- Use the _M_ARM and _M_ARM64 macros provided by Visual Studio for
compile-time detection of Arm builds, since __arm__ and __aarch64__
are only present in GNU-compatible compilers.
- Neon/intrinsics: Use the _CountLeadingZeros() and
_CountLeadingZeros64() intrinsics provided by Visual Studio, since
__builtin_clz() and __builtin_clzl() are only present in
GNU-compatible compilers.
- Neon/intrinsics: Since Visual Studio does not support static vector
initialization, replace static initialization of Neon vectors with the
appropriate intrinsics. Compared to the static initialization
approach, this produces identical assembly code with both GCC and
Clang.
- Neon/intrinsics: Since Visual Studio does not support inline assembly
code, provide alternative code paths for Visual Studio whenever inline
assembly is used.
- Build: Set FLOATTEST appropriately for AArch64 Visual Studio builds
(Visual Studio does not emit fused multiply-add [FMA] instructions by
default for such builds.)
- Neon/intrinsics: Move temporary buffer allocation outside of nested
loops. Since Visual Studio configures Arm builds with a relatively
small amount of stack memory, attempting to allocate those buffers
within the inner loops caused a stack overflow.
Closes#461Closes#475
Otherwise, because the file begins with an ASCII header, Git will
erroneously treat is as an ASCII file, and if Git for Windows is
configured with default options (specifically, "Checkout windows-style,
commit Unix-style line endings"), it will add carriage return characters
to all of the "linefeed" characters in the PPM file, thus corrupting it
and causing libjpeg-turbo's regression tests to fail.
There doesn't seem to be any performance or compatibility downside to
this, and it has the advantages of simplicity and consistency between
the PR and official builds.
- Rename IOS_ARMV8_BUILD to ARMV8_BUILD.
- Rename install_ios() to install_subbuild() in makemacpkg.
- Wordsmith the build instructions accordingly.
- Use xcode12.2 image in Travis CI.
This error occurs at the call to (*cinfo->cconvert->color_convert)() in
sep_upsample() whenever cinfo->upsample->need_context_rows == TRUE
(i.e. whenever h2v2 or h1v2 fancy upsampling is used.) The error is
innocuous, since (*cinfo->cconvert->color_convert)() points to a dummy
function (noop_convert()) in that case.
Fixes#470
The "32bit" vs. "64bit" floating point test results actually have
nothing to do with the FPU. That was a fallacious assumption based on
the observation that, with multiple CPU types, 32-bit and 64-bit builds
produce different floating point test results. It seems that this is,
in fact, due to differing compiler behavior-- more specifically, whether
fused multiply-add (FMA) instructions are used to combine multiple
floating point operations into a single instruction ("floating point
expression contraction".) GCC does this by default if the target
supports FMA instructions, which PowerPC and AArch64 targets both do.
Fixes#468
This allows the Neon intrinsics code to be built successfully (albeit
likely with reduced run-time performance) with Xcode 5.0-6.2
(iOS/AArch64) and Android NDK < r19 (AArch32). Note that Xcode 5.0-6.2
will not build the Armv8 GAS code without gas-preprocessor.pl, and no
version of Xcode will build the Armv7 GAS code without
gas-preprocessor.pl, so we always use the full Neon intrinsics
implementation by default with macOS and iOS builds.
Auto-detecting the completeness of the compiler's set of Neon intrinsics
also allows us to more intelligently set the default value of
NEON_INTRINSICS, based on the values of HAVE_VLD1*. This is a
reasonable, albeit imperfect, proxy for whether a compiler has a full
and optimal set of Neon intrinsics. Specific notes:
- 64-bit RGB-to-YCbCr color conversion
does not use any of the intrinsics in question, regresses with GCC
- 64-bit accurate integer forward DCT
uses vld1_s16_x3(), regresses with GCC
- 64-bit Huffman encoding
uses vld1q_u8_x4(), regresses with GCC
- 64-bit YCbCr-to-RGB color conversion
does not use any of the intrinsics in question, regresses with GCC
- 64-bit accurate integer inverse DCT
uses vld1_s16_x3(), regresses with GCC
- 64-bit 4x4 inverse DCT
uses vld1_s16_x3(). I did not test this algorithm in isolation, so
it may in fact regress with GCC, but the regression may be hidden by
the speedup from the new SIMD-accelerated upsampling algorithms.
- 32-bit RGB-to-YCbCr color conversion:
uses vld1_u16_x2(), regresses with GCC
- 32-bit accurate integer forward DCT
uses vld1_s16_x3(), regression irrelevant because there was no
previous implementation
- 32-bit accurate integer inverse DCT
uses vld1_s16_x3(), regresses with GCC
- 32-bit fast integer inverse DCT
does not use any of the intrinsics in question, regresses with GCC
- 32-bit 4x4 inverse DCT
uses vld1_s16_x3(). I did not test this algorithm in isolation, so
it may in fact regress with GCC, but the regression may be hidden by
the speedup from the new SIMD-accelerated upsampling algorithms.
Presumably when GCC includes a full and optimal set of Neon intrinsics,
the HAVE_VLD1* tests will pass, and the full Neon intrinsics
implementation will be enabled automatically.
- Remove gas-preprocessor.pl. None of the compilers that can build the
new intrinsics implementation require gas-preprocessor.pl (tested
with Xcode and with Clang 3.9+ for Linux.)
