lkml.org 
[lkml]   [2002]   [Oct]   [24]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
    /
    Date
    From
    SubjectRe: [CFT] faster athlon/duron memory copy implementation
    Robert Love wrote:
    > On Thu, 2002-10-24 at 16:09, Ed Sweetman wrote:
    >
    >
    >>I seem to be seeing compiler optimizations come into play with the
    >>numbers and not any mention of them that i've seen has been talked
    >>about. That could be causing any discrepencies with predicted values. So
    >>not only would we have to look at algorithms, but also the compilers and
    >>what optimizations we plan on using them with. Some do better on
    >>certain compilers+flags than others. It's a mixmatch that seems to only
    >>get complicated the more realistic you make it.
    >
    >
    > The majority of the program is inline assembly so I do not think
    > compiler is playing a huge role here.
    >
    > Regardless, the numbers are all pretty uniform in saying the new no
    > prefetch method is superior so its a mute point.
    >
    > Robert Love

    With gcc 3.x i get

    495MB/s with -O3 -march=athlon-tbird -mcpu=athlon-tbird -falign-loops=4
    -falign-functions=4

    488MB/s with -O3 -march=athlon-tbird -mcpu=athlon-tbird -falign-loops=4

    467MB/s with -O0 -march=i686 -mcpu=i686

    which is almost a 30MB/s difference or 6% simply from compiler options
    of the same compiler. It may not mean much in 1 second. But few things
    where we care about performance are only run for one second.

    I'd expect something below 3% and realistically closer to 1%. Any ideas
    as to why it is making a difference? Does the execution path to the
    function in C really take up performance to drop 30MB/s of memory
    bandwidth because from the looks of it this program is very small and
    things should be really quick to the asm functions.

    -
    To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
    the body of a message to majordomo@vger.kernel.org
    More majordomo info at http://vger.kernel.org/majordomo-info.html
    Please read the FAQ at http://www.tux.org/lkml/

    \
     
     \ /
      Last update: 2005-03-22 13:30    [W:5.233 / U:0.064 seconds]
    ©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site