| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
| |
QImage::convertToFormat was ignoring its conversion flag argument, only
performing dithering when converting to indexed formats.
This patch updates the documentation and implements ordered dithering
for other conversions.
Change-Id: I807353d61669694185b7e595ef262d80d9fbb3f1
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>
|
|
|
|
|
|
|
|
| |
Removes the now unused QPixelLayout parameter, simplifies the colorTable
passing and prepares for adding dithering.
Change-Id: Iaf7698b248b857804d8921bf118e7cfabbabff87
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>
|
|
|
|
|
|
|
|
|
| |
Remove #ifdef sections for Q_OS_WINCE and wince .pro file clauses in library,
and tests.
Task-number: QTBUG-51673
Change-Id: I55f61845c3b54027c467a5c59c122e7d16955358
Reviewed-by: Lars Knoll <lars.knoll@theqtcompany.com>
|
|\
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Conflicts:
configure
mkspecs/common/wince/qplatformdefs.h
src/plugins/platforms/directfb/qdirectfbbackingstore.cpp
src/plugins/platforms/xcb/qxcbbackingstore.cpp
Change-Id: Ied4d31264a9afca9514b51a7eb1494c28712793c
|
| |
| |
| |
| |
| |
| |
| |
| | |
Reuses the optimized routines from qimage to make painting RGB888
images faster on SSSE3 and NEON.
Change-Id: I99116b318322ba4cb0ddc2cb90bcf17a0350ef99
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>
|
| |
| |
| |
| |
| |
| |
| |
| | |
Calling qMin often prevents effective vectorization, and it is only
necessary when converting from formats with mixed color-channel widths.
Change-Id: I2a0f3f3fb528d45be1fd025758f9d915ee1736c0
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>
|
| |
| |
| |
| |
| |
| |
| |
| | |
These four methods do not need to be separate, the compiler can figure
generate the optimal version from the template arguments.
Change-Id: I45b30a9c2f2ce4da46c47f2e6e1fbd7561213c4a
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>
|
|/
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
sizeof:
Operator : 144
QTransform : 88
QGlyphLayout : 48
QTextFrame::Iterator: 32
QRectF : 32
QCss::Value : 24
(Linux/x86_64)
Change-Id: I4965bf0959683c0bc8909428fca86d66851af92c
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>
Reviewed-by: Olivier Goffart (Woboq GmbH) <ogoffart@woboq.com>
|
|
|
|
|
|
|
|
| |
Using shuffle and align storing our quint24 format can be done much
faster. This in particular improves conversions to RGB888.
Change-Id: I179748706a33a43fd6f60f5c40287317418c8867
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>
|
|\
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This also reverts commit 018e670a26ff5a61b949100ae080f5e654e7bee8.
The change was introduced in 5.6. After the refactoring, 14960f52,
in 5.7 branch and a merge, it is not needed any more.
Conflicts:
.qmake.conf
src/corelib/io/qstandardpaths_mac.mm
src/corelib/tools/qsharedpointer_impl.h
tests/auto/widgets/itemviews/qlistview/tst_qlistview.cpp
Change-Id: If4fdff0ebf2b9b5df9f9db93ea0022d5ee3da2a4
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Convert rgba64 color table to a new rgb32 color table only once. Use
this cache when required.
This patch can 2x speed up gradient painting using 32bit color format.
Task-number: QTBUG-50930
Change-Id: I9212e01e397c2e0127cdf3070cc49880a2d8df88
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@theqtcompany.com>
|
|\|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Conflicts:
src/corelib/io/qfilesystemwatcher_win.cpp
src/corelib/plugin/plugin.pri
src/plugins/platforms/cocoa/qcocoaaccessibility.mm
tests/auto/corelib/tools/qlocale/tst_qlocale.cpp
Change-Id: Id6824631252609a75eff8b68792e4d10095c8fc1
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
To calculate the real count we need to use the actual fixed point
increment and can not use the floating point value increment wass based
on since it might round differently.
Includes auto-test by Gabriel de Dietrich.
Task-number: QTBUG-50153
Change-Id: Ia973088f361c90370fa20bac14a4b8f373b5d234
Reviewed-by: Gabriel de Dietrich <gabriel.dedietrich@theqtcompany.com>
Reviewed-by: Olivier Goffart (Woboq GmbH) <ogoffart@woboq.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
I wrote a script to help find the files, but I reviewed the
contributions manually to be sure I wasn't claiming copyright for search
& replace, adding Q_DECL_NOTHROW or adding "We mean it" headers.
