 25MB=10 cores and 30MB=12 cores). I'm waiting on those to get out in the wild before upgrading the servers at my new company.

[TABLE="class: List"]
[TR="bgcolor: aliceblue"]
[TD]2011[/TD]
[TD]Xeon[/TD]
[TD]E5-2697 v2(C1)[/TD]
[TD]Ivy Bridge[/TD]
[TD]2.7GHz[/TD]
[TD]BLK 100MHz[/TD]
[TD]30MB[/TD]
[TD]130W[/TD]
[TD][URL="http://www.asrock.com/support/cpu.asp?s=2011&u=639"]List[/URL][/TD]
[/TR]
[TR="bgcolor: white"]
[TD]2011[/TD]
[TD]Xeon[/TD]
[TD]E5-2697 v2(C0)[/TD]
[TD]Ivy Bridge[/TD]
[TD]2.7GHz[/TD]
[TD]BLK 100MHz[/TD]
[TD]30MB[/TD]
[TD]130W[/TD]
[TD][URL="http://www.asrock.com/support/cpu.asp?s=2011&u=619"]List[/URL][/TD]
[/TR]
[TR="bgcolor: aliceblue"]
[TD]2011[/TD]
[TD]Xeon[/TD]
[TD]E5-2695V2(C0)[/TD]
[TD]Ivy Bridge[/TD]
[TD]2.4GHz[/TD]
[TD]BLK 100MHz[/TD]
[TD]30MB[/TD]
[TD]115W[/TD]
[TD][URL="http://www.asrock.com/support/cpu.asp?s=2011&u=618"]List[/URL][/TD]
[/TR]
[TR="bgcolor: white"]
[TD]2011[/TD]
[TD]Xeon[/TD]
[TD]E5-2690 v2(M1)[/TD]
[TD]Ivy Bridge[/TD]
[TD]3.0GHz[/TD]
[TD]BLK 100MHz[/TD]
[TD]25MB[/TD]
[TD]130W[/TD]
[TD][URL="http://www.asrock.com/support/cpu.asp?s=2011&u=653"]List[/URL][/TD]
[/TR]
[TR="bgcolor: aliceblue"]
[TD]2011[/TD]
[TD]Xeon[/TD]
[TD]E5-2690 v2(M0)[/TD]
[TD]Ivy Bridge[/TD]
[TD]3GHz[/TD]
[TD]BLK 100MHz[/TD]
[TD]25MB[/TD]
[TD]130W[/TD]
[TD][URL="http://www.asrock.com/support/cpu.asp?s=2011&u=631"]List[/URL][/TD]
[/TR]
[TR="bgcolor: white"]
[TD]2011[/TD]
[TD]Xeon[/TD]
[TD]E5-2687W v2(M0)[/TD]
[TD]Ivy Bridge[/TD]
[TD]3.4GHz[/TD]
[TD]BLK 100MHz[/TD]
[TD]25MB[/TD]
[TD]150W[/TD]
[TD][URL="http://www.asrock.com/support/cpu.asp?s=2011&u=634"]List[/URL][/TD]
[/TR]
[TR="bgcolor: aliceblue"]
[TD]2011[/TD]
[TD]Xeon[/TD]
[TD]E5-2687W v2(M1)[/TD]
[TD]Ivy Bridge[/TD]
[TD]3.4GHz[/TD]
[TD]BLK 100MHz[/TD]
[TD]25MB[/TD]
[TD]150W[/TD]
[TD][URL="http://www.asrock.com/support/cpu.asp?s=2011&u=654"]List[/URL][/TD]
[/TR]
[TR="bgcolor: white"]
[TD]2011[/TD]
[TD]Xeon[/TD]
[TD]E5-2680 v2(M1)[/TD]
[TD]Ivy Bridge[/TD]
[TD]2.8GHz[/TD]
[TD]BLK 100MHz[/TD]
[TD]25MB[/TD]
[TD]115W[/TD]
[TD][URL="http://www.asrock.com/support/cpu.asp?s=2011&u=652"]List[/URL][/TD]
[/TR]
[TR="bgcolor: aliceblue"]
[TD]2011[/TD]
[TD]Xeon[/TD]
[TD]E5-2670 v2(M1)[/TD]
[TD]Ivy Bridge[/TD]
[TD]2.5GHz[/TD]
[TD]BLK 100MHz[/TD]
[TD]25MB[/TD]
[TD]115W[/TD]
[TD][URL="http://www.asrock.com/support/cpu.asp?s=2011&u=650"]List[/URL][/TD]
[/TR]
[TR="bgcolor: white"]
[TD]2011[/TD]
[TD]Xeon[/TD]
[TD]E5-2670 v2(M0)[/TD]
[TD]Ivy Bridge[/TD]
[TD]2.5GHz[/TD]
[TD]BLK 100MHz[/TD]
[TD]25MB[/TD]
[TD]115W[/TD]
[TD][URL="http://www.asrock.com/support/cpu.asp?s=2011&u=632"]List[/URL][/TD]
[/TR]
[TR="bgcolor: aliceblue"]
[TD]2011[/TD]
[TD]Xeon[/TD]
