On 09/18/2012 01:24 AM, Richard Earnshaw wrote: > Quite possibly. In order to determine what is right, we first need to > understand the specification. My reading of that is that the semantics > should be endian independent, but I was hoping that someone would know > for certain and be able to chip in.
It is supposed to be endian independant. The vector elements as visible to the user are in memory order. I *believe* we've fixed all of the middle-end interfaces have been fixed to be properly endian independant. There could easily be backend bugs that accidentally number vector elements as per their reference manual, as opposed to in memory order. We probably don't have enough test cases to ferret all of those out. r~