Change-Id: I7a9e11d7b64a4cc78e24ffff142b506368fc8842
Reviewed-by: Lars Knoll <lars.knoll@theqtcompany.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
From Qt 5.7 -> LGPL v2.1 isn't an option anymore, see
http://blog.qt.io/blog/2016/01/13/new-agreement-with-the-kde-free-qt-foundation/
Updated license headers to use new LGPL header instead of LGPL21 one
(in those files which will be under LGPL v3)
Change-Id: I046ec3e47b1876cd7b4b0353a576b352e3a946d9
Reviewed-by: Lars Knoll <lars.knoll@theqtcompany.com>
|
|\|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Conflicts:
src/gui/painting/painting.pri
src/plugins/platforms/xcb/qxcbconnection.cpp
tests/auto/corelib/thread/qthreadstorage/qthreadstorage.pro
tests/auto/corelib/tools/qlocale/test/test.pro
tests/auto/gui/kernel/qwindow/tst_qwindow.cpp
tools/configure/environment.cpp
Change-Id: I9c40f458b89b2c206de2d2c24e90b5f679c93495
|
| |
| |
| |
| |
| |
| |
| |
| | |
Use the macro we have that does exactly what we want without the boiler
plate code.
Change-Id: I25c67a71f83f91f50128c8a54033ee53ad28e88a
Reviewed-by: Erik Verbruggen <erik.verbruggen@theqtcompany.com>
|
| |
| |
| |
| |
| |
| | |
Change-Id: I8636993571b04d8cda74708777ba3ef2bac09fb5
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@theqtcompany.com>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@theqtcompany.com>
|
|/
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Adds NEON version of interpolate_4_pixels used by smooth upscaling, and
bilinear sampling.
The SSE2 version is reordered to match the NEON version so they have
the same order of operations and a faster version that loads directly
into vector registers.
Testing is extended so we have a test of smoothness that can catch more
possible mistakes.
Change-Id: I0de4aecf5cb79468e7c8f19f421aa24b2955547c
Reviewed-by: Erik Verbruggen <erik.verbruggen@theqtcompany.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
QImage can be used without QGuiApplication, but wihout QGuiApplication
the accelerated methods would not be used.
This patch changes the initialization of the QtGui function tables to
happen in static object constructors on QtGui load, instead of in
the QGuiApplication init.
Change-Id: Ib58982fbc50fd79122352d021d93a3c99aca5043
Reviewed-by: Erik Verbruggen <erik.verbruggen@theqtcompany.com>
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>
|
|
|
|
|
|
|
|
|
| |
Fix the ARGB32 NEON-optimized drawhelpers so they also build on AArch64.
The RGB16 NEON-drawhelpers are not converted as they use more assembly.
Change-Id: I8b75fadf5bad74360a2ab6aec9a6bf50df80c1b9
Reviewed-by: Erik Verbruggen <erik.verbruggen@theqtcompany.com>
|
|
|
|
|
|
|
|
|
|
|
| |
This information is already registered by the QMessageLogger ctor.
Where, by dropping the << Q_FUNC_INFO in ostream-style qDebug(), only
a string literal remained, converted to printf-style qDebug() on the
go.
Change-Id: I3f261c98fd7bcfa1fead381a75a82713bb75e6f3
Reviewed-by: Olivier Goffart (Woboq GmbH) <ogoffart@woboq.com>
|
|
|
|
|
|
|
|
|
|
|
|
| |
The __ARM_NEON is the standard define for NEON instructions support
__ARM_NEON__ is only legacy, and specifically not defined in
AArch64 builds, which causes us not to detect NEON support there.
The NEON assembler files doesn't build with AArch64, so the NEON
drawhelper methods must be excluded for now.
Change-Id: Ie32f855bde94ee7efd8a8ddb7766c931778e729b
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
|
|
|
|
|
|
| |
error C2397: conversion from 'uint' to 'uchar' requires a narrowing conversion
Change-Id: Ic1b9e0a78b2faa5a0ae4e6de33ab2ee4b654af5f
Reviewed-by: Olivier Goffart (Woboq GmbH) <ogoffart@woboq.com>
|
|
|
|
|
|
|
|
| |
_mm_load_si128 requires 16 byte alignment. This crashes on 32 bit Windows
builds.
