OrionWorks wrote:
The web page states rather cryptically: "The only C language
environment that unlocks the many-core processing power of GPUs to
solve the world's most computationally-intensive challenges". Ok...
Any C compiler??? ...or does one need to purchase a specially
designed C language compiler.
It looks like their own compiler that produces parallel code. It is
the "CUDA C" processor with a set of optimized libraries. In other
words, if you already have your problem coded in C you just recompile
it. Or I guess most people know how to write in C these days.
Not long ago this would have been Fortran instead of C.
See:
http://uuu.enseirb.fr/~pelegrin/enseignement/enseirb/archsys/documents/architectures/tesla_technical_brief.pdf
QUOTE:
The key to CUDA is the C compiler for the GPU. This first-of-its-kind
programming environment simplifies coding parallel applications.
Using C, a language familiar to most developers, allows programmers
to focus on creating a parallel program instead of dealing with the
complexities of graphics APIs. To simplify development, the CUDA C
compiler lets programmers combine CPU and GPU code into one
continuous program file. Simple additions to the C program tell the
CUDA compiler which functions reside on the CPU and which to compile
for the GPU. The program is then compiled with the CUDA compiler for
the GPU, and then the CPU host code is compiled with the developer's
standard C compiler.
Developers use a novel programming model to map parallel data
problems to the GPU. CUDA programs divide information into smaller
blocks of data that are processed in parallel. This programming model
allows developers to code once for GPUs with more multiprocessors and
for lower-cost GPUs with fewer multiprocessors. . . .
MPP architecture has been a long time coming but I am convinced it is
the wave of the future. I think it would help things like voice input
and translation software, and artificial intelligence of course. See
chapter 10 of my book. I do not know whether programmers will ever
become good at writing parallel algorithms. Perhaps they cannot do
this because they are not used to parallel architecture, or perhaps
because the human mind deals with problems in a serial step-by-step
fashion only (even though the brain itself is an MPP processor par
excellence). But whether people ever get good at it or not, compilers
will eventually make the process automatic.
This gadget splits the object code between the conventional CPU and
the parallel processors. In most programs, a small section of the
code runs most of the time. A large body of code sets up the problem,
and then a small section iterates to solve the problem. A compiler
that outputs code for parallel processors can make most of the code
and most functions sequential (ordinary); only the innermost
iterative code needs to be made parallel. (But perhaps in the future
sequential code will not be considered ordinary.)
- Jed
