SKPMFloat: we can beat the naive loops when clamping

SKPMFloat: we can beat the naive loops when clamping

Clamping 4 at a time is now about 15% faster than 1 at a time with SSSE3.
Clamping 4 at a time is now about 20% faster with SSE2,
and this applies to non-clamping too (we still just clamp there).

In all cases, 4 at a time is never worse than 1 at a time,
and not clamping is never slower than clamping.

Here's all the bench results, with the numbers for portable code as a fun point
of reference:

SSSE3:
maxrss  loops   min median  mean    max stddev  samples     config  bench
  10M   2291    4.66ns  4.66ns  4.66ns  4.68ns  0%  ▆█▁▁▁▇▁▇▁▃  nonrendering    SkPMFloat_get_1x
  10M   2040    5.29ns  5.3ns   5.3ns   5.32ns  0%  ▃▆▃▃▁▁▆▃▃█  nonrendering    SkPMFloat_clamp_1x
  10M   7175    4.62ns  4.62ns  4.62ns  4.63ns  0%  ▁▄▃████▃▄▇  nonrendering    SkPMFloat_get_4x
  10M   5801    4.89ns  4.89ns  4.89ns  4.91ns  0%  █▂▄▃▁▃▄█▁▁  nonrendering    SkPMFloat_clamp_4x

SSE2:
maxrss  loops   min median  mean    max stddev  samples     config  bench
  10M   1601    6.02ns  6.05ns  6.04ns  6.08ns  0%  █▅▄▅▄▂▁▂▂▂  nonrendering    SkPMFloat_get_1x
  10M   2918    6.05ns  6.06ns  6.05ns  6.06ns  0%  ▂▇▁▇▇▁▇█▇▂  nonrendering    SkPMFloat_clamp_1x
  10M   3569    5.43ns  5.45ns  5.44ns  5.45ns  0%  ▄█▂██▇▁▁▇▇  nonrendering    SkPMFloat_get_4x
  10M   4168    5.43ns  5.43ns  5.43ns  5.44ns  0%  █▄▇▁▇▄▁▁▁▁  nonrendering    SkPMFloat_clamp_4x

Portable:
maxrss  loops   min median  mean    max stddev  samples     config  bench
  10M   500     27.8ns  28.1ns  28ns    28.2ns  0%  ▃█▆▃▇▃▆▁▇▂  nonrendering    SkPMFloat_get_1x
  10M   770     40.1ns  40.2ns  40.2ns  40.3ns  0%  ▅▁▃▂▆▄█▂▅▂  nonrendering    SkPMFloat_clamp_1x
  10M   1269    28.4ns  28.8ns  29.1ns  32.7ns  4%  ▂▂▂█▂▁▁▂▁▁  nonrendering    SkPMFloat_get_4x
  10M   1439    40.2ns  40.4ns  40.4ns  40.5ns  0%  ▆▆▆█▁▆▅█▅▆  nonrendering    SkPMFloat_clamp_4x

SkPMFloat_neon.h is still one big TODO as far as 4-at-a-time APIs go.

BUG=skia:

Committed: https://skia.googlesource.com/skia/+/91fd7371ec80724ec53aae8f2d5a6753499d8963

[mtklein_C, 2015-03-05 20:03:26]

mtklein@chromium.org changed reviewers:
+ msarett@google.com, reed@google.com

[mtklein_C, 2015-03-05 20:03:26]

Turns out I was looking in the wrong places to beat the naive loop.  I still
haven't beaten it for From4PMColors or To4PMColors, but I have pretty soundly
for ClampTo4PMColors.

[msarett, 2015-03-06 13:48:03]

lgtm

I'm still looking to see if we can beat the naive loops.  One issue I'm having
is that my desktop is running the SSE2 version.  It looks like I have SSSE3 and
SSE4, so I'm not sure why that would be the case.

[mtklein, 2015-03-06 13:58:36]

On 2015/03/06 13:48:03, msarett wrote:
> lgtm
> 
> I'm still looking to see if we can beat the naive loops.  One issue I'm having
> is that my desktop is running the SSE2 version.  It looks like I have SSSE3
and
> SSE4, so I'm not sure why that would be the case.

We don't set -march=native by default:
  - Windows + Linux assume SSE2;
  - Mac and Android x86 assume SSSE3;
  - NEON depends on the device.

If you want to bump yourself up to whatever your machine supports, I'd do:
$ export CC="$CC -march=native"
$ export CXX="$CXX -march=native"

Or if CC and CXX aren't already defined, get things started:
$ export CC="$gcc -march=native"
$ export CXX="$g++ -march=native"

Then run ./gyp_skia to pick up the changes.

[mtklein, 2015-03-06 13:58:40]

The CQ bit was checked by mtklein@google.com

[commit-bot: I haz the power, 2015-03-06 14:10:19]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/982123002/20001

[commit-bot: I haz the power, 2015-03-06 14:15:47]

Committed patchset #2 (id:20001) as
https://skia.googlesource.com/skia/+/91fd7371ec80724ec53aae8f2d5a6753499d8963