benchmarking with and without "register" storage class specifier

Bernhard Fischer rep.nop at aon.at
Wed Mar 29 13:03:10 UTC 2006


On Wed, Mar 29, 2006 at 07:54:42AM -0500, Robert P. J. Day wrote:
>
>  i did several runs of:
>
>	$ make distclean defconfig
>	$ time make busybox
>
>both before and after applying my two earlier submitted patches that
>removed a majority of the "register" storage class specifiers, and the
>results are indistinguishable in terms of build time.

How is benchmarking *compile-time* of any relevance when it comes to try
to hint the register allocator on what to?
>
>rday



More information about the busybox mailing list