fastaudio_decode() computes subframes = pkt->size / (40 * channels); frame->nb_samples = subframes * 256; both as 32-bit signed multiplications.
Vertically-subsampled inputs (nv12, nv21, yuv420p) share a chroma row across two output rows; compute the chroma -> RGB offsets once and apply to both luma rows.
pack_rgb16_2l uses v26-v29 as scratch (luma temps, dead by then) instead of v20-v23, so v20-v25 chroma survives the pack step.
An IFUNC resolver in a DSO that is being loaded by dlopen is allowed to read its own TLS storage during the resolver call.
The -unscaled parameter has been removed in favour of "-scaler none". Some legacy scalers cannot be selected with these options (i.e.: SWS_X and SWS_FAST_BILINEAR).
This patch adds an RVV-optimized implementation of memcmp for RISC-V and enables it for both multiarch (IFUNC) and non-multiarch builds. The implementation integrates Hau Hsu's 2023 RVV work under a unified ifunc-based framework.
This lets us test specific scaling operations, for example: $ .
Reduces sizeof(AACPCEInfo) from 296 to 120 bytes.
This allows a libspice-server consumer call spice_server_set_port(.
This adds a new protocol shared:URI which is distinct from the existing `cache:` in that it is explicity designed to be thread-safe and cross-process, enabling multiple ffmpeg processes (or multiple ffmpeg decoders within the same process) to share a single cache file, for e.g.
This will effectively disable the cache but allows the cache layer to verify cached files against the original input file.
The checkasm tool originated in x264.
This allows constraining the set of available backends.
Not currently publicly visible, but useful inside the test framework nonetheless.
Needed by h2645_sei.
Adds missing log of struct mfxInfoMFX member: codec level.
Useful to make sure the memcpy_in/out paths work as expected.
__memcmpeq (added in glibc 2.35) was previously an alias to memcmp on POWER10 via strong_alias.