[Openicc] GSoC & oyranos
Bob Friesenhahn
bfriesen at simple.dallas.tx.us
Sat Apr 4 15:14:27 PDT 2009
On Sat, 4 Apr 2009, Kai-Uwe Behrmann wrote:
> Ah, that is a point I did not really understand. So a library can add a
> (GPU) backend to LLVM at runtime? E.g. thus enabling GPU support in
> gallium3d while OpenGTL has no benefit?
> So a (longterm) plan could be:
> CTL/Shiva -> IL -> GLSL -> IL -> GPU/CPU assembler code ?
It seems that the LLVM VM just needs to be ready to accept
instructions. Based on what I saw while building LLVM, it seemed like
assembly output for all supported targets is available in one build.
Not having any experience with LLVM before, I installed it on a
FreeBSD/Xeon system and built GraphicsMagick with it. Then I ran a
benchmark suite which uses OpenMP. Usually native GCC was somewhat
faster than LLVM generated code, but for some floating-point intensive
algorithms, LLVM was faster. I would say that performance was pretty
competitive.
OpenMP was detected and worked.
The only anomally I noticed was that C++ exceptions did not work in
the build.
There is no autotools support for building LLVM IL files so I was
unable to test its bitstream VM capability.
Bob
--
Bob Friesenhahn
bfriesen at simple.dallas.tx.us, http://www.simplesystems.org/users/bfriesen/
GraphicsMagick Maintainer, http://www.GraphicsMagick.org/
More information about the openicc
mailing list