- Document that Xcode 6.3.x or later is now required for iOS builds
(older versions of Xcode do not have a full set of Neon intrinsics.)
- Add a change log entry.
- Do not enable the ASM CMake language unless NEON_INTRINSICS is false.
- Add a Clang/Arm64 test to .travis.yml in order to test the new
intrinsics implementation.
Closes#455
There was no previous GAS implementation.
NOTE: This doesn't produce much of a speedup when using -O3, because -O3
already enables Neon autovectorization, which works well for the scalar
C implementation of plain upsampling. However, the Neon SIMD
implementation will benefit other optimization levels.
There was no previous GAS implementation.
This commit also reverts 40557b2301 and
7723d7f7d0.
7723d7f7d0 was only necessary because
there was no Neon implementation of merged upsampling/color conversion,
and 40557b2301 was only necessary because
of 7723d7f7d0.
The previous AArch64 GAS implementation has been removed, since the
intrinsics implementation provides the same or better performance.
There was no previous AArch32 GAS implementation.
The previous AArch32 and AArch64 GAS implementations are retained by
default when using GCC, in order to avoid a performance regression. The
intrinsics implementation can be forced on or off using the new
NEON_INTRINSICS CMake variable.
The previous AArch32 GAS implementation is retained by default when
using GCC, in order to avoid a performance regression. The intrinsics
implementation can be forced on or off using the new NEON_INTRINSICS
CMake variable. The previous AArch64 GAS implementation has been
removed, since the intrinsics implementation provides the same or better
performance.
The previous AArch32 GAS implementation of h2v1 fancy upsampling has
been removed, since the intrinsics implementation provides the same or
better performance. There was no previous GAS implementation of h2v2
fancy upsampling, and there was no previous AArch64 GAS implementation
of h2v1 fancy upsampling.
The previous AArch64 GAS implementation is retained by default when
using GCC, in order to avoid a performance regression. The intrinsics
implementation can be forced on or off using the new NEON_INTRINSICS
CMake variable. The previous AArch32 GAS implementation has been
removed, since the intrinsics implementation provides the same or better
performance.
The previous AArch64 GAS implementation is retained by default when
using GCC, in order to avoid a performance regression. The intrinsics
implementation can be forced on or off using the new NEON_INTRINSICS
CMake variable. The previous AArch32 GAS implementation has been
removed, since the intrinsics implementation provides the same or better
performance.
The previous AArch64 GAS implementation is retained by default when
using GCC, in order to avoid a performance regression. The intrinsics
implementation can be forced on or off using the new NEON_INTRINSICS
CMake variable. The previous AArch32 GAS implementation has been
removed, since the intrinsics implementation provides the same or better
performance.
The previous AArch64 GAS implementation is retained by default when
using GCC, in order to avoid a performance regression. The intrinsics
implementation can be forced on or off using the new NEON_INTRINSICS
CMake variable. There was no previous AArch32 GAS implementation.
The previous AArch64 GAS implementation has been removed, since the
intrinsics implementation provides the same or better performance.
There was no previous AArch32 GAS implementation.
The previous AArch32 and AArch64 GAS implementations are retained by
default when using GCC, in order to avoid a performance regression. The
intrinsics implementation can be forced on or off using a new
NEON_INTRINSICS CMake variable.
Regression caused by
a46c111d9f
Because of 7723d7f7d0, which was
introduced in libjpeg-turbo 1.5.1 in response to #81, merged upsampling/
color conversion is disabled on platforms that have SIMD-accelerated
YCbCr -> RGB color conversion but not SIMD-accelerated merged
upsampling/color conversion. This was intended to improve performance
with the Neon SIMD extensions, since those are the only SIMD extensions
for which those circumstances apply. Under normal circumstances, the
separate "plain" (non-fancy) upsampling and color conversion routines
will produce bitwise-identical output to the merged upsampling/color
conversion routines, but that is not the case when skipping scanlines
starting at an odd-numbered scanline. The modified test introduced in
a46c111d9f does precisely that in order to
validate the fixes introduced in
9120a24743 and
a46c111d9f.
Because of 7723d7f7d0, the segfault fixed
in 9120a24743 and
a46c111d9f didn't affect the Neon SIMD
extensions, so this commit effectively reverts the test modifications in
a46c111d9f when using those SIMD
extensions. We can get rid of this hack, as well as
7723d7f7d0, once a Neon implementation of
merged upsampling/color conversion is available.
- Refer to the "slow" [I]DCT algorithms as "accurate" instead, since
they are not slow under libjpeg-turbo.
- Adjust documentation claims to reflect the fact that the "slow" and
"fast" algorithms produce about the same performance on AVX2-equipped
CPUs (because of the dual-lane nature of AVX2, it was not possible to
accelerate the "fast" algorithm beyond what was achievable with SSE2.)
Also adjust the claims to reflect the fact that the "fast" algorithm
tends to be ~5-15% faster than the "slow" algorithm on
non-AVX2-equipped CPUs, regardless of the use of the libjpeg-turbo
SIMD extensions.
- Indicate the legacy status of the "fast" and float algorithms in the
documentation and cjpeg/djpeg usage info.
- Remove obsolete paragraph in the djpeg man page that suggested that
the float algorithm could be faster than the "fast" algorithm on some
CPUs.
It is our convention to use the term "region" when referring to crop
specs, since this is more consistent with the terminology used by the
rest of the image processing community.
