[LLVMdev] Dependence Analysis [was: Flow-Sensitive AA]

[Chris Lattner]

On Aug 22, 2008, at 3:05 PM, Vikram S. Adve wrote:

> On Aug 22, 2008, at 4:49 PM, John Regehr wrote:
>
>> Has anyone quantified the optimizations afforded by undefined signed
>> overflow?  I'd expect that the benefits are minimal for most codes.
>
> In most cases, I agree.  But for codes that depend heavily on
> dependence analysis, I would think that being conservative with index
> expressions would really kill any disambiguation capability and make
> many loop optimizations and other dependence-based optimizations much
> weaker.  For example, static scheduling of array intensive loops seems
> vulnerable to this.

There are many 'instcombine' level optimizations that become safe when  
overflow can't happen, e.g. x*2/2 -> x.

However, by far the most important case is loop induction variables  
for realistic code.

-Chris