On Windows, __builtin_clzl() only handles 32-bit inputs due to LLP64. This causes problems when larger values are used. Changing this to __builtin_clzll() fixes the crashes. This most notably fixes the crash when running the demo scope on Windows, as next_pow2() is used in the code that performs the downscaling using FFTS.
On Windows, __builtin_clzl() only handles 32-bit inputs due to LLP64. This causes problems when larger values are used. Changing this to __builtin_clzll() fixes the crashes. This most notably fixes the crash when running the demo scope on Windows, as next_pow2() is used in the code that performs the downscaling using FFTS.