Why did the 80bit precision calculation go away?
This happened because the C++ Standards committee defined the long double type to be equivalent to the double (64bits each). The long double use to be 80bits, and is supported by the Intel hardware. However, after the new standard, the compiler manufacturers, for some reason, quickly dropped all support for the 80 bit data type. Hopefully as the processing bandwidth of the PC improves, newer data types will be defined and supported. A 124 bit long double (as in other architectures), or even a 256 long long double would be nice. Such a feature is expected to significantly improve the convergence performance of the algorithms used by GSOVLER.
