Commit Graph

2609 Commits

Author SHA1 Message Date
skal
399b916d27 lossy decoding: correct alpha-rescaling for YUVA format
The luminance needs to be pre- and post- multiplied by
the alpha value in case of rescaling, for proper averaging.

Also:
- removed util/alpha_processing and moved it to dsp/
- removed WebPInitPremultiply() which was mostly useless
and merged it with the new function WebPInitAlphaProcessing()

Change-Id: If089cefd4ec53f6880a791c476fb1c7f7c5a8e60
2014-05-27 15:27:13 -07:00
James Zern
78c12ed8e6 Merge "Makefile.vc: add rudimentary avx2 support" 2014-05-27 11:13:40 -07:00
skal
dc5b122f23 try to remove the spurious warning for static analysis
Change-Id: Ib81f16c70a0bfad05021401c1cf6788c974b63bd
2014-05-26 18:31:00 +02:00
James Zern
ddfefd624c Makefile.vc: add rudimentary avx2 support
similar to makefile.unix:
> nmake /f Makefile.vc CFG=release-static HAVE_AVX2=1

from the msdn:
The /arch:AVX2 option and __AVX2__ macro were introduced in Visual
Studio 2013 Update 2, version 12.0.34567.1
(Update 2, version 12.0.30501.00 seems to work)

Change-Id: I649ee47c9fdc399fc71a8ac8464728608d9b6412
2014-05-23 20:52:02 -07:00
Pascal Massimino
a891164398 Merge "simplify VP8LInitBitReader()" 2014-05-22 22:36:41 -07:00
Pascal Massimino
fdbcd44dd3 simplify VP8LInitBitReader()
gcc was generating very complex code, one for each case of br->len_ values!

also, pretty-fy the mask constants

Change-Id: If62b1e8266f3fe5334517305113038d2ea8a6b42
2014-05-22 21:44:16 -07:00
James Zern
7c004287af makefile.unix: add rudimentary avx2 support
$ make -f makefile.unix HAVE_AVX2=1

will define -mavx2 for src/dsp/*_dsp.c

Change-Id: Id9651bda54da057cb051dc70f7dcd008a3f803f4
2014-05-22 18:38:40 -07:00
James Zern
515e35cfb1 Merge "add stub dsp/enc_avx2.c" 2014-05-22 18:28:38 -07:00
skal
a05dc1402c SSE2: yuv->rgb speed-up for point-sampling
- use statically initialized tables (if WEBP_YUV_USE_SSE2_TABLES is defined)
 - use SSE2 row conversion for yuv->ARGB / RGBA / ABGR / RGB / BGR
 - clean-up and harmonize the WebpUpsamplers[] usage.

Change-Id: Ic5f3659a995927bd7363defac99c1fc03a85a47d
2014-05-22 09:56:47 +02:00
James Zern
178e9a69ae add stub dsp/enc_avx2.c
VP8EncDspInitAVX2 is included in sse2 builds for now, later a configure
flag should be added to avoid the stub when avx2 is unavailable/disabled

Change-Id: I6127b687c273f46f41652aaf8e3b86ae3cfb8108
2014-05-22 00:31:46 -07:00
James Zern
1b99c09cdc Merge "configure: add a test for -mavx2" 2014-05-22 00:30:10 -07:00
James Zern
fe72807112 configure: add a test for -mavx2
sets AVX2_FLAGS; currently unused

Change-Id: Ie07ee6c2fa7c1f0748430010a9f207b1723b6def
2014-05-21 23:17:21 -07:00
James Zern
e46a247c87 cpu: fix check for __cpuidex availability
__cpuidex was added in VS2008 /SP1/

Change-Id: Ie49b00b0246bd6537c0ed583412f17d6fd135baa
2014-05-21 22:59:47 -07:00
skal
176fda2650 fix the bit-writer for lossless in 32bit mode
Sometimes, we can write 18bit or more at time, and it would
overflow the 32bit accumulator.

Also clarified the num-bits limitations (and exposed
VP8L_MAX_NUM_BIT_READ in bit_reader.h)

fixes http://code.google.com/p/webp/issues/detail?id=200

Seems a bit faster (use of local fields for bits_ / used_)

also: added the __QNX__ bswap while at it.

Change-Id: I876db93a931db15b083cf1d838c70105effa7167
2014-05-22 07:19:22 +02:00
James Zern
541784c710 dsp.h: add a check for AVX2 / define WEBP_USE_AVX2
Change-Id: I90cc870f0bb4426af701779c367587dc2ae79c8b
2014-05-21 20:46:28 -07:00
James Zern
bdb151ee80 dsp/cpu: add AVX2 detection
currently unused.
https://software.intel.com/en-us/articles/how-to-detect-new-instruction-support-in-the-4th-generation-intel-core-processor-family
http://www.intel.com/content/dam/www/public/us/en/documents/manuals/64-ia-32-architectures-optimization-manual.pdf

Change-Id: I314200f890c58b9a587b902b214f90deb95f0579
2014-05-20 22:48:54 -07:00
Pascal Massimino
ab9f2f8685 Merge "revamp the point-sampling functions by processing a full plane" 2014-05-20 15:21:31 -07:00
Pascal Massimino
a2f8b28905 revamp the point-sampling functions by processing a full plane
-nofancy is slower than fancy upsampler, because the latter has SSE2 optim.

