By: Ricardo B (ricardo.b.delete@this.xxxxxx.xx), February 2, 2013 4:05 pm
Room: Moderated Discussions
Richard Cownie (tich.delete@this.pobox.com) on February 2, 2013 2:50 pm wrote:
> I don't think I'm underestimating it. But they can only do that tight
> integration of process/circuit/microarchitecture for one thing, which at the
> moment is x86. The right process for x86 laptop/desktop is not necessarily
> the right process for smartphone/tablet. And even over the last few years,
> when they try to push a mediocre design, they don't do so well: Atom-based
> low-power systems haven't matched up well against AMD's Bobcat-based products,
> and aren't selling well these days. And Itanium ...
No, they can do that integration for anything they can develop and fabricate in house, provided it generates enough revenue to sustain the R&D.
Ie, they can't do that integration for a ARM core from ARM, but they can do it if they develop their own.
> anon's argument is a bit circular: when Intel has a high-volume product
> on their world-beating fabs, they do great. But of course, the products
> that don't reach high volume are the ones with weak design. And Intel has
> had plenty of those (several generations of Itanium; original Pentium4; Atom).
> Can they design a great smartphone/tablet SoC ? It's not impossible; but
> I wouldn't put the odds any higher than 50%, based on the complexity of the
> challenge, and their track record. You're welcome to make a different judgment.
Your argument is a bit circular as well. :)
You keep focusing on how Intel has f**ked up. Yes, Intel does a lot of stuff and they do f**k up now and then. Most companies do eventually.
The Atom design was aimed at the netbook market, which was a flop.
And the (current) Atom design is too wimpy even for budget PCs, so it can't sell there either (unlike Bobcat).
But even with that Atom design, they've shown they can produce a smartphone SoC with good performance and power usage.
Itanium was another f**k up, but it was smashed mostly by Intel's own x86.
But Intel's ability to f**k up does not show they can't compete outside x86.
>
> > When the smartphone/tablet rose, it met neither of these requirements. What we have
> > was companies like Apple design their own SoCs for use in their own devices.
> > But now, as we have lots of companies buying large quantities of of-the-shelf SoC chips
> > to use in their devices, it begins to meet Intel's business model requirements.
>
> Not really. It seems to me there are two classes of smartphone/tablet product -
> the premium high-end high-margin products, mostly from Apple and Samsung; and the
> commodity products with so-so specs and razor-thin margins. There isn't an opportunity
> for Intel in the high-end products - at least for now - because Apple and Samsung
> want that margin for themselves; and while there might be an opportunity in the
> other segment, it's unlike to have the kind of margins that Intel wants to get
> from its leading-edge fabs.
NVIDIA and Qualcomm beg to differ.
Samsung does not own the high end non-iPhone market.
Besides Samsung, you have half a dozen companies competing to produce high end smartphones/tablets and they'll buy their SoCs from whoever can deliver.
> I don't think I'm underestimating it. But they can only do that tight
> integration of process/circuit/microarchitecture for one thing, which at the
> moment is x86. The right process for x86 laptop/desktop is not necessarily
> the right process for smartphone/tablet. And even over the last few years,
> when they try to push a mediocre design, they don't do so well: Atom-based
> low-power systems haven't matched up well against AMD's Bobcat-based products,
> and aren't selling well these days. And Itanium ...
No, they can do that integration for anything they can develop and fabricate in house, provided it generates enough revenue to sustain the R&D.
Ie, they can't do that integration for a ARM core from ARM, but they can do it if they develop their own.
> anon's argument is a bit circular: when Intel has a high-volume product
> on their world-beating fabs, they do great. But of course, the products
> that don't reach high volume are the ones with weak design. And Intel has
> had plenty of those (several generations of Itanium; original Pentium4; Atom).
> Can they design a great smartphone/tablet SoC ? It's not impossible; but
> I wouldn't put the odds any higher than 50%, based on the complexity of the
> challenge, and their track record. You're welcome to make a different judgment.
Your argument is a bit circular as well. :)
You keep focusing on how Intel has f**ked up. Yes, Intel does a lot of stuff and they do f**k up now and then. Most companies do eventually.
The Atom design was aimed at the netbook market, which was a flop.
And the (current) Atom design is too wimpy even for budget PCs, so it can't sell there either (unlike Bobcat).
But even with that Atom design, they've shown they can produce a smartphone SoC with good performance and power usage.
Itanium was another f**k up, but it was smashed mostly by Intel's own x86.
But Intel's ability to f**k up does not show they can't compete outside x86.
>
> > When the smartphone/tablet rose, it met neither of these requirements. What we have
> > was companies like Apple design their own SoCs for use in their own devices.
> > But now, as we have lots of companies buying large quantities of of-the-shelf SoC chips
> > to use in their devices, it begins to meet Intel's business model requirements.
>
> Not really. It seems to me there are two classes of smartphone/tablet product -
> the premium high-end high-margin products, mostly from Apple and Samsung; and the
> commodity products with so-so specs and razor-thin margins. There isn't an opportunity
> for Intel in the high-end products - at least for now - because Apple and Samsung
> want that margin for themselves; and while there might be an opportunity in the
> other segment, it's unlike to have the kind of margins that Intel wants to get
> from its leading-edge fabs.
NVIDIA and Qualcomm beg to differ.
Samsung does not own the high end non-iPhone market.
Besides Samsung, you have half a dozen companies competing to produce high end smartphones/tablets and they'll buy their SoCs from whoever can deliver.