On 11/20/2013 11:37 PM, Erik Schnetter wrote: > What should we do? Pre-relase clang-3.4 is rather important for us. Should > we omit -ffake-address-space-map? Or does this need to be addressed in > clang?
-ffake-address-space-map has been used to avoid losing the information of the different OpenCL address spaces after the frontend: http://portablecl.org/docs/html/memory_management.html If that is omitted, all pointers go to AS 0 for CPUs and then in our kernel compiler or other processing we cannot differentiate what was the original meaning. That fake-address-space map sets them to known fixed ones, regardless of the target. There was some discussion in LLVMdev about handling the "logical address spaces" in more sensible way to address the same issue. Unfortunately, I lost track of the discussion at some point, so I do not know if they implemented such and removed this switch as unneeded. That needs to be checked first. -- Pekka ------------------------------------------------------------------------------ Shape the Mobile Experience: Free Subscription Software experts and developers: Be at the forefront of tech innovation. Intel(R) Software Adrenaline delivers strategic insight and game-changing conversations that shape the rapidly evolving mobile landscape. Sign up now. http://pubads.g.doubleclick.net/gampad/clk?id=63431311&iu=/4140/ostg.clktrk _______________________________________________ pocl-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/pocl-devel
