So is IB-E as good as cancelled then or... what?

[Maximilian]

I read it was delayed to 2h 2013, by that time haswell will have been out, possibly for quite a few months. Will intel bother making an IB-E at all or just jump straight to a haswell-E?

 

[EltonL]

It really depends on the demand from the server markets. The thing is, IB-E would be a good stopgap, but it's clear that Haswell-E won't be using LGA2011 so it wouldn't be a very viable platform. That said, IB-E will at least offer IPC and power consumption improvements.That and a 8-core Intel chip perhaps.

But will IB-E actually be released? Seems like no if it overlaps with Haswell. It probably would though owing to their tick/tock strategy. At the very least they can milk the X79 a bit. Perhaps a revision?

 

[pantsaregood]

Ivy Bridge-E has never seen an official announcement to my knowledge. Ivy Bridge-EX/EP have been announced and engineering samples exist. Intel could always cancel them, but I doubt that will happen.

The existence of Ivy Bridge-E is dependent on the demand for rebranded Ivy Bridge-EX/EP Xeons, honestly.

 

[ShintaiDK]

There is no IB-E and no ES samples either. Next step is Haswell-E.

 

[pantsaregood]

http://www.cpu-world.com/news_2012/2012072301_Some_details_of_Ivy_Bridge-EP_and_-EN_Xeon_CPUs.html
http://news.softpedia.com/news/10-Core-Intel-Ivy-Bridge-EP-CPU-Tested-Has-20-Threads-250492.shtml

I really doubt Intel is going to throw the server market to the wolves like that.

 

[ShintaiDK]

http://forums.anandtech.com/showthread.php?t=2259585

The cpu-world.com article is an utter joke.

And even if we played along. The IB-E release then would be a short release before Haswell-E.

 

[pantsaregood]

Intel also generally doesn't seem to care about conflicting launches.

Westmere units were being launched until quite some time after Sandy Bridge hit.

 

[Arachnotronic]

Not cancelled...yeesh:

http://www.linkedin.com/in/terryu

"Working in a logic verification group on the Ivytown project"

http://www.linkedin.com/pub/jimmy-limqueco/a/20b/2b4

"Have been working as a verification to multiple Intel NB, and IOH projects. Current responsibility is to perform verification duties on Intel's Ivytown's Integrated I/O's DFx features such as error injection, performance counters, PCIE LTSSM FPLA and others. Presently still provide consultation level of support for Intel's Crystal Beach pre-Si verification for version 3.3"

I don't think Intel's going to spend all this time & money having people develop IVB-E just to "cancel" it.

 

[Rvenger]

Not cancelled...yeesh:

http://www.linkedin.com/in/terryu

"Working in a logic verification group on the Ivytown project"

http://www.linkedin.com/pub/jimmy-limqueco/a/20b/2b4

"Have been working as a verification to multiple Intel NB, and IOH projects. Current responsibility is to perform verification duties on Intel's Ivytown's Integrated I/O's DFx features such as error injection, performance counters, PCIE LTSSM FPLA and others. Presently still provide consultation level of support for Intel's Crystal Beach pre-Si verification for version 3.3"

I don't think Intel's going to spend all this time & money having people develop IVB-E just to "cancel" it.

Yeah, they will release IVB-E right before mainstream Haswell. We probably won't see Haswell-E until Q4 2013 or even Q1 2014.

 

[Arachnotronic]

Yup. The -E chips are meant for servers, so it takes a lot more time to validate them. It isn't "lack of competition" or "Intel being greedy", it's just these things take time.

 

[tynopik]

Yup. The -E chips are meant for servers, so it takes a lot more time to validate them. It isn't "lack of competition" or "Intel being greedy", it's just these things take time.

which of course explains why socket 1366 came first

 

[exar333]

Not sure why this is so surprising. SB-E was released only 4 months before IB was originally scheduled (before it was delayed a couple months). It's the new pattern so the newest mainstream chips are available for mobile markets, and the design can be refined and validated for serves.

 

[tynopik]

Not sure why this is so surprising. SB-E was released only 4 months before IB was originally scheduled (before it was delayed a couple months). It's the new pattern so the newest mainstream chips are available for mobile markets, and the design can be refined and validated for serves.

but that was never the plan

SB-E was massively delayed by problems they found

I fail to see why a huge screw-up should now be used to justify this as 'the new norm'

 

[Edrick]

but it's clear that Haswell-E won't be using LGA2011 so it wouldn't be a very viable platform.

Proof?

 

[Edrick]

There is no IB-E and no ES samples either. Next step is Haswell-E.

Again, Proof?

So many people make bold claims lately without any links or documentation to back it up.

 

[tynopik]

Proof?

for certain definitions of 'proof'

http://semiaccurate.com/2012/07/09/haswell-ep-to-use-the-same-socket-just-totally-different/

 

[Maximilian]

Sort of like socket 940 and socket AM2. Same number of pins but worlds apart.

 

[ShintaiDK]

Again, Proof?

So many people make bold claims lately without any links or documentation to back it up.

IB-E aint on the latest Intel roadmap for partners. And yes, the roadmap have been posted several times. Just look for it.

 

[Sweepr]

IB-E aint on the latest Intel roadmap for partners. And yes, the roadmap have been posted several times. Just look for it.

Current Intel roadmaps dont show products that will be launched H2/2013 or after. An IB-E launch after Haswell-DT could happen. Unlikely but possible. After all, Haswell-E is not coming until 2014, Romsley platform was launched in March and the desktop SB-E will EOL without competition.

 

[Arachnotronic]

IB-E aint on the latest Intel roadmap for partners. And yes, the roadmap have been posted several times. Just look for it.

That greater-than-or-equals-to symbol next to 3960X tells you what you need to know.

 

[tynopik]

That greater-than-or-equals-to symbol next to 3960X tells you what you need to know.

That Intel still has no clue what they're going to do?

 

[ShintaiDK]

That greater-than-or-equals-to symbol next to 3960X tells you what you need to know.

Yes, that we might get a 3970X speedbin update.

 

[BenchPress]

Ivy Bridge-E doesn't make any sense. Quad-core Haswell will outperform it at any workload consumers could possibly care about. More cores is useful for servers since they run many services, but enthusiast consumers are better off with higher throughput per core instead.

So I don't expect Intel to launch a product for which there won't be any informed demand. It would just confuse people and lead to unhappy customers who wished they went with Haswell instead.

 

[fixbsod]

think i already saw confirm of 3970x on vr-zone or legitreviews. however all i saw was a 100mhz bump and then a bump of tdp as well????

Yes, that we might get a 3970X speedbin update.

 

[cytg111]

CrystalBallMoment(tm) : Indeed we will see IB-E, and thus we shall see haswell delayed. Delayed at least 6 months to make room for much appreciated Coin in intels coffers.

Atleast, show me the money plan that makes this NOT make sense! You can trust one thing and one thing only : $$$$$$.