code review 101970047: testing: make benchmarking faster

testing: make benchmarking faster

Allow the number of benchmark iterations to grow faster for fast benchmarks, and don't round up twice.

Using the default benchtime, this CL reduces wall clock time to run benchmarks:

net/http        49s   -> 37s   (-24%)
runtime         8m31s -> 5m55s (-30%)
bytes           2m37s -> 1m29s (-43%)
encoding/json   29s   -> 21s   (-27%)
strings         1m16s -> 53s   (-30%)

[josharian, 2014-06-11 00:51:03]

Hello golang-codereviews@googlegroups.com,

I'd like you to review this change to
https://code.google.com/p/go

[crawshaw, 2014-06-11 19:13:18]

LGTM

Looks like all tests are still run with n=1 and n=101, which is protection
enough against mistiming.

[josharian, 2014-06-12 14:51:32]

*** Submitted as https://code.google.com/p/go/source/detail?r=30307cc8bef2 ***

testing: make benchmarking faster

Allow the number of benchmark iterations to grow faster for fast benchmarks, and
don't round up twice.

Using the default benchtime, this CL reduces wall clock time to run benchmarks:

net/http        49s   -> 37s   (-24%)
runtime         8m31s -> 5m55s (-30%)
bytes           2m37s -> 1m29s (-43%)
encoding/json   29s   -> 21s   (-27%)
strings         1m16s -> 53s   (-30%)

LGTM=crawshaw
R=golang-codereviews, crawshaw
CC=golang-codereviews
https://codereview.appspot.com/101970047

[gobot, 2014-06-12 15:52:17]

This CL appears to have broken the windows-amd64-race builder.
See http://build.golang.org/log/7be4b697f4a924e4f336bb110cfcb5e125ab427e

[rsc, 2014-06-13 05:33:47]

I'm pretty sure this is killing the windows-amd64-race builder.
There are lots of data points now, and it all started with this CL.