lib: Properly mark threads above 32 as available
arm32: Set .arch armv7-a just like we already set .fpu neon
meson: Try building with -Wshorten-64-to-32
cdef_apply_tmpl: Restore the y variables to int
obu: Initialize off_before_idx[0], fix MSVC warning about uninitialized use
Add casts to silence warnings about intended type conversions/shortenings
x86: Add w_mask_444 AVX2 asm
x86: Add w_mask_422 AVX2 asm
Use 64-bit versions of fseek and ftell
arm64: cdef: NEON optimized cdef filter function
x86: Fix 32-bit looprestoration SSSE3 asm on MSVC
dav1d_fuzzer_mem: test random pthread_* errors
clean up threads on dav1d_open failures
checkasm: Fix MC strides
looprestoration: add SSSE3 implementation
looprestoration: correct comment in AVX2 sgr_box5_v
looprestoration: fix macro undef in C code
Simplify dav1d_thread_picture_alloc()
Windows: Improve pthread wrapper
Add SSSE3 implementation for ipred_smooth, ipred_smooth_v and ipred_smooth_h
tools: Store return values from fread() in a size_t
Cast the output from strtoul to unsigned int
Use size_t as return value type for strlen() consistently
Avoid warnings about variables being truncated by changing the type of variables
decode: Use abs() instead of labs()
meson: Check for io.h and include it in dav1d.c if available
meson: Try adding -Wno-missing-field-initializers when building
arm: looprestoration: Fix register shifting at the end of a vertical run
CI: disable logging on test jobs
build: fix defining CONFIG_LOG when logging is disabled
cdef: Add a CDEF_ prefix to the HAVE_LEFT etc enum values
CI: Added armv7 build jobs.
checkasm: cdef: Don't use uninitialized buffer contents for the test
arm: looprestoration: Fix a comment to use the right instruction names
mc: Ensure high bitdepth intermediates fits in int16_t
arm64: looprestoration: Optimize loop termination checks in copy_narrow_neon
Make the test section easier to use
arm: looprestoration: NEON optimized wiener filter
arm: fix movrel macro for thumb
lib: disable logging in dav1d_parse_sequence_header()
arm64: looprestoration: Simplify the horizontal filtering of one pixel at a time
arm64: looprestoration: Simplify the setup of wiener_filter_v_neon
arm64: looprestoration: Fix the loop condition in copy_narrow_neon
arm64: looprestoration: Fix comment typos
arm64: looprestoration: Avoid unnecessary alignment of the mid buffer
replace fprintf uses with dav1d_log()
add a logging callback mechanism
x86inc: Support N_PEXT bit on Mach-O
Internalize the EC_PROB_SHIFT precision reduction
Add msac_decode_bool_equi() function
Fix broken link to wikis/task-list
arm: mc: Implement 8tap and bilin functions
arm: Fix the movrel macro for Apple with PIC
Don't filter top/left intra edge if intra_edge_filter=0
CI: Add CI jobs for armv7-w64-mingw32 and aarch64-w64-mingw32
arm64: mc: Optimize mc_8tap_regular_w4_hv_8bpc for A53
arm64: mc: Simplify the 8tap_2w_hv code slightly
arm64: mc: Optimize the mul_mla_8_* macros for Cortex A53
arm64: mc: Improve a comment
arm64: mc: Remove unused/unnecessary macro args
arm64: mc: Use ubfx instead of ubfm, for consistency with arm
arm: Create proper .rdata sections for COFF
data: factorize setting Dav1dDataProps to default values
Add SSSE3 implementation for the 4x16 and 16x4 blocks in itx
Add SGR optimizations
build: Change required nasm version to 2.13.02
allocate Tile Group cache dynamically
Add minor x86 bilin mc optimizations
Shrink dav1d_dr_intra_derivative[]
obu: fix separate_uv_delta_q for RGB
obu: set subsampling to 1 for monochrome
Set subsampling to 0 for RGB.
ref: properly free the data buffer in dav1d_ref_create on failure
pass the sequence header corresponding to the picture being allocated
Add SSSE3 put_bilin
picture: properly handle Dav1dUserData references in Dav1dPicture
Only set Dav1dRef's data field in dav1d_ref_create
Add a user data pointer to Dav1dDataProps
Add SSSE3 implementations for dav1d_ipred_top, dav1d_ipred_left and dav1d_ipred_128
Add SSSE3 implementation for the 8x8 blocks in itx
Add SSSE3 implementation for dav1d_ipred_v and dav1d_ipred_dc
Add tail call optimizations in SSSE3 itx
Add SSSE3 implementation for the 4x8 and 8x4 blocks in itx
Duplicate LR line if the bottom edge only has one line available
build: Silence some useless MSVC warnings
Use 16-byte stack alignment on 32-bit x86
Don't apply QM on lossless blocks
Perform stack realignment in every API entry point
arm: detect NEON on FreeBSD armv6
arm: assume NEON if built with it
intrabc: adjust right border to multiple of the block size
Only perform stack allocation on WIN64 in ipred_filter_avx2
memory sanitizer: check all blocks for fully initialized pixel data
Fix stack corruption in 32-bit emu_edge_ssse3
CI: update debian32 image to treat text relocations as error
add SSSE3 emu_edge