Re: How can RISC OS computers most effectively survive in the future?

On Sep 5, 12:13 pm, druck <n...@xxxxxxxxxxxxxxxx> wrote:
The main problem is where are the new applications to take advantage
of it? There are only a few applications still in active development,
and an even smaller number would be extensively re-written to take
advantage of any new system architecture. That would leave the vast
majority of applications running under some form of emulator, working
exactly as they do now, only slower.

They basic idea, druck, was a new system architecture, to encourage
developers to come back and develop applications on the platform, as
the java idea is too... about java you said "it would run few apps..."
well in this "few" (but they are not few anyway) there is Eclipse,
which (even if with all the problems of the world, and don't let start
a discussion about Eclipse pleaseeeeee) is used by many, many
developers and it is a good start point from where people may have a
better IDE on RISC OS...


Apple moved processor architectures twice and were able to use
emulation as a stop gap for existing applications, because people
actually bough new hardware when it came out (as opposed to still
having 68K Macs classics 14 years later and whinging about them), and
the developers of large expensive MAC applications made a lot of money
from selling upgrades to faster native versions (as opposed to saying
Draw does everything they need, so they have no need for Artworks).

well... with Virtual Acorn and other projects it sounds like that
there is people in the ROS community which are tired of the "forced"
join to ARM, which (IMHO) is a good platform, but for desktop
applications it will never be competitive with the Intel (AMD etc...)
side of the moon ;) (even the AMBA bus is great, I love it, but can
not compare with technologies like Hyper Threading or Hyper Transport
from AMD - just a note, actually I am getting bored with manuals about
hyper threading code optimization and the true is that no OS, at the
moment, is really taking advantage by this technology, even Windoze,
which, in the better case will get the 10% and/or not more than the
20% of speed improvement by Hyper Threading).


With a kernel more advanced than the actual one also the emulation
would be faster, plus the new HW would help in that sense.

Its wishful thinking such as system would be faster, and what new
hardware is this?
Com'on druck ROS architecture is lacking a lot of optimizations (and
you know I am not talking about simple kernel hacks to improve
performance, but about algorithms changing and even architectural
changes which will change the actual situation a lot). Of course
software can't go much far from what hardware allows it to go, but ROS
is really NOT an OS CPU throughput oriented as other OSes does. (an
example of this is all the people which are still asking for not
blocking I/O, which will comes out only from a deep system
refactoring).

As new hardware... I just want to mention the new development lines in
ARM Ltd about the "A" processors which are going to be optimized for
Applications Execution, and YES, once again, Cortex A9 could represent
a good start point... (sorry but I didn't change my opinion about it)
(and there is already some boards with flash memory on, PCIe, PCI,
multi processors and SDRAM...).

-Paolo
.

Relevant Pages

  • Re: Windows Vista - worst OS yet?
    ... both the hardware and software level. ... applications developers willing to sign all the NDAs, ... to be hostile to he rights of the protected content on the platform. ...
    (sci.electronics.design)
  • Re: Is .NET going to be dropped by Microsoft?
    ... For line-of-business applications using .NET's paint optimization features ... >and developers influential in the decision process. ... Most Microsoft embedded device development has or is transtitioning to .NET ... > My sources are from some relatively large organizations such as Bank of> the West, Bank of America, SBC, ATI, and a few others -- from senior> management and developers influential in the decision process. ...
    (microsoft.public.dotnet.languages.vb)
  • Re: Is that a joke ?
    ... than C++ apps for most uses. ... It probably looks fast to Java or VB developers but what ... > What about CAD, CAE, CAM, scientific applications, utilities, math, ... a garbage collector doesn't make good programmers or safe ...
    (microsoft.public.dotnet.general)
  • RE: Is that a joke ?
    ... > native code. ... It probably looks fast to Java or VB developers but what ... > about real time applications? ... > programmers but not the rest of us who used to hand-optimise our code, ...
    (microsoft.public.dotnet.general)
  • Re: Is that a joke ?
    ... avoid getting slapped around by .Net gotchas. ... > native code. ... It probably looks fast to Java or VB developers but what ... > What about CAD, CAE, CAM, scientific applications, utilities, math, ...
    (microsoft.public.dotnet.general)