ARM Cast Cost Table

[Nadav Rotem]

On Jan 29, 2013, at 11:20 AM, Arnold Schwaighofer <aschwaighofer at apple.com> wrote:

> In the case of func0 we pay for sign-extension while in func1 it can be merged into the arithmetic. The question is now which cost should we return: the optimistic or the pessimistic one? I would lean towards the optimistic one (as you do) but I think we should get an agreement on which one to use in such cases.
> 

The cost model is accumulative, and the costs of instructions are added. So if we miss the scalar version then we will also miss the vector one. Right ? My opinion on this is that it does not matter as long as there is consistency between the scalar and vector operations. 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20130129/8eafacf5/attachment.html>