[Pixman] [PATCH v2 00/11] Implement more vmx fast paths

[Oded Gabbay]

Hi,

This is v2 of the vmx fast paths patch-set.

The following are the changes:

- Add 3 patches that were sent as a seperate patch-set. They are:
  vmx: implement fast path composite_over_8888_8888
  vmx: implement fast path composite_add_8_8
  vmx: implement fast path composite_add_8888_8888

- Removed [PATCH 07/12] vmx: implement fast path vmx_composite_over_n_8_8888 
  because it produced more slowdowns than speedups in cairo

- Removed [PATCH 11/12] vmx: implement fast path iterator vmx_fetch_r5g6b5 
  because it caused performance regression in "lowlevel-blt-bench
  src_0565_8888" test on Playstation3:

- Removed a not-used helper function expand_pixel_32_1x128()

- Removed [PATCH 04/12] vmx: implement fast path vmx_blt 
  and [PATCH 05/12] vmx: implement fast path vmx_composite_copy_area
  because they caused performance regression in "lowlevel-blt-bench
  src_8888_8888" test on POWER8 ppc64le and POWER7 ppc64

- Added missing benchamrks in some of the patches

The original cover letter:
http://lists.freedesktop.org/archives/pixman/2015-July/003771.html

Cover letter of the other patch-set that I added to this patch-set:
http://lists.freedesktop.org/archives/pixman/2015-June/003751.html

Oded Gabbay (11):
  vmx: add LOAD_VECTOR macro
  vmx: add helper functions
  vmx: implement fast path vmx_fill
  vmx: implement fast path composite_over_8888_8888
  vmx: implement fast path composite_add_8_8
  vmx: implement fast path composite_add_8888_8888
  vmx: implement fast path vmx_composite_over_n_8888_8888_ca
  vmx: implement fast path vmx_composite_src_x888_8888
  vmx: implement fast path scaled nearest vmx_8888_8888_OVER
  vmx: implement fast path iterator vmx_fetch_x8r8g8b8
  vmx: implement fast path iterator vmx_fetch_a8

 pixman/pixman-vmx.c | 1183 +++++++++++++++++++++++++++++++++++++++++++++++++--
 1 file changed, 1157 insertions(+), 26 deletions(-)

-- 
2.4.3