3 ms·
It was most likely removed from x86_64 because it is less useful! 32 bit overflows are possible (I have seen them in scientific algorithms for instance). But I
by hbbio 10y ago
It was most likely removed from x86_64 because it is less useful!
32 bit overflows are possible (I have seen them in scientific algorithms for instance). But I have yet to see 64 bits overflows.
- dikaiosune 10y agoOverflows in subtraction remain common for 64-bit unsigned integers, however.
- dspillett 10y agoIt is going to be less common in many fields but I can imagine 64-bit overflow is a concern in a fair few scientific number crunching situations. Also, if you were to use the scaled integer method of representing fractional numbers (like the fixed-point data types seen in many SQL implementations) with a scale factor of 2^32 then a 64-bit overflow is a 32-bit overflow effectively (though I'll admit that is going to be no less rare than genuinely needing the full range of 64-bit integers). And I'm guessing all this applies to underflow as much as it is does to overflow which could be a concern for unsigned integers of any size.