[llvm-commits] [llvm] r49928 - /llvm/trunk/lib/Analysis/ScalarEvolution.cpp

[Chris Lattner]

On Apr 19, 2008, at 5:32 PM, Dale Johannesen wrote:

>
> On Apr 19, 2008, at 5:22 PM, Nick Lewycky wrote:
>
>> Chris Lattner wrote:
>>> Shouldn't this just be:
>>>
>>>  case ICmpInst::ICMP_ULT: {
>>>    SCEVHandle TC = HowManyLessThans(LHS, RHS, L, false);
>>>    if (!isa<SCEVCouldNotCompute>(TC)) return TC;
>>>    break;
>>>  }
>>>  case ICmpInst::ICMP_UGT: {
>>>    SCEVHandle TC = HowManyLessThans(RHS, LHS, L, false);
>>>    if (!isa<SCEVCouldNotCompute>(TC)) return TC;
>>>    break;
>>>  }
>>>
>>> x > y  -->  y < x
>
> First thing I tried, but:
>
>> The first thing HowManyLessThans does is notice that your the RHS
>> isn't
>> loop invariant and give up.
>>
>> Nick
>
> Yep.
> I then considered adding HowManyGreaterThans but it turns out UMin
> isn't there either....

Ah ok, good catch :).  I don't think howfartozero is safe to use in  
any case.  If it were, we could just use it in the ULT case too, right?

-Chris