Re^5: Our perl/xs/c app is 30% slower with 64bit 5.24.0, than with 32bit 5.8.9. Why?

[dcmertens]

Are you sure these speedups are due to Perl's hash improvements, and not improvements in Perl's IO handling? Because that latter would have been my first guess. A more interesting comparison might be to time the script under two modes, one with a simple counter increment and one with the hash addition. The difference between these two running times would be more illuminating, I think.

[dave_the_m]

There's a tool in the perl src repository which uses cachegrind behind the scenes to accurately measure how many CPU instructions, data reads etc a small snippet of code uses. With the following initial setup (so the hash already exists and has some keys):

my %h = qw(a 1 b 2 c 3 d 4); my $key = "foo";

Running the following benchmark (using a non-constant key so the key's hash gets recalculated each time):

$h{$key} = 1; delete $h{$key}

Shows the following results on various perls:

Key:
    Ir   Instruction read
    Dr   Data read
    Dw   Data write
    COND conditional branches
    IND  indirect branches
    _m   branch predict miss
    _m1  level 1 cache miss
    _mm  last cache (e.g. L3) miss
    -    indeterminate percentage (e.g. 1/0)

The numbers represent raw counts per loop iteration.


       perl589o perl5101o perl5125o perl5144o perl5163o perl5184o perl
+5203o perl5222o perl5240o perl5258o
       -------- --------- --------- --------- --------- --------- ----
+----- --------- --------- ---------
    Ir   1348.0    1340.4    1378.0    1383.0    1423.0    1453.0    1
+466.0    1368.0    1356.0    1300.0
    Dr    414.0     403.0     411.0     404.0     408.0     403.0     
+411.0     379.0     373.0     362.0
    Dw    226.0     214.0     222.0     227.0     228.0     231.0     
+231.0     208.0     206.0     196.0
  COND    202.0     210.1     210.0     204.0     213.0     204.0     
+210.0     199.0     197.0     188.0
   IND     16.0      16.0      17.0      18.0      18.0      18.0     
+ 17.0      14.0      12.0      14.0

COND_m      2.0       1.0       4.0       2.0       3.0       3.0     
+  1.0       2.0       2.0       3.0
 IND_m      9.0       9.0      11.0       9.0       9.0      11.0     
+  9.0       5.0       5.0       5.0

 Ir_m1      0.0       0.0       0.0       0.0       0.0       0.0     
+  0.0      -0.1       0.0       0.0
 Dr_m1      0.0       0.0       0.0       0.0       0.0       0.0     
+  0.0       0.0       0.0       0.0
 Dw_m1      0.0       0.0       0.0       0.0       0.0       0.0     
+  0.0       0.0       0.0       0.0

 Ir_mm      0.0       0.0       0.0       0.0       0.0       0.0     
+  0.0       0.0       0.0       0.0
 Dr_mm      0.0       0.0       0.0       0.0       0.0       0.0     
+  0.0       0.0       0.0       0.0
 Dw_mm      0.0       0.0       0.0       0.0       0.0       0.0     
+  0.0       0.0       0.0       0.0

Which shows everything being much the same before 5.22 (and in particular no significant slowdown in 5.16), and things getting better since.

Dave.