Change-Id: Ibf22e5a8ea1de86a54248d4a4ecc63d514c01b88
2014-05-20 15:13:44 -07:00
Pascal Massimino
ef076026af use decoder's DSP functions for autofilter
-af is now faster (6-7%), since we're using the SSE2 variant
Output is binary the same as before.

Change-Id: If75694594c9501cd486b8f237a810ddcc145cadd
2014-05-20 14:55:05 -07:00
pascal massimino
2b5cb32612 Merge "dsp/cpu: add AVX detection" 2014-05-20 01:10:18 -07:00
James Zern
df08e67e06 dsp/cpu: add AVX detection
currently unused.
https://software.intel.com/en-us/articles/introduction-to-intel-advanced-vector-extensions

similar checks exist in ffmpeg, libyuv. the visual studio inline asm is
based off of libyuv.

Change-Id: I3e233de3492172434e482607a94b99c617f11aad
2014-05-20 00:25:12 -07:00
Pascal Massimino
e2f405c969 Merge "clean-up and slight speed-up in-loop filtering SSE2" 2014-05-20 00:08:40 -07:00
Pascal Massimino
f60957bfd2 clean-up and slight speed-up in-loop filtering SSE2
* remove some sign-bit flipping
* turn some macro into inline functions
* fix some 'const' in signatures
* clarify the int8/uint8 usage

Change-Id: Ib04459ac34cb280c57579c5d79a5efd2f8d5e99d
2014-05-19 23:23:47 -07:00
James Zern
9fc3ae469f .gitattributes: treat .ppm as binary
Change-Id: I4da7b846f6255078f0ce97fc7e8df9f29271f52a
2014-05-15 23:18:35 -07:00
James Zern
3da924b5b4 Merge "dsp/WEBP_USE_NEON: test for __aarch64__" 2014-05-14 20:16:18 -07:00
James Zern
c7164490da Android.mk: always include *_neon.c in the build
the inclusion of the files is harmless when NEON is not enabled and will
allow them to be built with NEON for APP_ABI=arm64-v8a which currently
does not use the '.neon' suffix

Change-Id: I39377876b1b68822c38f4e2396da93c56145fc0f
2014-05-14 00:11:46 -07:00
James Zern
a577b23a0a dsp/WEBP_USE_NEON: test for __aarch64__
__ARM_NEON__ is unset by current linux gcc/clang + android toolchains
for aarch64/arm64 builds.

Change-Id: Ib2ca172ea6fcf046e4ced19a431088674c99b7f6
2014-05-14 00:07:13 -07:00
skal
54bfffcabc move RemapBitReader() from idec.c to bit_reader code
mostly for coherency and later patch.

Change-Id: Ica8352d67845b6c5b3153435edfb4646c6f24341
2014-05-14 07:07:08 +02:00
James Zern
34168ecbe4 Merge "remove all unused layer code" 2014-05-08 22:51:13 -07:00
Pascal Massimino
f1e771735a remove all unused layer code
Change-Id: I220590162b24c70f404fe3087f19dd3e6cac3608
2014-05-08 22:37:38 -07:00
Vikas Arora
b0757db7c6 Code cleanup for VP8LGetHistoImageSymbols.
Fix comments and few nits.
Change-Id: I8fa25ed523f12c6a7bfe125f0e4d638466ba4304
2014-05-08 14:13:47 -07:00
skal
5fe628d35d make the token page size be variable instead of fixed 8192
also changed the token-page layout a little bit to remove
a not-needed field.

This reduces the number of malloc()/free() calls substantially
with minimal increase in memory consumption (~2%).
For the tail of large sources, the number of malloc calls goes
typically from ~10000 to ~100 (e.g.: bryce_big.jpg: 22711 -> 105)

Change-Id: Ib847f41e618ed8c303d26b76da982fbc48de45b9
2014-05-05 14:26:14 -07:00
skal
f948d08c81 memory debug: allow setting pre-defined malloc failure points
MALLOC_FAIL_AT flag can be used to set-up a pre-determined failure
point during malloc calls. The counter value is retrieved using
getenv().
Example usage: export MALLOC_FAIL_AT=37 && cwebp input.png
will make 'cwebp' report a memory allocation error the 37th time
malloc() or calloc() is called.

MALLOC_MEM_LIMIT can be used similarly to prevent allocating more
than a given amount of memory. This is usually less convenient to
use than MALLOC_FAIL_AT since one has to know in advance the typical
memory size allocated.

