code review 5491064: strconv: reduce buffer size for multi-precision decimals.

strconv: reduce buffer size for multi-precision decimals.

The longest numbers we have to represent are the smallest denormals.
Their decimal mantissa is not longer than 5^1100. Taking into
account some extra size for in-place operations, 800 digits are
enough. This saves time used for zero intiialization of extra
bytes.

                                        old ns/op  new ns/op    delta
strconv_test.BenchmarkAtof64Decimal           521        334   -35.9%
strconv_test.BenchmarkAtof64Float             572        391   -31.6%
strconv_test.BenchmarkAtof64FloatExp        10242      10036    -2.0%
strconv_test.BenchmarkAtof64Big              4229       4029    -4.7%
strconv_test.BenchmarkFormatFloatDecimal     1396        934   -33.1%
strconv_test.BenchmarkFormatFloat            4295       3341   -22.2%
strconv_test.BenchmarkFormatFloatExp        12035      11181    -7.1%
strconv_test.BenchmarkFormatFloatBig         4213       3229   -23.4%
strconv_test.BenchmarkAppendFloatDecimal     1031        600   -41.8%
strconv_test.BenchmarkAppendFloat            3971       3044   -23.3%
strconv_test.BenchmarkAppendFloatExp        11699      11003    -5.9%
strconv_test.BenchmarkAppendFloatBig         3836       2915   -24.0%

[remyoudompheng, 2011-12-16 23:08:23]

Hello golang-dev@googlegroups.com (cc: golang-dev@googlegroups.com,
remy@archlinux.org),

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

[bradfitz, 2011-12-16 23:10:48]

Update the TODO that's in that struct?

On Fri, Dec 16, 2011 at 3:08 PM, <remyoudompheng@gmail.com> wrote:

> Reviewers: golang-dev_googlegroups.com,
>
> Message:
> Hello golang-dev@googlegroups.com (cc: golang-dev@googlegroups.com,
> remy@archlinux.org),
>
> I'd like you to review this change to
> https://go.googlecode.com/hg/
>
>
> Description:
> strconv: reduce buffer size for multi-precision decimals.
>
> The longest numbers we have to represent are the smallest denormals.
> Their decimal mantissa is not longer than 5^1100. Taking into
> account some extra size for in-place operations, 800 digits are
> enough. This saves time used for zero intiialization of extra
> bytes.
>
>                                        old ns/op  new ns/op    delta
> strconv_test.**BenchmarkAtof64Decimal           521        334   -35.9%
> strconv_test.**BenchmarkAtof64Float             572        391   -31.6%
> strconv_test.**BenchmarkAtof64FloatExp        10242      10036    -2.0%
> strconv_test.**BenchmarkAtof64Big              4229       4029    -4.7%
> strconv_test.**BenchmarkFormatFloatDecimal     1396        934   -33.1%
> strconv_test.**BenchmarkFormatFloat            4295       3341   -22.2%
> strconv_test.**BenchmarkFormatFloatExp        12035      11181    -7.1%
> strconv_test.**BenchmarkFormatFloatBig         4213       3229   -23.4%
> strconv_test.**BenchmarkAppendFloatDecimal     1031        600   -41.8%
> strconv_test.**BenchmarkAppendFloat            3971       3044   -23.3%
> strconv_test.**BenchmarkAppendFloatExp        11699      11003    -5.9%
> strconv_test.**BenchmarkAppendFloatBig         3836       2915   -24.0%
>
> Please review this at
http://codereview.appspot.com/**5491064/<http://codereview.appspot.com/5491064/>
>
> Affected files:
>  M src/pkg/strconv/decimal.go
>
>
> Index: src/pkg/strconv/decimal.go
> ==============================**==============================**=======
> --- a/src/pkg/strconv/decimal.go
> +++ b/src/pkg/strconv/decimal.go
> @@ -14,9 +14,9 @@
>  type decimal struct {
>        // TODO(rsc): Can make d[] a bit smaller and add
>        // truncated bool;
> -       d   [2000]byte // digits
> -       nd  int        // number of digits used
> -       dp  int        // decimal point
> +       d   [800]byte // digits
> +       nd  int       // number of digits used
> +       dp  int       // decimal point
>        neg bool
>  }
>
>
>
>

[rsc, 2011-12-16 23:12:11]

The TODO is for a different idea, which may or may not be possible.

[rsc, 2011-12-19 20:03:50]

LGTM

[rsc, 2011-12-19 20:03:57]

*** Submitted as http://code.google.com/p/go/source/detail?r=ac6c5970cc33 ***

strconv: reduce buffer size for multi-precision decimals.

The longest numbers we have to represent are the smallest denormals.
Their decimal mantissa is not longer than 5^1100. Taking into
account some extra size for in-place operations, 800 digits are
enough. This saves time used for zero intiialization of extra
bytes.

                                        old ns/op  new ns/op    delta
strconv_test.BenchmarkAtof64Decimal           521        334   -35.9%
strconv_test.BenchmarkAtof64Float             572        391   -31.6%
strconv_test.BenchmarkAtof64FloatExp        10242      10036    -2.0%
strconv_test.BenchmarkAtof64Big              4229       4029    -4.7%
strconv_test.BenchmarkFormatFloatDecimal     1396        934   -33.1%
strconv_test.BenchmarkFormatFloat            4295       3341   -22.2%
strconv_test.BenchmarkFormatFloatExp        12035      11181    -7.1%
strconv_test.BenchmarkFormatFloatBig         4213       3229   -23.4%
strconv_test.BenchmarkAppendFloatDecimal     1031        600   -41.8%
strconv_test.BenchmarkAppendFloat            3971       3044   -23.3%
strconv_test.BenchmarkAppendFloatExp        11699      11003    -5.9%
strconv_test.BenchmarkAppendFloatBig         3836       2915   -24.0%

R=golang-dev, bradfitz, rsc
CC=golang-dev, remy
http://codereview.appspot.com/5491064

Committer: Russ Cox <rsc@golang.org>