Le 06/12/2015 14:09, Andreas Schwab a écrit : > Laurent Vivier <[email protected]> writes: > >> I think qemu-m68k is emulating precisely the FPU. I've checked the >> result of testfloat on qemu-m68k, and moreover I've compared the result >> of softloat on my Quadra 800 (68040 FPU) against the result of qemu-m68k. > > That doesn't implement Motorola 80-bit float. How are the results from > the glibc math tests?
80-bit float is only a compact 96-bit float. In fact this work is 4 years old, so I don't remember all the details. The first FPU bugs have been found when I have started to try to run gtk-demo that uses some FPU functions. I've then of course, tested with the glibc math tests and I think also with gcc tests, especially gfortran tests (I think...) and I have compared the result on a real machine (and fixed the issues). If it is important for you to know the accuracy of the emulation, I can try to re-run all these stuffs. Laurent