Change-Id: Ib6c30eba726747bbab56467eada820521981a80c
Reviewed-by: Olivier Goffart (Woboq GmbH) <ogoffart@woboq.com>
|
|
|
|
|
|
|
|
|
|
|
| |
qConvertARGB32PMToARGB64PM_sse2
On 32 bit platforms the pointers may end up being 4 byte aligned.
Happens with MSVC on 32 bit Windows. _mm_store_si128 is documented to
require 16 byte alignment.
Change-Id: I80737fedf9e7f436a51a83924117cc0bc63017cc
Reviewed-by: Jędrzej Nowacki <jedrzej.nowacki@theqtcompany.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Instead of trying to detect one bit and set another, let's just use the
bits from the x86 CPUID instruction on x86. This makes use of the full
64-bit space now.
Since MSVC doesn't like enums bigger than 32-bit, we have to store the
bit number instead of the actual bit value in the constant. For that
reason, I also renamed the constants, to catch anyone who was using them
directly, instead of through qCpuHasFeature.
Change-Id: Ib306f8f647014b399b87ffff13f1d587692d827a
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@theqtcompany.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Just like from RGB32 to RGB30 we must also repremultiply when converting
from RGB64 because the alpha channel loses more precision than the other
color channels.
Since this is not approximated accurately in the simple blending
functions and the functions are no longer needed now the main render
engine supports higher accuracy, the simple blending routines for RGB30
have been removed.
Change-Id: I2b7b8eb015e330a487848fc4370ad3a1e966be91
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>
|
|\
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Conflicts:
src/corelib/global/qglobal.cpp
src/corelib/global/qglobal.h
src/corelib/global/qsysinfo.h
src/corelib/global/qsystemdetection.h
src/corelib/kernel/qobjectdefs.h
src/plugins/plugins.pro
tests/auto/widgets/itemviews/qlistview/qlistview.pro
Change-Id: Ib55aa79d707c4c1453fb9d697f6cf92211ed665c
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
When a premultiplied alpha changes value because it is rounded to fewer
bits the premultiplied colors may need to be recalculated with the new
value. Otherwise the color will both be wrong and potentially invalid.
Change-Id: I9ec74a22aac73cd7ffab04e180cf2bf35bb4c315
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>
|
|\|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Conflicts:
src/corelib/global/qnamespace.qdoc
src/corelib/io/qwindowspipereader.cpp
src/corelib/io/qwindowspipereader_p.h
src/corelib/statemachine/qstatemachine.cpp
src/corelib/statemachine/qstatemachine_p.h
src/plugins/platforms/xcb/qxcbconnection.h
tests/auto/network/access/qnetworkreply/tst_qnetworkreply.cpp
tests/auto/tools/qmake/tst_qmake.cpp
tests/manual/touch/main.cpp
Change-Id: I917d694890e79ee3da7d65134b5b085e23e0dd62
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Moves some of them to the .rodata section, the rest at least to
.data.rel.ro[.local].
Change-Id: I85676ddf22b0c0097f3f0dce4c3dc018dc29d045
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
|
|\|
| |
| |
| |
| |
| |
| |
| | |
Conflicts:
src/tools/qdoc/tree.cpp
tests/auto/gui/painting/qcolor/tst_qcolor.cpp
Change-Id: Iaa78f601a63191fa643aabf853520f913f2f0fdc
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Could causedSSE4 instructions to be used on non SSE4 machines
in cases when qUnpremultiplywas not inlined.
This reverts commit 964ccc58534aac436529007000d1c38d76c88834.
Change-Id: Ic676ade8f75129e8d37c4d96cbfb2bdb5b794919
Task-number: QTBUG-45741
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
|\|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Conflicts:
src/corelib/statemachine/qstatemachine.cpp
src/corelib/statemachine/qstatemachine_p.h
src/gui/painting/qdrawhelper.cpp
src/plugins/platforms/xcb/qxcbnativeinterface.cpp
src/plugins/platforms/xcb/qxcbwindow.cpp
src/plugins/platforms/xcb/qxcbwindow.h
src/testlib/qtestblacklist.cpp
src/tools/qdoc/node.cpp
src/tools/qdoc/node.h
tests/auto/gui/painting/qcolor/tst_qcolor.cpp
Change-Id: I6c78b7b162001712d5774293f501b06b4ff32684
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Merges the SSE4 specific unpremultiply with the normal version, and
adds a SSE2 fallback. There was no reason to split the two since
compile time options will ensure the right version is inlined.
Also adds short-cut for 0 and 255 values.