[TD]E5-2667 v2(M1)[/TD]
[TD]Ivy Bridge[/TD]
[TD]3.3GHz[/TD]
[TD]BLK 100MHz[/TD]
[TD]25MB[/TD]
[TD]130W[/TD]
[TD][URL="http://www.asrock.com/support/cpu.asp?s=2011&u=651"]List[/URL][/TD]
[/TR]
[TR="bgcolor: white"]
[TD]2011[/TD]
[TD]Xeon[/TD]
[TD]E5-2660 v2(M0)[/TD]
[TD]Ivy Bridge[/TD]
[TD]2.2GHz[/TD]
[TD]BLK 100MHz[/TD]
[TD]25MB[/TD]
[TD]95W[/TD]
[TD][URL="http://www.asrock.com/support/cpu.asp?s=2011&u=635"]List[/URL][/TD]
[/TR]
[TR="bgcolor: aliceblue"]
[TD]2011[/TD]
[TD]Xeon[/TD]
[TD]E5-2660 v2(M1)[/TD]
[TD]Ivy Bridge[/TD]
[TD]2.2GHz[/TD]
[TD]BLK 100MHz[/TD]
[TD]25MB[/TD]
[TD]95W[/TD]
[TD][URL="http://www.asrock.com/support/cpu.asp?s=2011&u=649"]List[/URL][/TD]
[/TR]
[TR="bgcolor: white"]
[TD]2011[/TD]
[TD]Xeon[/TD]
[TD]E5-2650 v2(M0)[/TD]
[TD]Ivy Bridge[/TD]
[TD]2.6GHz[/TD]
[TD]BLK 100MHz[/TD]
[TD]20MB[/TD]
[TD]95W[/TD]
[TD][URL="http://www.asrock.com/support/cpu.asp?s=2011&u=637"]List[/URL][/TD]
[/TR]
[TR="bgcolor: aliceblue"]
[TD]2011[/TD]
[TD]Xeon[/TD]
[TD]E5-2650 v2(M1)[/TD]
[TD]Ivy Bridge[/TD]
[TD]2.6GHz[/TD]
[TD]BLK 100MHz[/TD]
[TD]20MB[/TD]
[TD]95W[/TD]
[TD][URL="http://www.asrock.com/support/cpu.asp?s=2011&u=658"]List[/URL][/TD]
[/TR]
[TR="bgcolor: Lavender"]
[TD]2011[/TD]
[TD]Xeon[/TD]
[TD]E5-2650L v2(M1)[/TD]
[TD]Ivy Bridge[/TD]
[TD]1.7GHz[/TD]
[TD]BLK 100MHz[/TD]
[TD]25MB[/TD]
[TD]70W[/TD]
[TD][URL="http://www.asrock.com/support/cpu.asp?s=2011&u=656"]List[/URL][/TD]
[/TR]
[TR="bgcolor: aliceblue"]
[TD]2011[/TD]
[TD]Xeon[/TD]
[TD]E5-2643 v2(M1)[/TD]
[TD]Ivy Bridge[/TD]
[TD]3.5GHz[/TD]
[TD]BLK 100MHz[/TD]
[TD]25MB[/TD]
[TD]130W[/TD]
[TD][URL="http://www.asrock.com/support/cpu.asp?s=2011&u=655"]List[/URL][/TD]
[/TR]
[TR="bgcolor: white"]
[TD]2011[/TD]
[TD]Xeon[/TD]
[TD]E5-2643 v2(M0)[/TD]
[TD]Ivy Bridge[/TD]
[TD]3.5GHz[/TD]
[TD]BLK 100MHz[/TD]
[TD]25MB[/TD]
[TD]130W[/TD]
[TD][URL="http://www.asrock.com/support/cpu.asp?s=2011&u=633"]List[/URL][/TD]
[/TR]
[TR="bgcolor: aliceblue"]
[TD]2011[/TD]
[TD]Xeon[/TD]
[TD]E5-2640 v2(M0)[/TD]
[TD]Ivy Bridge[/TD]
[TD]2GHz[/TD]
[TD]BLK 100MHz[/TD]
[TD]20MB[/TD]
[TD]95W[/TD]
[TD][URL="http://www.asrock.com/support/cpu.asp?s=2011&u=620"]List[/URL][/TD]
[/TR]
[TR="bgcolor: white"]
[TD]2011[/TD]
[TD]Xeon[/TD]
[TD]E5-2640 v2(M1)[/TD]
[TD]Ivy Bridge[/TD]
[TD]2.0GHz[/TD]
[TD]BLK 100MHz[/TD]
[TD]20MB[/TD]
[TD]95W[/TD]
[TD][URL="http://www.asrock.com/support/cpu.asp?s=2011&u=657"]List[/URL][/TD]
[/TR]
[TR="bgcolor: aliceblue"]
[TD]2011[/TD]
[TD]Xeon[/TD]
[TD]E5-2650L v2(M0)[/TD]
[TD]Ivy Bridge[/TD]
[TD]1.7GHz[/TD]
[TD]BLK 100MHz[/TD]
[TD]25MB[/TD]
[TD]70W[/TD]
[TD][URL="http://www.asrock.com/support/cpu.asp?s=2011&u=636"]List[/URL][/TD]
[/TR]
[/TABLE]

