[LLVMdev] Performance regression on ARM

[Stephen Checkoway]

On Oct 20, 2014, at 6:10 AM, Joerg Sonnenberger <joerg at britannica.bec.de> wrote:

> On Sun, Oct 19, 2014 at 06:10:26PM -0700, Stephen Checkoway wrote:
>> In general, multiplying complex numbers in rectangular form is pretty straight-forward:
>> (a+ib)(c+id) = ac - bd + i(ad+bc).
>> 
>> Division is a little more of a hassle:
>> (a+ib)/(c+id) = (a+ib)(c-id)/(c^2 + d^2)
>> = (ac + bd)/(c^2 + d^2) + i(bc - ad)/(c^2 + d^2).
> 
> Both forms are problematic if you are about overflow and NaN. Which is
> exactly what my question is about.

My apologies. I misunderstood. When you said you meant real in the mathematical sense, I thought you were asking about the mathematical nature of the operation, not language guarantees.

-- 
Stephen Checkoway