On 7/25/07, Paolo Bonzini <[EMAIL PROTECTED]> wrote:
> For performance small arrays should be the same as individual members > (I can see the annoying fact that initialization is a headache - this has > annoyed me as well). For larger arrays (>4 members), aliasing will > make a difference possibly, making the array variant slower. Any union > variant is expected to be slower for aliasing reasons (we do not do > field-sensitive aliasing for unions). I wonder if the fabled "union as VIEW_CONVERT_EXPR" patch would help... Richi, do you know?
I don't think so. As long as our memory optimizers are weak and only work well if they are faced with single SFT.xx VUSEs and VDEFs then using a union will be hit by create_variable_info_for not creating field variables for unions at all (and thus memory optimizers not be able to tell apart fields). Richard.
