On 2/5/19 11:56 PM, ToddAndMargo via perl6-users wrote:
On 2/5/19 8:34 PM, ToddAndMargo via perl6-users wrote:


On Sun, Feb 3, 2019 at 9:36 PM ToddAndMargo via perl6-users <perl6-users@perl.org <mailto:perl6-users@perl.org>> wrote:

    Hi All,

    If I have a variable of type Buf which 10000000 bytes in it
    and I find the five bytes I want, is it faster, slower,
    or no difference in speed to overwrite the same variable
    with the five bytes?  Or is it faster to put the five bytes
    from the first variable into a second variable?

    Many thanks,
    -T


On 2/5/19 8:42 AM, yary wrote:
There are modules to time two pieces of code and show the difference
https://github.com/perl6-community-modules/perl6-Benchy
https://github.com/tony-o/perl6-bench

You can write up the two versions you're thinking of, feed them to the benchmark module, and show us what you find!

-y


Hi Yary,

Thank you!

Apparently, overwriting the original buffer only change the
structures pointers, which is almost instantaneous.

And you taught me something new today!

-T


<code VarTest.pl6>

#!/usr/bin/env perl6

use Bench;

my IO::Handle $HaystackHandle = open( "/home/temp/procexp64.exe", :bin, :ro );
my Buf $Haystack              = $HaystackHandle.read( 3000000 );
$HaystackHandle.close;

my Buf $Needle;

my $b = Bench.new;

sub Another() { $Needle   = $Haystack.subbuf( 0x14FFAC .. 0x145FAC ); }
sub Same()    { $Haystack = $Haystack.subbuf( 0x14FFAC .. 0x145FAC ); }

say "first copies to a new variable; second overwrites the same variable";
$b.timethese( 100000, {
   first  => sub { Another; },
   second => sub { Same; },
});

</code VarTest.pl6>


$ VarTest.pl6
first copies to a new variable; second overwrites the same variable
Benchmark:
Timing 100000 iterations of first, second...
      first: 0.021 wallclock secs (0.021 usr 0.000 sys 0.021 cpu) @ 4676612.262/s (n=100000)
     second: 0.000 wallclock secs (0.000 usr 0.000 sys 0.000 cpu)


You guys catch my mistake?  It is only valid for one iteration.
Chuckle.  A rewrite is in order.


<code VarTest.pl6>
#!/usr/bin/env perl6

use Bench;

my IO::Handle $HaystackHandle = open( "/home/temp/procexp64.exe", :bin, :ro );
my Buf $Haystack              = $HaystackHandle.read( 3000000 );
$HaystackHandle.close;

my Buf $SubBuf;
my $b = Bench.new;

sub Another() { my Buf $b = $Haystack; my Buf $Needle = $b.subbuf( 0x14FFAC .. 0x145FAC ); } sub Same() { my Buf $b = $Haystack; $b = $b.subbuf( 0x14FFAC .. 0x145FAC ); }

say "first copies to a new variable; second overwrites the same variable";
$b.timethese( 100000, {
  first  => sub { $SubBuf = Another(); },
  second => sub { $SubBuf = Same(); },
});
</code VarTest.pl6>



$ VarTest.pl6
first copies to a new variable; second overwrites the same variable
Benchmark:
Timing 100000 iterations of first, second...
first: 0.051 wallclock secs (0.054 usr 0.003 sys 0.057 cpu) @ 1965370.177/s (n=100000) second: 0.052 wallclock secs (0.045 usr 0.001 sys 0.046 cpu) @ 1926856.526/s (n=100000)


Virtually no difference

Reply via email to