Change-Id: Ie5aa262f6964219fd3062d4a498f697cf79a4595
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
| |
| |
| |
| |
| |
| |
| |
| | |
Reading directly from an array instead of calling a function pointer
is much faster.
Change-Id: I833b33448bad064d6f38d2f9ff44138d90206822
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Cleaning up smoothscale code. Upscaling is improved using existing
optimized interpolation methods, and downscale is given SSE4.1
optimized versions.
Change-Id: I7cdc256c26850948aef7dae26fda1622be6b8179
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Also changes the way the SSE2 composition modes are inserted into the table,
so it is handled like all the other tables and doesn't require duplicating most
of the table.
Change-Id: I8de383caece0367bc7466d7a1b145aa820c3bd6a
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Adds the last two missing source types to rgb64 rendering.
Conical and radial gradients. At the same time linear
gradients are moved to a template form to increase code
sharing.
Change-Id: I30fdd0837b0da03e3447683856ebbe4d7f48df6c
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
A previous commit added 16-bit per color-channel precision to
solids and linear gradients. This patch extends that to include
texture data.
To support that pixellayouts now have a method to convert to
RGBA64.
Change-Id: I661ae91bd7038085787003608a0af4add057e478
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>
|
| |
| |
| |
| |
| | |
Change-Id: I3bacecbabdf2d7b2de1acd86ab9383e69924a390
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Some tests were failing because the color was not read correctly from
a QImage. To make it possibly to read more accurate colors a pixel
accessor returing QColor has been added.
Some tests also had the wrong order of arguments, confusing dest and src
formats. This has been corrected, so they test what they claim to test.
A test for RGB30 linear gradients is also added.
Change-Id: Ic623ae1b8e0bf7383056b641c6e8230a1d7dd0dd
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>
|
|/
|
|
|
|
|
|
|
|
|
|
|
| |
This patch updates the internal color precisions of solids and
gradients to 16bit per color. This makes it possible to render
at higher precision on non-premultiplied ARGB32, the RGB30
formats and any other hi-color formats if more are added.
[ChangeLog][QtGui][Painting] Internal precision of solids and gradients
is now up to 16bit per color.
Change-Id: Ieae5468bd6de1f56adfa4cb9fa966faf2ed824fd
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>
|
|
|
|
|
|
|
|
|
|
|
| |
Most of the CPU time was spending in fetching pixel values instead of
calculating the bilinear sample. The access to vectors unions turns
out to be very slow, and should be avoided. This patch removes the
uses of vector,int array union in the bilinear sampling for SSE2.
Change-Id: Ie765a80963c4899db59a583ea9a59d15e05f2b13
Reviewed-by: Kai Koehne <kai.koehne@theqtcompany.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
|
|
|
|
|
|
| |
Found with GCC's -Wcast-qual.
Change-Id: Ia0aac2f09e9245339951ffff13c946844bc31eb8
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@theqtcompany.com>
|
|
|
|
|
|
|
|
|
|
|
|
| |
Following up on using GCC's autovectorizing for faster SSE4.1
premultiply, this patch adds specialized autovectorized versions
of premultiply for AVX2, giving another almost doubling in speed.
To make the speed up for AVX2 and also SSE4_1 available to non-GCC
compilers, the target-specific methods have been moved to separate
files.
Change-Id: I97ce05be67f4adeeb9a096eef80fd5fb662099f3
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>
|
|
|
|
|
|
|
|
|
|
|
| |
Removes the non-templated pixellayout methods that were used for
non-c++11 builds, and instead initialize with the templated ones
verbatim.
These versions are faster, and using only those removes duplicate code.
Change-Id: I91a7d9122d2601ad282287a2f5ed6b88877d52fb
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>
|
|
|
|
|
|
|
|
|
| |
The patch removes several uses of the private Q_PI constant and removes
the qmath_p.h from the includes. A few places are optimized to multiply
with reciprocals instead of dividing by Q_PI.
Change-Id: I097af6a929e0609d6935563064e81c856005f4bc
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Including math.h can pollute the default namespace, and break
some compilers if cmath versions of the method are declared as using.
Switching to C++ math functions also greatly simplifies handling of
float qreal as C++ automatically chooses the right method.
[ChangeLog][QtCore][QtMath] qmath.h no longer includes math.h, so any
sources depending on that indirect inclusion may fail to build.
Change-Id: I4d0e331dafba354ec05dc5052e61ef4ff8d387fe
Reviewed-by: Rafael Roquetto <rafael.roquetto@kdab.com>
|