G'day Danny,

I can confirm and reproduce the types of results you've presented. I have Cygwin running on Win10 with the latest updates as of a few days ago.

$ uname -a
CYGWIN_NT-10.0-19045 titan 3.4.10-1.x86_64 2023-11-29 12:12 UTC x86_64
+ Cygwin
[download]

I have a variety of Perl versions installed using Perlbrew:

$ perlbrew list
 * perl-5.39.3
   perl-5.38.0
   perl-5.36.0
   perl-5.34.0
   perl-5.33.5
   perl-5.32.0
   perl-5.30.0
[download]

I used perl-5.39.3, perl-5.36.0 and perl-5.30.0 for the tests below.

I didn't know too much about your test data. I created these:

• test_data

  $ perl -E 'my $test_rec = "X" x 45; open my $fh, ">", "test_data"; $fh
  +->say($test_rec) for 1..769114;'
  
  $ ls -l test_data
  -rw-r--r-- 1 ken None 35379244 Jan 23 17:09 test_data
  
  $ wc test_data
    769114   769114 35379244 test_data
  
  $ head -1 test_data
  XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
  [download]

• test_data_Q

  $ perl -E 'my $test_rec = "Query" . "X" x 40; open my $fh, ">", "test_
  +data_Q"; $fh->say($test_rec) for 1..769114;'
  
  $ ls -l test_data_Q
  -rw-r--r-- 1 ken None 35379244 Jan 23 17:23 test_data_Q
  
  $ wc test_data_Q
    769114   769114 35379244 test_data_Q
  
  $ head -1 test_data_Q
  QueryXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
  [download]

I made a verbatim copy of your posted code: "op_code.pl". Here's the test results:

v5.39.3

$ perl -v | head -2 | tail -1
This is perl 5, version 39, subversion 3 (v5.39.3) built for cygwin-th
+read-multi

$ ./op_code.pl test_data
0.111368 read lines from disk and do RE.
0.096441 read lines from in-memory file and do RE.

$ ./op_code.pl test_data_Q
0.356852 read lines from disk and do RE.
398.759730 read lines from in-memory file and do RE.
[download]

v5.36.0

$ perl -v | head -2 | tail -1
This is perl 5, version 36, subversion 0 (v5.36.0) built for cygwin-th
+read-multi

$ ./op_code.pl test_data
0.118364 read lines from disk and do RE.
0.094379 read lines from in-memory file and do RE.

$ ./op_code.pl test_data_Q
0.362871 read lines from disk and do RE.
399.115542 read lines from in-memory file and do RE.
[download]

v5.30.0

$ perl -v | head -2 | tail -1
This is perl 5, version 30, subversion 0 (v5.30.0) built for cygwin-th
+read-multi

$ ./op_code.pl test_data
0.112352 read lines from disk and do RE.
0.096379 read lines from in-memory file and do RE.

$ ./op_code.pl test_data_Q
0.327187 read lines from disk and do RE.
398.730810 read lines from in-memory file and do RE.
[download]

I have ~32GB memory available; at no time during the running of these tests was more than 6.5GB used.

I see throughout this thread that both you and others have not encountered this anomaly on other operating systems: I can't readily discern a reason for it.