Both these flags are meant to be used for debugging only!

Also: added a 'total_mem_allocated' to record the overall memory allocated

Change-Id: I9d408095ee7d76acba0f3a31b1276fc36478720a
2014-05-05 14:01:33 -07:00
skal
ca3d746e39 use block-based allocation for backward refs storage, and free-lists
Non-photo source produce far less literal reference and their
buffer is usually much smaller than the picture size if its compresses
well. Hence, use a block-base allocation (and recycling) to avoid
pre-allocating a buffer with maximal size.

This can reduce memory consumption up to 50% for non-photographic
content. Encode speed is also a little better (1-2%)

Change-Id: Icbc229e1e5a08976348e600c8906beaa26954a11
2014-05-05 11:11:55 -07:00
James Zern
1ba61b09f9 enable NEON intrinsics in aarch64 builds
avoids functions that use vtbl? as in iOS builds these are marked
unavailable

Change-Id: I17aedc3c7dc8f1d5be0941205de0b22c3772ef1b
2014-05-03 12:37:42 -07:00
James Zern
b9d2bb67d6 dsp/neon.h: coalesce intrinsics-related defines
Change-Id: Ifadd41a5bbf7f99eeb6d75d2b67daa25e0544946
2014-05-03 11:34:07 -07:00
James Zern
b5c7525897 iosbuild: add support for iOSv7/aarch64
Change-Id: I3a51c77276e245cd871acb18d9d70d109aac000b
2014-05-03 11:14:37 -07:00
Vikas Arora
9383afd5c7 Reduce number of memory allocations while decoding lossless.
This change reduces the number of calls to WebPSafeMalloc from 200 to
100. The overall memory consumption is down 3% for Lenna image.

Change-Id: I1b351a1f61abf2634c035ef1ccb34050b7876bdd
2014-05-02 01:01:43 -07:00
James Zern
888e63edc9 Merge "dsp/lossless: prevent signed int overflow in left shift ops" 2014-05-02 00:29:54 -07:00
Pascal Massimino
8137f3edbd Merge "instrument memory allocation routines for debugging" 2014-05-02 00:23:48 -07:00
Pascal Massimino
2aa187360d instrument memory allocation routines for debugging
Some tracing code is activated by PRINT_MEM_INFO flag.
For debugging only! (not thread-safe, and slow).

Change-Id: I282c623c960f97d474a35b600981b761ef89ace9
2014-05-02 00:19:55 -07:00
skal
d3bcf72bf5 Don't allocate VP8LHashChain, but treat like automatic object
the unique instance of VP8LHashChain (1MB size corresponding to hash_to_first_index_)
is now wholy part of VP8LEncoder, instead of maintaining the pointer to VP8LHashChain
in the encoder.

Change-Id: Ib6fe52019fdd211fbbc78dc0ba731a4af0728677
2014-04-30 14:10:48 -07:00
James Zern
bd6b8619dd dsp/lossless: prevent signed int overflow in left shift ops
force unsigned when shifting by 24.

Change-Id: I453601f33fdf01c516ef66ad23399ae6cbe032b3
2014-04-30 00:10:49 -07:00
James Zern
b7f19b8311 Merge "dec/vp8l: prevent signed int overflow in left shift ops" 2014-04-29 15:56:54 -07:00
Pascal Massimino
29059d5178 Merge "remove some uint64_t casts and use." 2014-04-29 14:15:40 -07:00
James Zern
e69a1df4b7 dec/vp8l: prevent signed int overflow in left shift ops
force unsigned when shifting by 24.

Change-Id: I6f9ca5fa2109e59b1d46a909136384fc6dc8ca0b
2014-04-29 14:12:38 -07:00
Pascal Massimino
cf5eb8ad19 remove some uint64_t casts and use.
We use automatic int->uint64_t promotion where applicable.

(uint64_t should be kept only for overflow checking and memory alloc).

Change-Id: I1f41b0f73e2e6380e7d65cc15c1f730696862125
2014-04-29 09:08:25 -07:00
Djordje Pesut
38e2db3e16 MIPS: MIPS32r1: Added optimization for HistogramAdd.
Change-Id: I39622a9c340c4090f64dd10e515c4ef2aa21d10a
2014-04-29 08:36:51 -07:00
James Zern
e0609ade15 dwebp: fix exit code on webp load failure
on ExUtilLoadWebP() failure no allocated memory will be returned, so
it's safe to exit immediately. additionally, any webp specific problems
will already have been reported as part of the call to
WebPGetFeatures().

broken since:
4a0e739 dwebp: move webp decoding to example_util

Change-Id: Ibc632015a1f52bae7f96d063252624123fa7c2da
2014-04-28 17:22:29 -07:00
James Zern
bbd358a8e7 Merge "example_util.h: avoid forward declaring enums" 2014-04-28 15:10:12 -07:00