So I guess all the people defending him in the thread take it back?
They're going back to the previous article and feverishly editing their comments.
So I guess all the people defending him in the thread take it back?
They're going back to the previous article and feverishly editing their comments.![]()
64bit is old news since 1970s (2003 for consumers)
time for 128bit!
This has been discussed ad nauseum now. The 64 bit ARM is so much more then a 64-bit addressing space. It is FACT that the A64 instruction set provides performance gains over the AArch32 instruction set for many applications, and it has been proven by the benchmarks. People who claim that ARMv8 is a 'gimmick' simply show that they have no idea about ARM CPU architecture whatsoever.
exactly. seems all too common that marketing people with little to no technical experience speak out. if anything this just makes him look like a fool that has no idea what he is actually talking about. i wonder why they bother keeping him..oh wait, cuz he is an executive. -_-Marketing person comments on something technical and is totally wrong. Nothing to see here.
64bit is old news since 1970s (2003 for consumers)
time for 128bit!
Its been a while but when I hear 64bit I think addressing space and instructions. I was in school a little after the time most pcs were moving from 32 to 64bit. While 64 has more addressing space it also often has more advanced instructions. So this can be good for some processes and bad for others? I could be totally off but my understanding is that there can be more overhead in 64bit since the blocks are larger and often there are more advanced instruction sets means longer processing time for simple tasks.
It is correct that some applications which rely on pointer-heavy data structures can see performance degradation. As to your second claims - more advanced instruction set means faster processing times. The A64 instructions set has more registers (which means more local variables can be kept on the CPU, avoiding the very expensive store/load operations), wider SIMD operations (which usually means faster math operations) and some specialised instructions which can improve certain algorithms. Creating instructions which make code be executed slower is a pretty dumb thing, so CPU designers don't do it anymoreBTW, the execution speed of floating-point instructions via the SSE instruction set is MUCH faster then what the old x87 can do, and SSE can operate on multiple values at the same time.
Most people are missing the point here.
If there is anyone more worried than Apple about NOT making this look like a marketing gimmick, it is Qualcomm. They sell those chips, for christ sakes, and they want to sell more of them! What else did you expect them to say?
Of course, there IS a technical reason to move to 64bit in mobile devices. I think Apple has done it in part because ARM8 is more power efficient (and in case you didn't know, that's a big deal for computing devices), and ARM8 is 64bit only. That's one good reason. AND in ADDITION, it is a really good marketing gimmick because most people won't notice that the 64bits themselves are not providing any benefit whatsoever. They just think OMG 64BIT MY PHONE IS MORE FUTUREPROOF, not realizing THAT's a marketing gimmick, and not realizing the real benefit.
So there, two sides to the coin.
The 99.99% of people who have no idea what a microprocessor does heard "marketing gimmick" and many of them believe it. Yes, even now, and they will months from now. A footnote correcting his statement won't be heard. He got what he wanted.
Seems you know your stuff and thanks for the clarification (macrumors forum saves the day again!). It had been a long while since I took a very basic course which just touched on the matter so its good to have some updated information.