In article <slrnh9vidv.2v5.jedi@xxxxxxxxxxxxx>,
JEDIDIAH <jedi@xxxxxxxxxxxxx> wrote:
> > 2. Support for multiple CPUs/cores (see page 10-13). Let's see you take
> > an existing C program in Linux, and make it work as easily and
> > efficiently on multiple cores or CPUs as the example on page 13.
>
> ...you say this as if Unix hasn't been working with multiple processors
> since before any form of MacOS ever existed. The idea that Linux can't
> readily exploit multiple processors is beyond absurd.
You should try reading the cited article, and you'll see what's
different about the way Linux (and OS X before Snow Leopard) handles
this and the way Snow Leopard handles it, and why it is a big
improvement.
> > 3. GPU programming (see page 14 and 15). Apple has OpenCL. Fortunately
> > for Linux, Apple turned this over to the same group that oversees
> > OpenGL, and AMD has a beta available for Linux. However, Apple is still
> > ahead at this point.
>
> There are entire companies dedicated to producing OpenCL tuned solutions
> for Linux. If you want to rant about an area where Apple is "ahead" this
> is certainly not it. If anything, OpenCL is a good demonstration of why
> a "closed" platform is bad and tends to stiffle innvoation.
I said OpenCL, not OpenGL. OpenCL was invented and initially developed
at Apple, then developed further by Apple, Intel, AMD, and Nvidia. Apple
them turned it over to the Khronos Group in mid-2008. A public
specification was released on Dec 8 2008. It is only in the last few
months that some companies, such as Intel and AMD, have started making
beta implementations available for other platforms, such as Windows and
Linux.
--
--Tim Smith
|
|