 

[Handruin]

We just requested a new quote for our new blades with Sandybridge (E5-2690) for a lab tech refresh because we couldn't wait for these E5 Ivybridge-EP chips. Not only do we have to await the release of the actual chip but also for Cisco to get them into blades and certified. By the time all that happens we've lost our quarterly budget spending. I'd love 2-socket 12-core hyperthreaded for our next gen build. When would you see the change in your new company? Would that happen 2014?

 

[Mercutio]

For you guys the biggest difference is presumably power requirements anyway. If you're short a few cores you can just migrate a system on to another host, right?

 

[Handruin]

For you guys the biggest difference is presumably power requirements anyway. If you're short a few cores you can just migrate a system on to another host, right?

Power consumption has yet to influence our purchase and planning. Density is more important due to the cost of lab space. The product we are testing consumes a lot of virtual resources so its not always easy to migrate. We test for performance so migrating during a performance test influences results so we disable features that automatically rebalance.

 

[CougTek]

When would you see the change in your new company? Would that happen 2014?

I hope we'll buy the new equipment in October this Fall. I'm doing everything I can not to give any reason to anyone in the decision making process to delay the server upgrade. I've made the studies, spec'ed the systems, contacted suppliers and written the plans and reports before anyone even thought to ask me to. They've asked me to help them to modernize their aging setup and they will, in a timely manner, weither it's their habit or not to do so. At least at the new place I work, no one complains about me actively pushing the project forward.

