yocto llvm clang for epiphany?
Posted: Thu Mar 02, 2017 10:16 amHi,
Just wondered if anyone is interested in using llvm in a yocto environment as an alternative to e-gcc. Now that an Epiphany backend is in progress thanks to @upcFrost (https://github.com/upcFrost/Epiphany - viewtopic.php?f=23&t=3966&p=18474&sid=413db2c74d20953d31d614cc52e73960#p18474) I am considering what it would take to bring this into the yocto environment (see https://layers.openembedded.org/layerin ... eta-clang/).
It seems that progress on improving support for multiple cores for Epiphany or other multi core processors will likely go hand in hand with tools using llvm to create code. From my point of view I think that the construction and optimization of the DAG for Epiphany needs to take into account the small memory size. Not sure how this is achieved but it seem like a area of development that might be of use. Using yocto allows an x86 build machine to be used to build the llvm / clang tools.
Peter.
Just wondered if anyone is interested in using llvm in a yocto environment as an alternative to e-gcc. Now that an Epiphany backend is in progress thanks to @upcFrost (https://github.com/upcFrost/Epiphany - viewtopic.php?f=23&t=3966&p=18474&sid=413db2c74d20953d31d614cc52e73960#p18474) I am considering what it would take to bring this into the yocto environment (see https://layers.openembedded.org/layerin ... eta-clang/).
It seems that progress on improving support for multiple cores for Epiphany or other multi core processors will likely go hand in hand with tools using llvm to create code. From my point of view I think that the construction and optimization of the DAG for Epiphany needs to take into account the small memory size. Not sure how this is achieved but it seem like a area of development that might be of use. Using yocto allows an x86 build machine to be used to build the llvm / clang tools.
Peter.