By: juanrga (nospam.delete@this.juanrga.com), August 18, 2014 2:48 pm
Room: Moderated Discussions
David Kanter (dkanter.delete@this.realworldtech.com) on August 16, 2014 11:31 am wrote:
> juanrga (nospam.delete@this.juanrga.com) on August 16, 2014 4:34 am wrote:
> > David Kanter (dkanter.delete@this.realworldtech.com) on August 15, 2014 6:54 pm wrote:
> > > Alexko (lexoka.delete@this.gmail.com) on August 15, 2014 5:43 pm wrote:
> > > > David Kanter (dkanter.delete@this.realworldtech.com) on August 15, 2014 3:01 pm wrote:
> > > > > 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!).
> > > > >
> > > > > > 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.
> > > > >
> > > > > I happen to know the differences between those two designs. I'm not really sure it's
> > > > > going to translate into a significant performance delta. My guess is maybe 10%.
> > > >
> > > > Interesting! Are you planning on writing an article on K12 and/or its x86 variant?
> > >
> > > Right now there aren't really sufficient details. I look forward
> > > to when AMD will disclose a bit more about the architectures.
> > >
> > > David
> >
> >
> > So far as I know AMD is discussing to disclose details of the new arch sometime in the next year.
> > Meanwhile could you share some basic details, such as if the core will be big and fast (Haswell-like)
> > or smaller and efficient? CMP or CMT? Is true that three pipes are shared each two cores?
>
> I can't really share anything unfortunately. However, I can
> say that they will be ditching Bulldozer and its lineage.
>
> David
Here some guesses/leaks:
The frequency will be around 3.0--3.5GHz.
Someone said that K12 will be made on TSMC 16FF+.
The same person mentions that CMT is abandoned by CMP, the cores will be small (he used the term "minicore Jaguar/Puma style") rather than big as Haswell, up to 16 core on die, the SoC design will be lego-like, with different custom options for L3 cache, stacked RAM, core count, iGPU, and even SIMD wide.
Some other person mentioned something about SMT3, but this number did look weird to me.
It seems that Keller has been working hard on the cache subsystem, and that part of his work will be already reused in Excavator.
> juanrga (nospam.delete@this.juanrga.com) on August 16, 2014 4:34 am wrote:
> > David Kanter (dkanter.delete@this.realworldtech.com) on August 15, 2014 6:54 pm wrote:
> > > Alexko (lexoka.delete@this.gmail.com) on August 15, 2014 5:43 pm wrote:
> > > > David Kanter (dkanter.delete@this.realworldtech.com) on August 15, 2014 3:01 pm wrote:
> > > > > 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!).
> > > > >
> > > > > > 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.
> > > > >
> > > > > I happen to know the differences between those two designs. I'm not really sure it's
> > > > > going to translate into a significant performance delta. My guess is maybe 10%.
> > > >
> > > > Interesting! Are you planning on writing an article on K12 and/or its x86 variant?
> > >
> > > Right now there aren't really sufficient details. I look forward
> > > to when AMD will disclose a bit more about the architectures.
> > >
> > > David
> >
> >
> > So far as I know AMD is discussing to disclose details of the new arch sometime in the next year.
> > Meanwhile could you share some basic details, such as if the core will be big and fast (Haswell-like)
> > or smaller and efficient? CMP or CMT? Is true that three pipes are shared each two cores?
>
> I can't really share anything unfortunately. However, I can
> say that they will be ditching Bulldozer and its lineage.
>
> David
Here some guesses/leaks:
The frequency will be around 3.0--3.5GHz.
Someone said that K12 will be made on TSMC 16FF+.
The same person mentions that CMT is abandoned by CMP, the cores will be small (he used the term "minicore Jaguar/Puma style") rather than big as Haswell, up to 16 core on die, the SoC design will be lego-like, with different custom options for L3 cache, stacked RAM, core count, iGPU, and even SIMD wide.
Some other person mentioned something about SMT3, but this number did look weird to me.
It seems that Keller has been working hard on the cache subsystem, and that part of his work will be already reused in Excavator.