By: Simon Farnsworth (simon.delete@this.farnz.org.uk), February 23, 2021 9:42 am
Room: Moderated Discussions
Jukka Larja (roskakori2006.delete@this.gmail.com) on February 23, 2021 7:12 am wrote:
> dmcq (dmcq.delete@this.fano.co.uk) on February 23, 2021 5:35 am wrote:
>
> > I certainly wondered what on earth AMD were up to when they
> > made an ARM server chip. There is certainly no advantage
> > to them being a leader in that market, best push x86 for the
> > moment, and they can always be top class as a follower
> > in the future. The place I thought they might have a good
> > chance with ARM though is in games machines, it probably
> > isn't a huge money spinner but it certainy would be a nice market to be the go to leader in.
>
> AMD has gotten their x86 chips to last two-and-half generations of Playstations
> and Xboxes. Hard to see they could make more money with ARM based chips.
>
> Maybe you were thinking about some mobile gaming consoles and the edge ARM CPU could have there?
> Considering nearly everyone is already carrying a small games machine (smart phone) around all
> the time, it's hard to see much money can be made there. Only one to ever succeed there is Nintendo,
> and it's not because they have great hardware. They aren't going to pay large sums for 20, 50 or
> 100 % more CPU performance, as long as there are good enough SoCs available for less.
>
> -JLarja
Looking at the timelines on K12 and the consoles, it looks like K12 early design work began at about the same time as AMD won the console bids for PlayStation 4 and XBox One. It would thus seem plausible to me that K12 was also about hedging their bets if Microsoft and Sony (who had not been tied to an ISA in previous generations) started looking seriously at ARM-based competitors for the XBox Series X and PlayStation 5 generation of consoles. At least with a decent ARM core, if you hear that the decision is ARM, then you can still bid - albeit now competing with many ARM ecosystem chips.
Once it becomes clear that Zen + RDNA will be competitive for the next generation against any ARM ecosystem chip, then AMD don't need K12 for a future console bid - they can stick with x86 and beat the competition. At that point, they don't need the ARM core, so might as well scrap it.
> dmcq (dmcq.delete@this.fano.co.uk) on February 23, 2021 5:35 am wrote:
>
> > I certainly wondered what on earth AMD were up to when they
> > made an ARM server chip. There is certainly no advantage
> > to them being a leader in that market, best push x86 for the
> > moment, and they can always be top class as a follower
> > in the future. The place I thought they might have a good
> > chance with ARM though is in games machines, it probably
> > isn't a huge money spinner but it certainy would be a nice market to be the go to leader in.
>
> AMD has gotten their x86 chips to last two-and-half generations of Playstations
> and Xboxes. Hard to see they could make more money with ARM based chips.
>
> Maybe you were thinking about some mobile gaming consoles and the edge ARM CPU could have there?
> Considering nearly everyone is already carrying a small games machine (smart phone) around all
> the time, it's hard to see much money can be made there. Only one to ever succeed there is Nintendo,
> and it's not because they have great hardware. They aren't going to pay large sums for 20, 50 or
> 100 % more CPU performance, as long as there are good enough SoCs available for less.
>
> -JLarja
Looking at the timelines on K12 and the consoles, it looks like K12 early design work began at about the same time as AMD won the console bids for PlayStation 4 and XBox One. It would thus seem plausible to me that K12 was also about hedging their bets if Microsoft and Sony (who had not been tied to an ISA in previous generations) started looking seriously at ARM-based competitors for the XBox Series X and PlayStation 5 generation of consoles. At least with a decent ARM core, if you hear that the decision is ARM, then you can still bid - albeit now competing with many ARM ecosystem chips.
Once it becomes clear that Zen + RDNA will be competitive for the next generation against any ARM ecosystem chip, then AMD don't need K12 for a future console bid - they can stick with x86 and beat the competition. At that point, they don't need the ARM core, so might as well scrap it.