By: Klimax (danklima.delete@this.gmail.com), August 15, 2014 1:57 pm
Room: Moderated Discussions
juanrga (nospam.delete@this.juanrga.com) on August 15, 2014 11:39 am wrote:
> David Kanter (dkanter.delete@this.realworldtech.com) on August 15, 2014 9:41 am wrote:
>
> > I also am skeptical that anyone can match Intel in performance while trailing behind by a node. To do that,
> > it is necessary to offer a very different product with a
> > different system architecture. E.g., target workloads
> > where cache does not help and simply slap down more memory controllers and cores (hint: that's a GPU!).
> >
>
> There are broad evidence on how low-end ARM processors on 28nm planar node match
> or outperform best Intel designs (for the same target) on 22nm FinFET.
So far nobody got there. Marketing slides are really bad for running real load...
And we know that slides will survive anything.
Also contrived benchmarks don't make real workload.
> The ISA advantage will be greatly reduced in the top-end side of the performance spectrum, but will not vanish.
> Keller mentioned during Core Day conference that his K12 core will have a "bigger engine" than its x86 sister
> thanks to the advantages of ARMv8 over x86-64, which allows to spend more transistors on compute.
Considering how nobody is close to full Xeons, that tiny advantage won't count for much.
(POWER is sort of exception, but not in vast majority of servers/markets due to its price)
Assuming there is any real advantage.
> Another key is that Intel process advantage will be reduced. Those server-class ARM SoCs that I mentioned will
> be made on 14/16 nm FinFET. Broadwell-EP and Skylake-EP on 14nm FinFET will not have a full node advantage.
You do know that no foundry is so far anywhere close to Intel in that regard? We will talk when they will finally ship, which considering history doesn't make positive case for you.
> David Kanter (dkanter.delete@this.realworldtech.com) on August 15, 2014 9:41 am wrote:
>
> > I also am skeptical that anyone can match Intel in performance while trailing behind by a node. To do that,
> > it is necessary to offer a very different product with a
> > different system architecture. E.g., target workloads
> > where cache does not help and simply slap down more memory controllers and cores (hint: that's a GPU!).
> >
>
> There are broad evidence on how low-end ARM processors on 28nm planar node match
> or outperform best Intel designs (for the same target) on 22nm FinFET.
So far nobody got there. Marketing slides are really bad for running real load...
And we know that slides will survive anything.
Also contrived benchmarks don't make real workload.
> The ISA advantage will be greatly reduced in the top-end side of the performance spectrum, but will not vanish.
> Keller mentioned during Core Day conference that his K12 core will have a "bigger engine" than its x86 sister
> thanks to the advantages of ARMv8 over x86-64, which allows to spend more transistors on compute.
Considering how nobody is close to full Xeons, that tiny advantage won't count for much.
(POWER is sort of exception, but not in vast majority of servers/markets due to its price)
Assuming there is any real advantage.
> Another key is that Intel process advantage will be reduced. Those server-class ARM SoCs that I mentioned will
> be made on 14/16 nm FinFET. Broadwell-EP and Skylake-EP on 14nm FinFET will not have a full node advantage.
You do know that no foundry is so far anywhere close to Intel in that regard? We will talk when they will finally ship, which considering history doesn't make positive case for you.
