Hal Murray via devel writes: > Are the specs and implementation for IEEE floating point tight enough so that > I should get the exact same result if I run a test on a different CPU > chip?
Formally yes, if you aren't straying into denormals and you keep yourself to elementary operations that actually are specified by the standard. In practise, every code generated for a particular FPU deviates a little from the standard in certain ways that sometimes can be mitigated by setting flags nobody ever uses or even knows about. > Or is there room for things like holding extra bits in temporary results so > the bottom bits might be different due to round off or such? The assumption is that the rounding happens as prescribed in the standard after each operation. Things like FMA, some compiler optimizations and the extra mantissa bits on Intel throw a spanner into this. Regards, Achim. -- +<[Q+ Matrix-12 WAVE#46+305 Neuron microQkb Andromeda XTk Blofeld]>+ Factory and User Sound Singles for Waldorf Q+, Q and microQ: http://Synth.Stromeko.net/Downloads.html#WaldorfSounds _______________________________________________ devel mailing list devel@ntpsec.org http://lists.ntpsec.org/mailman/listinfo/devel