By: Patrick Chase (patrickjchase.delete@this.gmail.com), May 20, 2013 9:29 am
Room: Moderated Discussions
Wilco (Wilco.Dijkstra.delete@this.ntlworld.com) on May 19, 2013 5:19 am wrote:
> David Kanter (dkanter.delete@this.realworldtech.com) on May 17, 2013 9:38 pm wrote:
> > > Second, do you have any support for this claim that A15
> > > was designed for servers and repurposed for mobiles?
> >
> > It's something I've heard from half a dozen people. ALthough I've also
> > heard contrary evidence (recently).
>
> What about reading ARM's press release when A15 was originally announced, ARM's A15
> product pages or any articles about A15? Not a single resource claims that A15 is
> server-only. I'm not sure who is trying to claim that (your Intel contacts?), but
> it is just BS, and frankly you should know better than to fall for it.
Whoa, easy. I tend to agree with you that David got this particular (and fairly irrelevant/trivial) point wrong, but that happens to everybody from time to time and he's already acknowledged that possibility. At that point you'd won the debate as much as you were ever going to, since there's no way to conclusively resolve a question of "initial design intent" without getting an ARM architect to violate their employment contract. The public flogging isn't serving any purpose.
I think you can objectively state that the A15 has server-oriented features, most notably ARM's version of PAE (can't remember what they called it). They wouldn't have put that in if A15 had been targeted purely as a mobile application processor, so there appears to be at least some underlying grain of truth to David's argument. It's a matter of temporal order (which came first) and degree.
What I actually suspect happened is that A15 started out targeted as a mobile application processor, but then microservers happened and ARM tried to stretch it in that direction during development. PAE in particular seemed to be a late-ish addition. They would have been silly if they hadn't done that, so this isn't any sort of negative comment towards/about ARM. This is one case where I agree 100% with their execution.
> Still incorrect - tablets aren't mentioned anywhere in any A15 announcement.
That announcement reflects the *outcome* of the design process, not its genesis. David's argument was about design intent (i.e. ARM's intentions when they *started* to design A15) so you'd have to go back at least a few years further than that announcement to prove anything one way or another.
> Originally the design point was 1-1.5GHz single/dual cores for phones, 1.5-2.GHz dual
> core for home entertainment, and 1.5-2.5GHz quad/octal core servers. That's still on
> the rel="nofollow">A15 product pages.
The content you reference is all VERY recent relative to the development timeline of A15/Eagle. Can you come up with something from back when ARM first started discussing Eagle with licensees and Si partners (under CDA of course)? That would be the earliest that anybody who doesn't work for ARM could realistically go.
> > We already know Intel is quite capable of producing rotten CPU designs (see Atom),
> > but oddly enough those still seem to have pretty good performance relative to
> > something like the A9.
>
> Certainly not on single threaded performance. The only redeeming feature of
> Atom is Hyperthreading [snip]
I'd say that the main redeeming feature of Atom relative to A9 is its memory subsystem (this goes back to a point Linus made in another post). A9 is blazing fast out of cache, but you can get into trouble really quickly when you go out to the bus. In addition to various inefficiencies/latencies, a cache line size of 32b is on the small side for modern bus fabrics and DDR2/3 memories.
> David Kanter (dkanter.delete@this.realworldtech.com) on May 17, 2013 9:38 pm wrote:
> > > Second, do you have any support for this claim that A15
> > > was designed for servers and repurposed for mobiles?
> >
> > It's something I've heard from half a dozen people. ALthough I've also
> > heard contrary evidence (recently).
>
> What about reading ARM's press release when A15 was originally announced, ARM's A15
> product pages or any articles about A15? Not a single resource claims that A15 is
> server-only. I'm not sure who is trying to claim that (your Intel contacts?), but
> it is just BS, and frankly you should know better than to fall for it.
Whoa, easy. I tend to agree with you that David got this particular (and fairly irrelevant/trivial) point wrong, but that happens to everybody from time to time and he's already acknowledged that possibility. At that point you'd won the debate as much as you were ever going to, since there's no way to conclusively resolve a question of "initial design intent" without getting an ARM architect to violate their employment contract. The public flogging isn't serving any purpose.
I think you can objectively state that the A15 has server-oriented features, most notably ARM's version of PAE (can't remember what they called it). They wouldn't have put that in if A15 had been targeted purely as a mobile application processor, so there appears to be at least some underlying grain of truth to David's argument. It's a matter of temporal order (which came first) and degree.
What I actually suspect happened is that A15 started out targeted as a mobile application processor, but then microservers happened and ARM tried to stretch it in that direction during development. PAE in particular seemed to be a late-ish addition. They would have been silly if they hadn't done that, so this isn't any sort of negative comment towards/about ARM. This is one case where I agree 100% with their execution.
> Still incorrect - tablets aren't mentioned anywhere in any A15 announcement.
That announcement reflects the *outcome* of the design process, not its genesis. David's argument was about design intent (i.e. ARM's intentions when they *started* to design A15) so you'd have to go back at least a few years further than that announcement to prove anything one way or another.
> Originally the design point was 1-1.5GHz single/dual cores for phones, 1.5-2.GHz dual
> core for home entertainment, and 1.5-2.5GHz quad/octal core servers. That's still on
> the rel="nofollow">A15 product pages.
The content you reference is all VERY recent relative to the development timeline of A15/Eagle. Can you come up with something from back when ARM first started discussing Eagle with licensees and Si partners (under CDA of course)? That would be the earliest that anybody who doesn't work for ARM could realistically go.
> > We already know Intel is quite capable of producing rotten CPU designs (see Atom),
> > but oddly enough those still seem to have pretty good performance relative to
> > something like the A9.
>
> Certainly not on single threaded performance. The only redeeming feature of
> Atom is Hyperthreading [snip]
I'd say that the main redeeming feature of Atom relative to A9 is its memory subsystem (this goes back to a point Linus made in another post). A9 is blazing fast out of cache, but you can get into trouble really quickly when you go out to the bus. In addition to various inefficiencies/latencies, a cache line size of 32b is on the small side for modern bus fabrics and DDR2/3 memories.