Regarding the processor certification, I couldn't care less when either HP or Supermicro certifies the new chips for their servers. Intel, those making the chipsets and the processors, say the Ivy Bridge-EP will be compatible with the current generation of chipsets. As far as I'm concerned, if it fits the thermal envelop, it's a go. This means up to the E5-2680v2 or E5-2695v2 for SuperMicro's blades and everything goes for HP's blades and SuperMicro's Twin² superservers (except the odd 150W TDP models, which are targeted at workstations and not servers anyway).

The way I see it, HP and other OEMs won't have a choice to certify the new processor models anyway, because Intel does. Waiting for brand X's certification is ok to cover your ass with your boss, but it has no real technical benefit. IMO, it's time wasted.

 

[CougTek]

Now the prices have been leaked too. The 12-core models are more expensive than what I thought they would be. They are worth buying anyway.

 

[Stereodude]

Yeah, I'm not so sure about that part.

 

[Mercutio]

Coug is talking about setting up some pretty massive systems for his datacenter. Given his employer's size and financial health, I suspect that it makes more sense to cram as much hardware as possible in a single physical machine as possible so that the cost of software licensing is minimized. I'm sure they'd rather pay for eight extra threads running on one physical machine than an extra physical machine that has to be licensed separately.

 

[CougTek]

Indeed. No matter if you opt to go with VSphere (~1000$ per year per socket) or Server 2012 R2 (~6000$ once per system), the cost of licensing always outweights the cost of going for premium hardware (versus targetting mid-range hardware if more licences are needed).

 

[ddrueding]

So next crazy workstation build should be a Xeon E5-2687W v2 looks like. That will be fun.

 

[LunarMist]

What is the backup for when the server craps out?

 

[ddrueding]

What is the backup for when the server craps out?

If it is that important, having two is clearly necessary. Once you have dozens, the actual number of spares on hand becomes a function of the time to get new parts vs. failure rate.

 

[Mercutio]

One nice thing about having a managed virtual environment is that you can often migrate the load on to one of the other systems in a fairly transparent fashion. Maybe that means being oversubscribed on RAM for a while or having to throttle CPU cycles per machine, but that's probably better than having a nonfunctional service in the first place.

 

[Howell]

If it's that important you should have enough hardware to have one down for maintenance and one down for failure. Whatever the bane of your infrastructure is, it will come to visit with friends when you can least afford it.

 

[Chewy509]

IVB-E Reviews (i7-4960X) are hitting the various websites...

http://www.anandtech.com/show/7255/intel-core-i7-4960x-ivy-bridge-e-review
http://hardocp.com/news/2013/09/03/intel_core_i74960x_extreme_edition_review_roundup

 

[ddrueding]

Is it time already to let this 4770 go? Looks like it...

 

[CougTek]

I'm surprised that Intel chose to launch the consumer-grade Ivy Bridge-E before the Server variant. I would have expected them to launch both simultaneously. Anyway, less than a week before the new Xeon are out.

In the meanwhile, both the peddlers we deal with (two companies) and the company's management are grabbing their asses. Our IT room was supposed to be redone prior to the servers purchase, but it still hasn't been done and no one's in a rush to push it. I've been waiting for official quotes from the peddlers for a while (with the previous generation of processors), but they keep me hanging. I thought I would be able to start the upgrading process by the end of the month, but it doesn't look good at this point.

On the other hand, if they let the project rot long enough, I might be able to use the upcoming E7-48xx v2 with 15 cores per socket instead of the 12-core E5-26xx v2. We'll see.

 

[CougTek]

Only one day left.

 

[CougTek]

In stock for sale!

 

[CougTek]

We just requested a new quote for our new blades with Sandybridge (E5-2690) for a lab tech refresh because we couldn't wait for these E5 Ivybridge-EP chips. Not only do we have to await the release of the actual chip but also for Cisco to get them into blades and certified. By the time all that happens we've lost our quarterly budget spending. I'd love 2-socket 12-core hyperthreaded for our next gen build.

