Intel Developer Forum (IDF) 2013 on September 10-12, what can we expect?

[Fjodor2001]

So IDF 2013 (Intel Developer Forum) is happening on 10-12 September. There's some info on that website. But what do you think we can actually expect?

* Will there be details about Broadwell?
-Will it be BGA only?
-Will it have some eDRAM solution similar to Iris/GT3e/HD5200 on Haswell? Will it be on-die, larger size, or what improvements can we expect?
-Will we get any details about the new Gen 8 IGPU?

* Will there be info on Skylake?
-Do Intel usually present any info on CPUs 2 generations ahead?
-Perhaps only briefly?

* Will there be info on Skymont?
-Will this be the focus of IDF, since Intel seems to have been prioritizing low power mobile CPUs nowadays?

* Any other interesting topics that are likely to be presented? In that case what?

Please let me know what you think.

 

[MisterMac]

Im hoping on some Airmont Info.

And that HW-E will have some extra perf up it's sleeve somehow.

 

[Sweepr]

I'm expecting lots of Bay Trail and Haswell-Y/U based products + details on: Airmont and Skymont, Intel Gen 8 Graphics Architecture, Broadwell and Skylake.

 

[mikk]

There are no specific Broadwell, Skylake, Gen8 sessions announced. It's too far away so no surprise that we won't get detailed infos. This IDF is mostly about Bay Trail and Haswell.

If we have luck we get some Broadwell infos from this session:
https://intel.activeevents.com/sf13/connect/sessionDetail.ww?SESSION_ID=1199

 

[Blandge]

Rackscale architecture will probably be big at IDF 2013.

 

[Phynaz]

IDF 2013 will be about power consumption, power consumption and power consumption. Time permitting they will throw in something about power consumption.

 

[SiliconWars]

Surely there has to be at least one and possibly multiple sources of 14nm on show. It's unheard of for Intel to be so close to a new node and never having shown it by this very late stage.

 

[jpiniero]

I could see them talking about AVX3.

 

[Khato]

Surely there has to be at least one and possibly multiple sources of 14nm on show. It's unheard of for Intel to be so close to a new node and never having shown it by this very late stage.

Yup, it really is quite peculiar... but it also makes sense. How much competitive advantage did Intel lose by telling all the foundries that finfets were the way to go a year before they had any need to do so? Sure all of the players have their own pathfinding efforts going on, but there's a marked difference between lab feasible research and mass production - with Intel's announcement finfets went from one of many possibilities to being something that they knew could be done.

The above is why I actually wouldn't be surprised if Intel doesn't go beyond vague information with respect to their 14nm process... even though I'd quite enjoy it if they give some details.

Anyway, my guess for IDF is just a lot more information on the mobile side of things - a lot of Baytrail, a bit of Airmont, a bit of Merrifield, maybe some LTE modem information as well. Thene there will likely be a decent amount on the server side between Avoton and the forthcoming IVB server parts. I'm not expecting a whole lot of information on the client side - it's possible that they'll say a bit about Broadwell, but if they do I wouldn't be surprised if it's with respect to lowering the power envelope to push it into the premium tablet space.

 

[tweakboy]

Macys all the way. Broadway blows.

 

[Idontcare]

Yup, it really is quite peculiar... but it also makes sense. How much competitive advantage did Intel lose by telling all the foundries that finfets were the way to go a year before they had any need to do so? Sure all of the players have their own pathfinding efforts going on, but there's a marked difference between lab feasible research and mass production - with Intel's announcement finfets went from one of many possibilities to being something that they knew could be done.

The finfet PR talk was done to perform a shock-and-awe campaign on the decision makers at OEMs who were (at the time) in the stages of making decisions regarding what MPU their 2014 and 2015 products would be built around.

Intel had to convince decision makers that the risk of going x86 instead of ARM wasn't going to be assured career suicide for those folks at the customer who were tasked with making those critical decisions. So you do what is necessary to make it a justifiable decision by making sure everyone thinks "well sure its x86 with its overhead, but they'll be on those fancy power-sipping 3D xtors whereas the ARM SoCs will be on bog-standard planar HKMG".

So from a risk-management standpoint Intel created the requirement that decision makers not simply dismiss them out of hand, it became risky (to one's personal career) to cross-off Intel on the short-list because if Intel's finfet stuff really was good stuff then your own company's products would be competing against it (if your competitors opted to endorse the Intel SoCs) on the marketplace.

It was a calculated, but necessary, trade-off for Intel to ensure they weren't dismissed out-of-hand for the 22/20nm cycle of handset form factor products.

Now that they have the people's attention, what needs to be divulged about 14nm will be done so behind closed doors without the full-on public PR campaign. (that will come later, once the OEMs are launching their 14nm-based Intel products to market, and Intel is subsidizing their marketing campaigns to make the consumer aware of how superior the products are compared to Samsung or foundry-based 16nm/14XM)

 

[TechFan1]

I predict stacked chips.

 

[Fjodor2001]

Surely there has to be at least one and possibly multiple sources of 14nm on show. It's unheard of for Intel to be so close to a new node and never having shown it by this very late stage.

I agree. Intel has been awfully quiet about 14 nm and Broadwell. Most info has come from non-official sources and rumors.

At last year's IDF 2012 there was lots of info on Haswell. See for example:

http://www.anandtech.com/show/6264/intel-haswell-architecture-slides-idf-2012

http://www.anandtech.com/show/6248/haswell-at-idf-2012-10w-is-the-new-17w

http://www.anandtech.com/show/6266/idf-2012-haswell-gt3-running-skyrim

http://www.anandtech.com/show/6263/intel-haswell-architecture-disclosure-live-blog

If Intel will not present info on 14 nm and Broadwell at IDF 2013, how should we interpret that? Is it delayed, or does Intel intentionally keep it secret for some reason?

 

[liahos1]

are any sites live blogging IDF keynotes?

 

[makken]

Is anyone even covering this event? Seems apple is all people care about today.

 

[Khato]

Only coverage of the keynote this morning I found - http://www.pcper.com/news/Processors/Intel-Developer-Forum-IDF-2013-Keynote-Live-Blog Which is annoying since it's in a format that allows those viewing to comment, bleh.

 

[makken]

Thanks, better than nothing.

 

[makken]

Interesting. Quark SoC @ 1/5 the size of atom and 1/10 the power. Destined for smart watches

 

[liahos1]

100 dollar and less tablets with atom inside for holiday season

 

[makken]

100 dollar and less tablets with atom inside for holiday season

I wonder who is manufacturing those...

 

[liahos1]

I wonder who is manufacturing those...

amazon D:

 

[Erenhardt]

DDR4 for servers from ADATA:
http://www.adata-group.com/index.ph...ist&pages=1&news_id=428&ny=2013&type=2&lan=en

 

[tential]

Not going to lie, I'll probably have an orgasm by the end of this IDF. CPUs are finally doing something interesting again.

 

[iMacmatician]

Interesting. Quark SoC @ 1/5 the size of atom and 1/10 the power. Destined for smart watches

Is Quark x86?

 

[erunion]

Is Quark x86?

Unknown so far. Intel said its an "open" architecture.