Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
It's exciting to see what machines will be able to do in the near-future. I've been really interested in Haswell for some time now. Broadwell at 14nm, will change computing forever.
 
Excuse my dumbness but arnt all transistors 3d? Even on a tiny proccessor they're not totally flat are they.

Guess im missing something but what is it?
 
Last edited:
Excuse my dumbness but arnt all transistors 3d? Even on a tiny proccessor they're not totally flat are they.

Guess im missing something but what is it?

You're right that the transistors are 3D, but the gate isn't; its a single plane, these new transistors will have 3D planes.
 
The standard name of the "3D transistors" is FinFETs.

Yeah, there's a lot of marketing mumbo-jumbo going on here. Transistors have always been "3D" structures by the nature of deposition and doping and simple physics. Intel has just managed a method whereby silicon and metal can share the same y space. I haven't actually researched how they do that part, but if you understand transistors, you can see the immediate benefits to having this much surface area control over a gate. It's like tripling the gate width (and thus, speeding up the transistor) with no sacrifice in footprint.
 
It's not about dislike what we have now. The current design is beautiful. It's about wanting more. like a greedy child lol. I'm just curious as to what apple comes up with next

Oh sure. I am too. And I'm sure whatever it is, whenever it is released, it'll make what we have now look dated. That said, I'd wager, we still have a bit longer to go with this enclosure. We had a good six years with the last style and part of the reasoning behind the changeover (and is usually a key reasoning for any Apple redesign) is that the new design was actually better engineered than the previous. So far, I haven't really heard of any problems with this enclosure.
 
I wonder the difference between this and the Arms chips now.

Can they actually surpass the Arm in capacity and battery life?
 
I wonder the difference between this and the Arms chips now.

Can they actually surpass the Arm in capacity and battery life?

No, there are power consumption differences inherent in the ISAs that Intel cannot overcome with process improvements alone.
 
... and those are?

for one Intel chips have a lot more cache on the CPU to keep the most used instructions on the CPU die. most of the processor die these days is cache

even with the A6, ARM CPU's will only be at the xbox 360/PS3 level of power while the new consoles that will come out soon will be even faster.
 
for one Intel chips have a lot more cache on the CPU to keep the most used instructions on the CPU die. most of the processor die these days is cache

even with the A6, ARM CPU's will only be at the xbox 360/PS3 level of power while the new consoles that will come out soon will be even faster.

Apple is known for having relatively low specs compared to the industry average yet STILL High Performing devices. Usually out performing their competitors. Take for instance the iPad 2 compared to the HP Slate (Apple makes better use of the resources they put into their devices)
 
This difference in decoding logic between a CISC and RISC architecture is the first that comes to mind.[/url]

But in the end - both will run similar numbers of operations to do the task. x64 reads fewer instructions, and expands them into primitive operations. RISC reads (and decodes) more primitive instructions.

There are some extremely power-hungry RISC chips out there (Alpha, POWER, ...).


There are other small differences such as how register dependent an architecture is.

If there's a measurable difference in the power consumption from a memory operation vs. a register operation, OK - but at the implementation level register renaming reduces the effect.
 
But in the end - both will run similar numbers of operations to do the task. x64 reads fewer instructions, and expands them into primitive operations. RISC reads (and decodes) more primitive instructions.

There are some extremely power-hungry RISC chips out there (Alpha, POWER, ...).

Sure, there are power hungry RISC architectures out there. But look at the performance/watt on an ARMv7 core versus an Atom core, for instance.

If there's a measurable difference in the power consumption from a memory operation vs. a register operation, OK - but at the implementation level register renaming reduces the effect.

True. I'm just pointing out that ISA differences necessarily translate into hardware differences as well.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.