This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: One line performance tweak


    Each instruction contains only a constant J number of pseudos.

    Therefore, if you can scan all at once, you do only J * N work where N
    is the number of instructions.

Yes, but the "scan" is against a list that's O(N) with the assumption that
the number of pseudos put into memory is linear.

Or is the idea that you can do the check by referencing some (e.g.) bitmap
that says which pseudos need to be put into memory and that converts it
to linear?

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]