Intel delays Sofia LTE (again)

NTMBK

Lifer
Nov 14, 2011
10,207
4,939
136

Idontcare

Elite Member
Oct 10, 1999
21,118
58
91
Damn, but I have to agree.

I'm curious to know if all these wheels coming off the bus at 14nm are attributable to things being setup to go south under Otellini's watch (mismanagement and he got out while the getting out was good), or if this is just a glimpse of what lay in store under BK's watch.
 

podspi

Golden Member
Jan 11, 2011
1,965
71
91
Regardless, 14nm is turning out to be a pretty terrible node for Intel, execution-wise.

Every once in a while I realize how old Haswell really is - my 4670K hasn't been outdated for so long... and still isn't!
 

Mondozei

Golden Member
Jul 7, 2013
1,043
41
86
^I think the influence of each CEO over large technical issues that require thousands and thousands of engineers are quite minimal.

But as to the article, it just surprises me more generally that it's still so hard for large companies to get a decent built-in LTE solution. I could understand it in 2012 or even 2013 but in 2015, this should have been solved a long time ago. People have been working on this for 5 years and not just Intel, I'm looking at all the non-Qualcomm players.

We've seen some improvement but just some.
The big player that looks set to challenge Qualcomm is Mediatek. No longer just doing value-oriented stuff. (Even if their addiction to MOAR COARES is problematic to say the least).
 
Aug 11, 2008
10,451
642
126
If you read the article carefully it doesnt say definitely that this is another delay. It poses the question whether it is or not, but the conclusion is somewhat vague. The main concern seems to be that sophia 2 will be delayed, but that is not definite.
 

sm625

Diamond Member
May 6, 2011
8,172
137
106
Yeah but when every mobile product you release costs you money, it makes perfect financial sense to delay. Their stock price probably goes up every time they delay.
 

NTMBK

Lifer
Nov 14, 2011
10,207
4,939
136
Yeah but when every mobile product you release costs you money, it makes perfect financial sense to delay. Their stock price probably goes up every time they delay.

The idea was that if they get 14nm Sofia out in time, they won't have to lose money on it any more. If they had kept to the original timeline then they would have a very competitive low cost product on the market when they get to that stage (14nm Atom vs. 28nm Mediatek stuff). Sadly, their roadmap keeps on sliding :(
 

liahos1

Senior member
Aug 28, 2013
573
45
91
Intel to launch quad-core SoFIA LTE chips in 4Q15, say Taiwan makers
Monica Chen, Taipei; Steve Shen, DIGITIMES [Wednesday 22 April 2015]
Intel will move ahead the launch of its 4G chip, the SoFIA LTE, to the fourth quarter of 2015 and the SoFIA LTE 2 in the second quarter of 2016, according to industry sources in Taiwan.

The SoFIA LTE, which integrates a 4G baseband and quad-core Atom CPU, comes after Intel and its partner Rockchip Electronics jointly released a 3G solution, the SoFIA 3G-R, recently.

To help mobile device makers shorten time to market, Intel will have its China-based ODM partner Wanlida and Taiwan's Elitegroup Computer Systems (ECS) offer corresponding reference designs for 6-inch models, the sources indicated.

Based on Intel's roadmap, the chip giant also plans to launch the 14nm quad-core SoFIA LTE 2 in the second quarter of 2016, supporting both Windows and Android OS. Furthermore, Intel also plans to release its 8-core solution, the SoFIA MID T3, in August 2016, the sources noted.

8 core mobile solution?????
 

Phynaz

Lifer
Mar 13, 2006
10,140
819
126
I said it before, Otellini was genius in picking his time of departure.

By the time Sofia is out it will be irrelevant. I think Intel needs to take that stock it's been buying back and just acquire what it needs (companies). Their NIH attitude isn't cutting it in today's market.
 

krumme

Diamond Member
Oct 9, 2009
5,952
1,585
136
I have been saying for at least 2 years Intels mobile strategy would fall flat and i have taken so much beating here for it here.

But man before we call out Otellini and what not cant the delay be as simple as eg some dopants not behaving as expected (you get the picture), 14nm yield suffer, and therefor mobile is prioritized down?

Even with sofia in hand i still dont know how they would sell it to eg apple when they have a9 or eg samsung. Or compete with qcom that is nothing like amd. Intel is historically good at what they do best - look at dcg profit where they protect an existing position - and nearly constantly fail at new product areas/ new tech. Trying to compete with apple with that is a bad start. Its as insane as amd trying to beat intel at dense servers. Its nearly unbearable to watch.
 

