This provides a significant speedup for encoding and decoding when
using many CPU cores.

name                        old time/op  new time/op  delta
EndToEndPipe                5.26µs ± 2%  5.38µs ± 7%     ~     (p=0.121 n=8+7)
EndToEndPipe-6              1.86µs ± 5%  1.80µs ±11%     ~     (p=0.442 n=8+8)
EndToEndPipe-48             1.39µs ± 2%  1.41µs ± 4%     ~     (p=0.645 n=8+8)
EndToEndByteBuffer          1.54µs ± 5%  1.57µs ± 5%     ~     (p=0.130 n=8+8)
EndToEndByteBuffer-6         620ns ± 6%   310ns ± 8%  -50.04%  (p=0.000 n=8+8)
EndToEndByteBuffer-48        506ns ± 4%   110ns ± 3%  -78.22%  (p=0.000 n=8+8)
EndToEndSliceByteBuffer      149µs ± 3%   153µs ± 5%   +2.80%  (p=0.021 n=8+8)
EndToEndSliceByteBuffer-6    103µs ±17%    31µs ±12%  -70.06%  (p=0.000 n=8+8)
EndToEndSliceByteBuffer-48  93.2µs ± 2%  18.0µs ± 5%  -80.66%  (p=0.000 n=7+8)
EncodeComplex128Slice       20.6µs ± 5%  20.9µs ± 8%     ~     (p=0.959 n=8+8)
EncodeComplex128Slice-6     4.10µs ±10%  3.75µs ± 8%   -8.58%  (p=0.004 n=8+7)
EncodeComplex128Slice-48    1.14µs ± 2%  0.81µs ± 2%  -28.98%  (p=0.000 n=8+8)
EncodeFloat64Slice          10.2µs ± 7%  10.1µs ± 6%     ~     (p=0.694 n=7+8)
EncodeFloat64Slice-6        2.01µs ± 6%  1.80µs ±11%  -10.30%  (p=0.004 n=8+8)
EncodeFloat64Slice-48        701ns ± 3%   408ns ± 2%  -41.72%  (p=0.000 n=8+8)
EncodeInt32Slice            11.8µs ± 7%  11.7µs ± 6%     ~     (p=0.463 n=8+7)
EncodeInt32Slice-6          2.32µs ± 4%  2.06µs ± 5%  -10.89%  (p=0.000 n=8+8)
EncodeInt32Slice-48          731ns ± 2%   445ns ± 2%  -39.10%  (p=0.000 n=7+8)
EncodeStringSlice           9.13µs ± 9%  9.18µs ± 8%     ~     (p=0.798 n=8+8)
EncodeStringSlice-6         1.91µs ± 5%  1.70µs ± 5%  -11.07%  (p=0.000 n=8+8)
EncodeStringSlice-48         679ns ± 3%   397ns ± 3%  -41.50%  (p=0.000 n=8+8)
EncodeInterfaceSlice         449µs ±11%   461µs ± 9%     ~     (p=0.328 n=8+8)
EncodeInterfaceSlice-6       503µs ± 7%    88µs ± 7%  -82.51%  (p=0.000 n=7+8)
EncodeInterfaceSlice-48      335µs ± 8%    22µs ± 1%  -93.55%  (p=0.000 n=8+7)
DecodeComplex128Slice       67.2µs ± 4%  67.0µs ± 6%     ~     (p=0.721 n=8+8)
DecodeComplex128Slice-6     22.0µs ± 8%  18.9µs ± 5%  -14.44%  (p=0.000 n=8+8)
DecodeComplex128Slice-48    46.8µs ± 3%  34.9µs ± 3%  -25.48%  (p=0.000 n=8+8)
DecodeFloat64Slice          39.4µs ± 4%  40.3µs ± 3%     ~     (p=0.105 n=8+8)
DecodeFloat64Slice-6        16.1µs ± 2%  11.2µs ± 7%  -30.64%  (p=0.001 n=6+7)
DecodeFloat64Slice-48       38.1µs ± 3%  24.0µs ± 7%  -37.10%  (p=0.000 n=8+8)
DecodeInt32Slice            39.1µs ± 4%  40.1µs ± 5%     ~     (p=0.083 n=8+8)
DecodeInt32Slice-6          16.3µs ±21%  10.6µs ± 1%  -35.17%  (p=0.000 n=8+7)
DecodeInt32Slice-48         36.5µs ± 6%  21.9µs ± 9%  -39.89%  (p=0.000 n=8+8)
DecodeStringSlice           82.9µs ± 6%  85.5µs ± 5%     ~     (p=0.121 n=8+7)
DecodeStringSlice-6         32.4µs ±11%  26.8µs ±16%  -17.37%  (p=0.000 n=8+8)
DecodeStringSlice-48        76.0µs ± 2%  57.0µs ± 5%  -25.02%  (p=0.000 n=8+8)
DecodeInterfaceSlice         718µs ± 4%   752µs ± 5%   +4.83%  (p=0.038 n=8+8)
DecodeInterfaceSlice-6       500µs ± 6%   165µs ± 7%  -66.95%  (p=0.000 n=7+8)
DecodeInterfaceSlice-48      470µs ± 5%   120µs ± 6%  -74.55%  (p=0.000 n=8+7)
DecodeMap                   3.29ms ± 5%  3.34ms ± 5%     ~     (p=0.279 n=8+8)
DecodeMap-6                 7.73ms ± 8%  7.53ms ±18%     ~     (p=0.779 n=7+8)
DecodeMap-48                7.46ms ± 6%  7.71ms ± 3%     ~     (p=0.161 n=8+8)

https://perf.golang.org/search?q=upload:20170426.4

Change-Id: I335874028ef8d7c991051004f8caadd16c92d5cc
Reviewed-on: https://go-review.googlesource.com/41872
Run-TryBot: Bryan Mills <bcmills@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>