code review 5496070: encoding/binary: add more benchmarks

encoding/binary: add more benchmarks
Also add a byte count to the varint benchmarks - this
isn't accurate, of course, but it allows a rough comparison to
the other benchmarks.

[rog, 2011-12-19 08:33:13]

Hello golang-dev@googlegroups.com,

I'd like you to review this change to
https://go.googlecode.com/hg/

[r, 2011-12-19 19:13:40]

http://codereview.appspot.com/5496070/diff/5001/src/pkg/encoding/binary/binar...
File src/pkg/encoding/binary/binary_test.go (right):

http://codereview.appspot.com/5496070/diff/5001/src/pkg/encoding/binary/binar...
src/pkg/encoding/binary/binary_test.go:174: func BenchmarkReadArray(b
*testing.B) {
ReadSlice?
also it's nice in the benchmark output to explain the test better, so I suggest
BenchmarkReadSlice1000 or even BenchmarkReadSlice1000int32s.
(i realize the other examples in this file don't self-identify as well)

http://codereview.appspot.com/5496070/diff/5001/src/pkg/encoding/binary/binar...
src/pkg/encoding/binary/binary_test.go:203: func BenchmarkReadInt(b *testing.B)
{
s/Int/Ints/ (at least) 
the number it will print is not the time to read an int

http://codereview.appspot.com/5496070/diff/5001/src/pkg/encoding/binary/binar...
src/pkg/encoding/binary/binary_test.go:234: func BenchmarkWrite(b *testing.B) {
this is an odd test. not your fault, but it's even odder than read.
do we expect the overhead of an indirection to matter? if so, split it out. if
not, make it the same complexity as Read. also call this WriteInts or whatever;
align it with Read.

[rog, 2011-12-20 10:50:44]

PTAL

http://codereview.appspot.com/5496070/diff/5001/src/pkg/encoding/binary/binar...
File src/pkg/encoding/binary/binary_test.go (right):

http://codereview.appspot.com/5496070/diff/5001/src/pkg/encoding/binary/binar...
src/pkg/encoding/binary/binary_test.go:174: func BenchmarkReadArray(b
*testing.B) {
On 2011/12/19 19:13:40, r wrote:
> ReadSlice?
> also it's nice in the benchmark output to explain the test better, so I
suggest
> BenchmarkReadSlice1000 or even BenchmarkReadSlice1000int32s.
> (i realize the other examples in this file don't self-identify as well)

Done.

http://codereview.appspot.com/5496070/diff/5001/src/pkg/encoding/binary/binar...
src/pkg/encoding/binary/binary_test.go:203: func BenchmarkReadInt(b *testing.B)
{
On 2011/12/19 19:13:40, r wrote:
> s/Int/Ints/ (at least) 
> the number it will print is not the time to read an int

Done.

http://codereview.appspot.com/5496070/diff/5001/src/pkg/encoding/binary/binar...
src/pkg/encoding/binary/binary_test.go:234: func BenchmarkWrite(b *testing.B) {
On 2011/12/19 19:13:40, r wrote:
> this is an odd test. not your fault, but it's even odder than read.
> do we expect the overhead of an indirection to matter? if so, split it out. if
> not, make it the same complexity as Read. also call this WriteInts or
whatever;
> align it with Read.

Done.

[r, 2011-12-20 16:49:37]

http://codereview.appspot.com/5496070/diff/3004/src/pkg/encoding/binary/binar...
File src/pkg/encoding/binary/binary_test.go (right):

http://codereview.appspot.com/5496070/diff/3004/src/pkg/encoding/binary/binar...
src/pkg/encoding/binary/binary_test.go:198: if !reflect.DeepEqual(s, t) {
do you want to stop the timer before this? or test it once beforehand?

http://codereview.appspot.com/5496070/diff/3004/src/pkg/encoding/binary/binar...
src/pkg/encoding/binary/binary_test.go:229: if !reflect.DeepEqual(ls, want) {
ditto.

http://codereview.appspot.com/5496070/diff/3004/src/pkg/encoding/binary/binar...
src/pkg/encoding/binary/binary_test.go:251: if !bytes.Equal(buf.Bytes(),
big[:30]) {
ditto

[rog, 2011-12-20 17:13:47]

PTAL

http://codereview.appspot.com/5496070/diff/3004/src/pkg/encoding/binary/binar...
File src/pkg/encoding/binary/binary_test.go (right):

http://codereview.appspot.com/5496070/diff/3004/src/pkg/encoding/binary/binar...
src/pkg/encoding/binary/binary_test.go:198: if !reflect.DeepEqual(s, t) {
On 2011/12/20 16:49:37, r wrote:
> do you want to stop the timer before this? or test it once beforehand?

Done.

http://codereview.appspot.com/5496070/diff/3004/src/pkg/encoding/binary/binar...
src/pkg/encoding/binary/binary_test.go:229: if !reflect.DeepEqual(ls, want) {
On 2011/12/20 16:49:37, r wrote:
> ditto.

Done.

http://codereview.appspot.com/5496070/diff/3004/src/pkg/encoding/binary/binar...
src/pkg/encoding/binary/binary_test.go:251: if !bytes.Equal(buf.Bytes(),
big[:30]) {
On 2011/12/20 16:49:37, r wrote:
> ditto

Done.

[r, 2011-12-20 17:25:51]

*** Submitted as c5b714cc9df6 ***

encoding/binary: add more benchmarks
Also add a byte count to the varint benchmarks - this
isn't accurate, of course, but it allows a rough comparison to
the other benchmarks.

R=golang-dev, r
CC=golang-dev
http://codereview.appspot.com/5496070

Committer: Rob Pike <r@golang.org>