Le 08/07/2012 22:46, Emil Lenngren a écrit : > Hmm. I just checked it up. Null pointer comparison ends up in a "cmp" > instruction, and bit test ends up in a "test" instruction. Both take the > same number of clock cycles.. > > So in theory it should make no difference at all. > Maybe the cache changes something then? -- Benoît Minisini