Cisco and HP have already certified the new Xeon E5 v2. I know for sure about HP because I'll have a quote with the new proc. tomorrow and regarding Cisco, this article states :

Cisco upgraded its Unified Computing System (UCS) lineup with the new Intel chips.

It's important to realize two things :

1. The server market is flat and has been for a while.
2. The manufacturer have the chips for testing since at least June.

There's a tremendous pressure to introduce the new processors ASAP in order to stimulate sales. They have every reason and no excuse to delay them. I'll have HP's part# for the E5-2697v2 tomorrow and I can look for Cisco's too if you want. The quarter isn't over yet so maybe your company still has some budget left.

 

[Handruin]

Cisco and HP have already certified the new Xeon E5 v2. I know for sure about HP because I'll have a quote with the new proc. tomorrow and regarding Cisco, this article states :


It's important to realize two things :

1. The server market is flat and has been for a while.
2. The manufacturer have the chips for testing since at least June.

There's a tremendous pressure to introduce the new processors ASAP in order to stimulate sales. They have every reason and no excuse to delay them. I'll have HP's part# for the E5-2697v2 tomorrow and I can look for Cisco's too if you want. The quarter isn't over yet so maybe your company still has some budget left.

This may work out good. Our initial plan for new hardware acquisition request got rejected back when I wrote that post a couple months back...so we got zero new hardware. They now tell us late Q1 or Q2 next year for new equipment purchases. This should give plenty of time for these new CPUs to saturate the market and make it viable for us to get some assuming the pricing is ok for those who budget for us. I appreciate your offer to look up the parts but I think I'll have to pass for now until I know more about when we can get new hardware.

 

[CougTek]

HP's part numbers for the BL460c Gen8 blade server are 718045-L21 and 718045-B21. For the DL380p (2U, non-blade), those are 715224-B21 and 715224-L21. I haven't found those for Cisco yet.

 

[Handruin]

HP's part numbers for the BL460c Gen8 blade server are 718045-L21 and 718045-B21. For the DL380p (2U, non-blade), those are 715224-B21 and 715224-L21. I haven't found those for Cisco yet.

Don't worry about finding the part number for now for Cisco unless it helps you. Did you make the final decision to go ahead with the HP BL460c Gen 8 blades based on your other thread?

We have 4 of the BL460c Gen 8's and they've been good for us so far. I also just noticed a power supply went out...time to go figure out what happened.

 

[CougTek]

No decision yet, but it should be done by this time next week. Both suppliers we've contacted don't give us what I asked them to, which is something that pisses me off. I want a c3000 with D2220sb DAS and BL460c blades. One sent a quote with a c7000 (it would be complicated for us to supply 240V to the servers) and a 3PAR SAN. The other went with a bunch of DL380p 2U servers and an MSA2040 SAN. Again, not what I asked.

The 3PAR is nice, but expensive as hell. As fast as it can be, it won't beat a D2220sb filled with SSDs. The 3PAR software is nice, but my collegue found a replication software made for VM and database (much like Veeam) that would achieve the same thing, but for so much less.

As for the DL380p, I've already worked with those and they are nothing special. If you can get blades, you don't want regular Proliant. The management, like you know and showed in the picture you've posted, is so much simpler. The cables connection also. I want everything inside a 6U chassis locally and a similar setup (with half the nodes) at our remote site. Simpler, faster and about the same price.

It doesn't look good so far for me to get what I want and some people are about to meet my temper. I'll be the one configuring it. I'll be the one maintaining it. Give me what I want or look elsewhere for someone to blame if a fuck happen afterwards.

 

[Chewy509]

First round of benchmarks:
http://www.anandtech.com/show/7285/intel-xeon-e5-2600-v2-12-core-ivy-bridge-ep

 

[CougTek]

I saw the review. I expected the advantage to be around 40%, not 30%. It's not bad, but it's less than I thought. Worth going for it over the old generation anyway, of course.

