This patch increases the efficiency of extraction of the new patches in
patch index update.
I have tested its effectiveness with gh's benchmark.
The new values:
==> timings-proto/PI_add_disable-patch-index <==
0.28
==> timings-proto/PI_add_patch-index <==
0.31
==> timings-proto/PI_amend_disable-patch-index <==
21.21
==> timings-proto/PI_amend_patch-index <==
25.43
==> timings-proto/PI_obliterate_disable-patch-index <==
18.80
==> timings-proto/PI_obliterate_patch-index <==
21.95
==> timings-proto/PI_record_disable-patch-index <==
115.16
==> timings-proto/PI_record_patch-index <==
144.02
The old values:
==> timings/PI_add_disable-patch-index <==
0.28
==> timings/PI_add_patch-index <==
0.31
==> timings/PI_amend_disable-patch-index <==
21.45
==> timings/PI_amend_patch-index <==
27.25
==> timings/PI_obliterate_disable-patch-index <==
18.50
==> timings/PI_obliterate_patch-index <==
22.25
==> timings/PI_record_disable-patch-index <==
117.02
==> timings/PI_record_patch-index <==
150.38
Clearly, I have not been able to reproduce the large pi update
inefficiency. So I suspect this patch will not help much. :-)
Nonetheless, It is still a (small) improvement.
Attachments
|