Bug 5386 – Initialising out float parameter causes FPU exception

The following short code snippet causes slow code caused by FPU exception: float f; func(f); void func(out float ff) { ff = 1; } The following code is produced for the function: func LABEL NEAR push ebp mov ebp, esp push eax fld dword ptr [_nan] fstp dword ptr [eax] fld dword ptr [_one] fstp dword ptr [eax] mov esp, ebp pop ebp ret It appears that in DMD2 a float NaN is 0x7FE00000 (in dword format) but when it initialises a float 'out' parameter it initialises it with 0x7FA00000. This causes an FPU trap which unnecessarily eats up around 250 ticks. I had a quick look at doubles too. This uses 7FF4000000000000 to initialise a double out parameter which also traps the FPU.

I see no such behavior, float is initialized with 7FC0.