code review 6625057: image/jpeg: move the level-shift and clip out of the id...

image/jpeg: move the level-shift and clip out of the idct function,
to be consistent with the fdct function, and to ease any future
idct rewrites in assembly.

The BenchmarkIDCT delta is obviously just an accounting change and not
a real saving, but it does give an indication of what proportion of
time was spent in the actual IDCT and what proportion was in shift and
clip. The idct time taken is now comparable to fdct.

The BenchmarkFDCT delta is an estimate of benchmark noise.

benchmark                   old ns/op    new ns/op    delta
BenchmarkFDCT                    3842         3837   -0.13%
BenchmarkIDCT                    5611         3478  -38.01%
BenchmarkDecodeRGBOpaque      2932785      2929751   -0.10%

[nigeltao, 2012-10-07 00:11:20]

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

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

[r, 2012-10-07 00:22:25]

LGTM

[nigeltao, 2012-10-07 00:32:08]

*** Submitted as http://code.google.com/p/go/source/detail?r=187194a52289 ***

image/jpeg: move the level-shift and clip out of the idct function,
to be consistent with the fdct function, and to ease any future
idct rewrites in assembly.

The BenchmarkIDCT delta is obviously just an accounting change and not
a real saving, but it does give an indication of what proportion of
time was spent in the actual IDCT and what proportion was in shift and
clip. The idct time taken is now comparable to fdct.

The BenchmarkFDCT delta is an estimate of benchmark noise.

benchmark                   old ns/op    new ns/op    delta
BenchmarkFDCT                    3842         3837   -0.13%
BenchmarkIDCT                    5611         3478  -38.01%
BenchmarkDecodeRGBOpaque      2932785      2929751   -0.10%

R=r
CC=golang-dev
http://codereview.appspot.com/6625057

[nigeltao, 2012-10-07 00:54:38]

On 7 October 2012 11:11,  <nigeltao@golang.org> wrote:
> The BenchmarkIDCT delta is obviously just an accounting change

FYI, I profiled decoding half a dozen different JPEG files, and after
this accounting change, idct is around 8-15% of CPU time for photo
images, and 20-30% for graphics like
$GOROOT/doc/gopher/appenginegopher{,color}.jpg