> > It depends on how you do it, on the architecture,
> > and on what you measure. Lmbench is an
> > invaluable hint, but it is not definitive. If kernel compiles speed
> > up by 20%, you knpw something.
>
> Assuming this is on PowerPC, how much of this saving comes from avoiding
> flush_page_to_ram()?
I can give figures for flush_page_to_ram() on MIPS where a cache
instructions costs like 13 cycles + eventual pipeline stalls due to
SysAD bus contention. Not counting loop overhead that's at least 1664
cycles or even 3228 cycles on machines with 16 byte l1 caches. Definately
worth to take care of.
Ralf
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/