By: Dummond D. Slow (mental.delete@this.protozoa.us), February 22, 2021 5:55 pm
Room: Moderated Discussions
Anon (no.delete@this.spam.com) on February 22, 2021 3:25 pm wrote:
> Dummond D. Slow (mental.delete@this.protozoa.us) on February 22, 2021 1:34 pm wrote:
> > I don't disagree on the principle but hurdles for getting into phone market are
> > so high it makes almost no sense to try to butt in forcefully - for AMD.
> > *Especially* for a company like AMD that usually has its own new initiatives shot-down
> > and ignored by the wider industry and such attempts almost never work for them. It's
> > as if they had some "not a cool kid don't invest in" paper glued on their back :)
>
> AMD is already on the mobile market, Samsung licensed AMD's GPU :) I think, if AMD had
> a working ARM core design they would be able to license the core to third party.
>
> Hey, look at what Samsung is calling "high end phone" it is essentially a joke compared to iPhone 12,
> high end Androids are in a much worse position than AMD was compared to Intel during Bulldozer era.
>
For a phone, almost everything else matters more and is a bigger barrier to entry compared to the CPU core AMD could tout as their advantage. Only GPU has lesser importance. These two items are not worth much - what you have to bring to the table is cellular connectivity (with allt he regulatory stuff handled!), wireless, power management, the low power platform stuff and software ecosystem. These days, also sensors, ISP, AI unit.
The assets AMD might have had to offer with K12 were those that are easiest to solve by licensing off-the-shelf ARM Holdings IP. So basically, they had nothing. GPU is something you can license but not use as a base for selling our SoC.
I doubt it was viable to make money by licensing K12 as a Cortex laternative. Even if AMD were able to pull that off and ARM would not push them out, the revenue stream would be small and not really worth AMD's time.
> > Look at how the processor debates go in the vein of "dumb old Intel needed Apple to wake them
> > up else the CPU market would go nowhere", as if AMD's influence and market presence (which
> > should not be overblown but is still significant) didn't exist at all or something.
> >
> > Bascially the problem with this topic is: ARM being a good idea doesn't mean it would have worked for AMD
> > Odds were decent that K12 would just be another waste of money a la Seattle/SeaMicro.
> > There is also the kicker that AMD's help with trailblazing server platform support for ARM
> > eventually only helped competitors that didn't have "not a cool kid" sticker on their back,
> > so it was a really dangerous mistake from them. You don't create a new market that kills your
> > old market, if only your competitors will profit while you paid the costs for them.
>
> Pushing ARM for server was dumb, for sure, but the mobile market is already
> consolidated, it is just the high end supplier position which is vacant.
>
> Dummond D. Slow (mental.delete@this.protozoa.us) on February 22, 2021 1:34 pm wrote:
> > I don't disagree on the principle but hurdles for getting into phone market are
> > so high it makes almost no sense to try to butt in forcefully - for AMD.
> > *Especially* for a company like AMD that usually has its own new initiatives shot-down
> > and ignored by the wider industry and such attempts almost never work for them. It's
> > as if they had some "not a cool kid don't invest in" paper glued on their back :)
>
> AMD is already on the mobile market, Samsung licensed AMD's GPU :) I think, if AMD had
> a working ARM core design they would be able to license the core to third party.
>
> Hey, look at what Samsung is calling "high end phone" it is essentially a joke compared to iPhone 12,
> high end Androids are in a much worse position than AMD was compared to Intel during Bulldozer era.
>
For a phone, almost everything else matters more and is a bigger barrier to entry compared to the CPU core AMD could tout as their advantage. Only GPU has lesser importance. These two items are not worth much - what you have to bring to the table is cellular connectivity (with allt he regulatory stuff handled!), wireless, power management, the low power platform stuff and software ecosystem. These days, also sensors, ISP, AI unit.
The assets AMD might have had to offer with K12 were those that are easiest to solve by licensing off-the-shelf ARM Holdings IP. So basically, they had nothing. GPU is something you can license but not use as a base for selling our SoC.
I doubt it was viable to make money by licensing K12 as a Cortex laternative. Even if AMD were able to pull that off and ARM would not push them out, the revenue stream would be small and not really worth AMD's time.
> > Look at how the processor debates go in the vein of "dumb old Intel needed Apple to wake them
> > up else the CPU market would go nowhere", as if AMD's influence and market presence (which
> > should not be overblown but is still significant) didn't exist at all or something.
> >
> > Bascially the problem with this topic is: ARM being a good idea doesn't mean it would have worked for AMD
> > Odds were decent that K12 would just be another waste of money a la Seattle/SeaMicro.
> > There is also the kicker that AMD's help with trailblazing server platform support for ARM
> > eventually only helped competitors that didn't have "not a cool kid" sticker on their back,
> > so it was a really dangerous mistake from them. You don't create a new market that kills your
> > old market, if only your competitors will profit while you paid the costs for them.
>
> Pushing ARM for server was dumb, for sure, but the mobile market is already
> consolidated, it is just the high end supplier position which is vacant.
>