Radu
|
2011-01-06
, 08:13
|
Posts: 303 |
Thanked: 146 times |
Joined on Aug 2009
|
#41
|
|
2011-01-06
, 08:18
|
Posts: 2,802 |
Thanked: 4,491 times |
Joined on Nov 2007
|
#42
|
Not to mention the fact that ARM is a lot more fragmented (v5 ? v7 ? NEON ? OpenGL (ES) ?) than X86, which is mostly incremental.
|
2011-01-06
, 10:52
|
Posts: 3,319 |
Thanked: 5,610 times |
Joined on Aug 2008
@ Finland
|
#43
|
x86 is just as much, if not more, fragmented. The issues are pretty much the same: "x" can refer to a number of different instruction sets, then you have a large set of SIMD extensions, FPU vs VFP, PAE, virtualisation extensions, hardware crypto and so on. ARM is also incremental, you can still run for example ARMv5 binaries on an N900. (OpenGL is an orthogonal issue)
It's just that on x86 most distros are doing a decent job picking build options that produce binaries that can run on most hardware that's still in use. The exceptions tend to be in the "squeeze every bit of performance out of Atom chips" category, such as MeeGo/x86 and Ubuntu's (now defunct) LPIA "architecture", or Gentoo's "optimise for this box and compatibility be damned" approach.
|
2011-01-06
, 11:25
|
Posts: 999 |
Thanked: 1,117 times |
Joined on Dec 2009
@ earth?
|
#44
|
|
2011-01-06
, 11:29
|
|
Posts: 1,789 |
Thanked: 1,699 times |
Joined on Mar 2010
|
#45
|
|
2011-01-06
, 17:15
|
Posts: 1,746 |
Thanked: 2,100 times |
Joined on Sep 2009
|
#46
|
|
2011-01-06
, 17:29
|
|
Posts: 2,355 |
Thanked: 5,249 times |
Joined on Jan 2009
@ Barcelona
|
#47
|
Ok, now I am totally confused. If DosBox uses code translation, rather than interpretation, then why is it so slow? With [proper] code translation, I would expect average speeds around 1/10 of the speed of an X86 CPU at the same frequency or so. But in my observations, DosBox is much, much slower than that.
The Following User Says Thank You to javispedro For This Useful Post: | ||
|
2011-01-06
, 19:43
|
Posts: 303 |
Thanked: 146 times |
Joined on Aug 2009
|
#48
|
Well. "1/10 of the speed" of the stock N900's 3430 that is around the performance of a Celeron 333Mhz is indeed around a 386.
|
2011-01-06
, 21:12
|
|
Posts: 2,355 |
Thanked: 5,249 times |
Joined on Jan 2009
@ Barcelona
|
#49
|
|
2011-01-07
, 00:58
|
Posts: 303 |
Thanked: 146 times |
Joined on Aug 2009
|
#50
|
Tags |
die ms die, microfail |
|