AllDayBreakfast

Junior Member
Feb 25, 2015
24
0
0
Broxton has been 2016 since Intel's Investor Meeting in Nov 14 (see Renee James' presentation slides).

SoFIA 3G was 1H2015, per the slides, and Rock chip announced 3G-R was shipping in March (see IDF Shenzhen preso) and is on sale this month (http://www.rock-chips.com/a/en/News/Rockchip_in_the_News/2015/0414/639.html). Intel also said they've started shipping this (see latest earnings call).

SoFIA LTE was "starting mid-2015", we aren't there yet.
 

Phynaz

Lifer
Mar 13, 2006
10,140
819
126
Intel is historically good at what they do best - look at dcg profit where they protect an existing position - and nearly constantly fail at new product areas/ new tech.

Good observation.

I remember all the stuff Intel tried getting into and then sold off - mostly because the margins weren't there. Set-top boxes, television, automotive, etc.
 

krumme

Diamond Member
Oct 9, 2009
5,952
1,585
136
Good observation.

I remember all the stuff Intel tried getting into and then sold off - mostly because the margins weren't there. Set-top boxes, television, automotive, etc.

As i see it Intel is pretty much aware of this problem. To its credit it always try. As idc says not trying is a risk in itself in this business.

But when you have a big corp that is especially tuned in organization and culture for tecnology development in production its damn hard to make true innovation on the market. Processes takes to long time and cost for everything is exorbitant. Even calculating cost for your exorbitant expensive project is very expensive. :)

I would try to adress that problem more fundamentally before trying more inventions that fail.
 
Last edited:

DrMrLordX

Lifer
Apr 27, 2000
21,570
10,762
136
So what this means in terms of product availability is that when Cherry Trail x5 and x7 hit the market, the x3 Intel launches alongside them will be SoFIA 3G?
 

Nothingness

Platinum Member
Jul 3, 2013
2,364
705
136
Wonder what's taking so long? It's been 4 years since they bought Infineon. Does it take at least that long to make an x86 modem?
It's possible Infineon modem was far from being ready.

OTOH wasting time and money switching to an x86 core sounds silly. As Phynaz wrote, they should stop with their NIH syndrome.
 
Last edited:
Mar 10, 2006
11,715
2,012
126
It's possible Infineon modem was far from being ready.

OTOH wasting time and money switching to an x86 core sounds silly. As Pynaz wrote, they should stop with their NIH syndrome.

Already done in SoFIA, though. Shouldn't be hard to run the modem firmware on a low-clocked Airmont in XMM 7460, especially if they're able to do it in SoFIA.
 

Dave2150

Senior member
Jan 20, 2015
639
178
116
Regardless, 14nm is turning out to be a pretty terrible node for Intel, execution-wise.

Every once in a while I realize how old Haswell really is - my 4670K hasn't been outdated for so long... and still isn't!

x99 says hello. Significantly faster than your i5 4670k.
 

erunion

Senior member
Jan 20, 2013
765
0
0
It's possible Infineon modem was far from being ready.

OTOH wasting time and money switching to an x86 core sounds silly. As Phynaz wrote, they should stop with their NIH syndrome.

As far as I know, intels plan with infineon was to keep to their original product roadmap until moving modems to Intel fab at 14nm. Porting ARM cores to Intel processes just to power their modems would be silly at that point, so switching to x86 was inevitable.

Sofia was added to the roadmap recently to accelerate to adoption of x86 in modems while still on TSMC. So you could say man-hours were wasted, but the far more valuable TTM was saved.

Using Atoms cores also radically changed the capability of the product. The modems Sofia replaces were sold into feature phones(and pseudosmartphones like the Nokia Asha) that used the modem's weak arm core to also run the OS/apps. Sofia on the other hand will power android and full windows devices.
 
Last edited:

mrmt

Diamond Member
Aug 18, 2012
3,974
0
76
Damn, but I have to agree.

I'm curious to know if all these wheels coming off the bus at 14nm are attributable to things being setup to go south under Otellini's watch (mismanagement and he got out while the getting out was good), or if this is just a glimpse of what lay in store under BK's watch.

I think BK would have had plenty of time to fix things up if they were brewed under Otelini. It seems that Intel corporate structure wasn't really tuned to the challenges the company faced.