News for hackers. Stuff that matters.
Re:Catching Integer Overflows in C by jensnockert on Tuesday August 02, 2011 @08:48AM (#2836908)
Or use saturating arithmetic, loads of common processors have instructions for that.
Re:Catching Integer Overflows in C by tedunangst on Tuesday August 02, 2011 @01:37PM (#2837956)
Yes, "don't use C" can solve lots of problems, but it doesn't solve them when using C.
Re:Catching Integer Overflows in C by carbonica on Tuesday August 02, 2011 @08:57AM (#2836930)
x86 only has SIMD instructions for it, IIRC.
Re:Catching Integer Overflows in C by onan_barbarian on Tuesday August 02, 2011 @09:03AM (#2836946)
You do recall correctly.

Further, x86 only provides these SIMD operations on 8-bit and 16-bit quantities, not on 32-bit or 64-bit quantities (although all these types can be packed into a SIMD register and operated on with at least some operations; for example regular, non-saturating add).

SSE has a menagerie of non-orthogonal oddities, unfortunately; this is only one of them.

Re:Catching Integer Overflows in C by adobriyan on Tuesday August 02, 2011 @09:03AM (#2836951)
And IIRC it's int16_t/uint16_t max.