95c2e5532b
BlitRow565: new NEON version of S32_D565_Blend This new implementation brings a good speedup in most cases and gives exact results (removes one mismatch in gm). Here are the benchmark results (speedup vs. existing S32A_D565_Blend): +-------+-----------+------------+ | count | Cortex-A9 | Cortex-A15 | +-------+-----------+------------+ | 1 | -26,7% | -27,5% | +-------+-----------+------------+ | 2 | 0% | +53% | +-------+-----------+------------+ | 4 | +38,3% | +26,5% | +-------+-----------+------------+ | 8 | +10,9% | -4,5% | +-------+-----------+------------+ | 16 | +18,2% | +1,6% | +-------+-----------+------------+ | 64 | +22,3% | +8,75% | +-------+-----------+------------+ | 256 | +12,3% | +11,2% | +-------+-----------+------------+ | 1024 | +79,2% | +10,9% | +-------+-----------+------------+ Signed-off-by: Kévin PETIT <kevin.petit@arm.com> BUG=skia: R=djsollen@google.com, mtklein@google.com Author: kevin.petit@arm.com Review URL: https://codereview.chromium.org/181523002 git-svn-id: http://skia.googlecode.com/svn/trunk@14103 2bbb7eff-a529-9590-31e7-b0007b416f81
65 lines
2.5 KiB
Plaintext
65 lines
2.5 KiB
Plaintext
# Failures of any GM tests listed in this file will be ignored [1], as
|
|
# if they had been marked "ignore-failure": true in the per-builder
|
|
# expected-results.json files.
|
|
#
|
|
# The idea is that, if you modify a GM test in such a way that you know it will
|
|
# require rebaselines on multiple platforms, you should add that test's name to
|
|
# this list within the same CL that modifies the test.
|
|
# Then, AS SOON AS POSSIBLE, you should rebaseline the tests (they will show up
|
|
# as status "failure-ignored") and remove the line you added to this file.
|
|
# If there are any lingering failures that need to be suppressed (ignored), you
|
|
# MUST move those suppressions into the per-builder expected-results.json files.
|
|
#
|
|
# The test name must match exactly (no partial string matches).
|
|
# See http://skbug.com/2022
|
|
#
|
|
# Any lines starting with '#' are comments and will be ignored.
|
|
# Other than that, each line must either be whitespace or list just one test.
|
|
#
|
|
# For more information, see https://code.google.com/p/skia/issues/detail?id=1600
|
|
# ('GM expectations: create tool to suppress failures until new baselines are
|
|
# checked in')
|
|
#
|
|
#
|
|
# [1] Differences between renderModes (e.g., normal rendering path and
|
|
# "serialized") are NEVER ignored. See
|
|
# https://code.google.com/p/skia/issues/detail?id=1748 ('gm: add new result
|
|
# type, RenderModeMismatch')
|
|
|
|
# EXAMPLES: (remove the first '#' on each line)
|
|
#
|
|
## Added as part of MADE-UP BUG
|
|
## http://skbug.com/123456 : ignoring failures on gradtext GM test
|
|
## epoger will rebaseline by 25 Dec 2013
|
|
#gradtext
|
|
|
|
# This GM has probably always drawn incorrectly. We need to clarify the API
|
|
# http://skbug.com/2222 : Inverse filled paths get inverse stroked
|
|
inverse_paths
|
|
|
|
# bsalomon: Additional test cases were added to these GMs
|
|
# https://codereview.chromium.org/193263002/
|
|
# The rrect_clip_aa and rrect_effect tests were additionally modified by:
|
|
# https://codereview.chromium.org/194603004
|
|
rrect_effect
|
|
rrect_draw_bw
|
|
rrect_draw_aa
|
|
rrect_clip_bw
|
|
rrect_clip_aa
|
|
|
|
# bsalomon: This test changed imperceptibly on my windows laptop with an intel gpu
|
|
# https://codereview.chromium.org/204983003/
|
|
filltypespersp
|
|
|
|
# reed: https://codereview.chromium.org/190723004/
|
|
# This change removes an API that this GM was testing. If/when it lands and sticks,
|
|
# I will likely just delete the GM.
|
|
canvas-layer-state
|
|
|
|
# yunchao: https://codereview.chromium.org/204143004/
|
|
# This change add some cases to bitmapshader to avoid potential drawing error for kA8_Config
|
|
bitmapshaders
|
|
|
|
# Added by kevin.petit@arm.com for https://codereview.chromium.org/181523002/
|
|
gradients_view_perspective
|