[llvm-commits] [llvm] r128545 - in /llvm/trunk: include/llvm/ADT/APFloat.h lib/Support/APFloat.cpp unittests/ADT/APFloatTest.cpp

[Benjamin Kramer]

On 30.03.2011, at 18:11, Stephen Canon wrote:

> What will this patch do on a platform where denormals are treated as zero?
> 
> Specifically, I'm interested in a case like this:
> 
> 	float a;
> 	float b = a / 0x1.0p127f;
> 
> If this is converted into the reciprocal multiply by 0x0.8p-126f (a denormal value), on platforms where denormals are treated as zero (the default state on many ARM cores, cell BE, and GPUs), this will become a multiplication by zero, which is incorrect.
> 
> The "safe" range of divisors that can be optimized into reciprocal multiplies with regard to this issue, for any IEEE-754 type, is +/-[minNormal, 1/minNormal].

Hi Steve,

Thanks for the quick review. I changed the code to avoid denormal multiplications in r128555.

I wonder how (llvm-)gcc deals with this, on x86 it turns divisions into subnormal multiplications, is there
some target specific setting that disables it on ARM and other "odd" platforms?