On Fri, Jun 05, 2015 at 09:09:26AM +0200, Hans Petter Selasky wrote: > Hi, > > I was going through some timer code and found some unnecessary while > loops in kern/kern_clocksource.c . > > I added some prints and found that during boot, "runs" can exceed 2000, > while during regular usage runs is typically 1. Do you think it is worth > to convert these loops into division and multiplications? It might make > the CPU pipeline a tiny bit faster, having to skip some conditionals? > And also possibly improve readability? For typical loop count of one, I think that setup time for the division is much greater than single substraction and non-taken conditional jump, which is also should be consistently predicted as not taken. You could try to run a microbenchmark to see if the patch makes any sense.Received on Fri Jun 05 2015 - 09:38:22 UTC
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:40:58 UTC