Another point is that Benchmark.pm tries hard to ignore most of the DB/network time so you need to look at the elapsed time. For a second I was puzzled because the second number was obscured:
new_u: 8 wallclock secs (n=508) old_u: 7 wallclock secs (n=217) new_u: 5 wallclock secs (n=508) old_u: 4 wallclock secs (n=227) new_u: 5 wallclock secs (n=528) old_u: 3 wallclock secs (n=227)
But after dividing, the wall-clock rates are (rather inaccurately):
new_u 63.5 old_u 31 new_u 101.6 old_u 56.75 new_u 105.6 old_u 75.67
Good work.
- tye
In reply to Re^5: patchable settings (elapsed time)
by tye
in thread patchable settings
by ysth
| For: | Use: | ||
| & | & | ||
| < | < | ||
| > | > | ||
| [ | [ | ||
| ] | ] |