search for: tt40041

...or half floating point because we are interested in it. we added from http://half.sourceforge.net/ a half-float library and changed some records in llvm and the way that handles half float. we also used the patch that was on this thread http://llvm.1065342.n5.nabble.com/PATCH-OpenCL-half-support-tt40041.html so the clang front-end could make IR with half. we created a target that has half float registers and tested on it. do you think it is a good idea to make a branch for it ?? or try to port it to llvm-3.2? thanks in advance Nikos Stavropoulos n.stavropoulos at think-silicon.com...