Unrelated to the review, but I looked at Intel's ARK specification sheet and I simply don't understand why the E5-2648Lv2 is numbered as a lower-end product than the E5-2650Lv2. Same TDP, same amount of cores, higher operating frequency and higher QPI speed. Same price too. I don't get it.

 

[ddrueding]

Hopefully I'll have time to fool with this today:

 

[Handruin]

Hopefully I'll have time to fool with this today:

View attachment 660

That should be fun. I have been debating going down the path of the Ivy Bridge-E to replace my Core i7 860 (Lynnfield). After reading up on it a bit I'm a little torn with going with the 4930K CPU mainly because of the slightly dated X79 chipset which is lacking a few things that I could get by going with the Haswell i7-4770K with a Z87 chipset. I don't think the added cost is worth it for me to get 2 more cores when I'd be taking a step back on USB, PCIe, etc.

 

[ddrueding]

I don't think the architecture difference is going to matter that much, at least in my application. I will be sticking a pair of titans onto it, at least as a test, just to see what's up.

 

[Mercutio]

Let's be honest: unless running 3DMark or Cinebench gives you an erection, the real-world difference between having the extra features of Z87 over X79 probably amounts to a couple extra native USB3 ports.

 

[P5-133XL]

I contend that the real difference between z87 (32GB) and x79 (128GB) is the amount of RAM allowed with x79 being the winner. All the other differences put together pale in significance.

 

[Mercutio]

Maybe. But honest, if you have a working set big enough to need more than 100GB RAM, you're not going to be working on flaky enthusiast hardware.

 

[P5-133XL]

Maybe. But honest, if you have a working set big enough to need more than 100GB RAM, you're not going to be working on flaky enthusiast hardware.

Are you sure you don't mean 32GB+ of ram? Otherwise the 32GB-100GB range is just missing in action. There are lots of enthusiasts that want more than 32GB even if they never really use it. I will agree that 100GB+ is undoubtedly multi-processor server territory especially considering how much 16GB simms cost.

 

[Mercutio]

Sure. Either way. About the time a pair of 16GB DIMMs starts to sound like a good idea, you should probably stop and re-consider the life choices that go along with a $400 motherboard that has $.80 worth of extra heat sinks and a dragon screen-printed on the PCB to distinguish it from the $250 version.

 

[Chewy509]

I contend that the real difference between z87 (32GB) and x79 (128GB) is the amount of RAM allowed with x79 being the winner. All the other differences put together pale in significance.

<nitpick>I would point out that the chipset has nothing to do with maximum RAM, as the memory controller is embedded in the CPU.... It's the IMC that defines the memory setup not the chipset...</nitpick>

 

[Stereodude]

Let's be honest: unless running 3DMark or Cinebench gives you an erection, the real-world difference between having the extra features of Z87 over X79 probably amounts to a couple extra native USB3 ports.

I would expect the CPU in the X79 to have an advantage to have an advantage over the the CPU in the Z87 in Cinebench. No idea on 3DMark.

Maybe. But honest, if you have a working set big enough to need more than 100GB RAM, you're not going to be working on flaky enthusiast hardware.

No, you'll be working on super picky enterprise hardware.

 

[Mercutio]

No, you'll be working on super picky enterprise hardware.

Yes, and that hardware will be like Old Yeller needing a bullet to be put down BUT ANYWAY this thread is about Xeon E5s; they generally don't hand those out with a free cold cathode lamp and subscription to Maximum PC.

 

[ddrueding]

Hopefully I'll have time to fool with this today:

View attachment 660

A week later and I've actually had a chance to throw something together. I have a sinking suspicion that this motherboard doesn't have a BIOS that supports this chip. Machine powers up for about a second (fans spin up) then shuts down for about two seconds, and repeats forever. Tried different RAM, GPU, PSU. Disconnected everything else. Don't have a spare 2011 CPU to test with. Damn.

 

[Mercutio]

The last Gigabyte board I stuck an unsupported CPU on gave the same beep code as unseated RAM.

 

[ddrueding]

Just ordered the cheapest 2011 CPU and motherboard I could find. I'll do some swapping on Monday.