| 1 |
On Sat, Dec 8, 2012 at 10:25 PM, Grant <emailgrant@×××××.com> wrote: |
| 2 |
> It seems like ARM processors will destroy x86 before too long. Does anyone |
| 3 |
> think this won't happen? |
| 4 |
|
| 5 |
It's looking promising. Not that I have a horse in the race, but I |
| 6 |
very much like ARM's low power consumption. The way I see it, they're |
| 7 |
only a short list of features away from obliterating x86: |
| 8 |
|
| 9 |
* I'd like to see fast division. |
| 10 |
I keep hearing about how this or that is slow because of ARM's lack of |
| 11 |
strong division. |
| 12 |
|
| 13 |
* I'd like to see a modern baseline of strong instructions. |
| 14 |
x86 kept continually improving in a very fragmented way, but there |
| 15 |
were, from time to time, baseline collections of feature sets you |
| 16 |
could expect all processors to have. i386 represented one. i686 |
| 17 |
represented one. Currently, it's x86_64, which implies not only a |
| 18 |
64-bit flattened address space and a departure from real mode, but |
| 19 |
also a collection of SIMD instruction sets and other features |
| 20 |
developed between the release of the Pentium Pro and AMD's Hammer |
| 21 |
architecture. |
| 22 |
|
| 23 |
ARM just feels...fragmented. And I don't have the impression I could |
| 24 |
write my code assuming the availability of SIMD (presuming I use |
| 25 |
things like OpenMP to expand my code to leverage it, rather than |
| 26 |
writing processor-specific code. Though OpenCL could very well |
| 27 |
alleviate that issue.) |
| 28 |
|
| 29 |
* I'd like to see virtualization be a thing. |
| 30 |
|
| 31 |
Productivity and efficiency on x86 *soared* with the |
| 32 |
compartmentalization that came with hardware-assisted (and therefore |
| 33 |
cheap! and fast!) virtualization. I haven't heard about the same on |
| 34 |
ARM, although Citrix is working hard on porting Xen there. Paravirt |
| 35 |
may well be the first common means of virtualization on ARM... |
| 36 |
|
| 37 |
-- |
| 38 |
:wq |
Archives