The release of the SVT-AV1 3.1.0 library (Scalable Video Technology AV1) has been published, featuring implementations of both encoder and decoder for the AV1 video coding format, utilizing the x86_64 and ARM hardware parallel computing extensions present in processors for acceleration. The project was created by Intel in partnership with Netflix, aiming to achieve a performance level suitable for real-time video transcoding and application in video-on-demand (VOD) services. Currently, development is being conducted under the auspices of the Alliance for Open Media (AOMedia), which oversees the development of the AV1 video coding format. Previously, the project was developed under the OpenVisualCloud project, which also develops the SVT-HEVC and SVT-VP9 encoders. The code is distributed under the BSD license.
SVT-AV1 can be built for systems based on any architectures for which a C99 compliant compiler is available, but the best performance is achieved on x86_64 systems, where assembler optimizations based on SIMD instructions are applied (CPU support for AVX2 is preferable, but SSE2 is the minimum requirement). Memory consumption depends on the number of processor cores engaged in encoding, regulated by the ‘—lp’ option. Due to the complexity of the algorithms used in AV1, significantly more resources are required for encoding in this format compared to others; for instance, the native encoder from the AV1 project requires 5721, 5869, and 658 times more computations compared to the x264 (main profile), x264 (high profile), and libvpx-vp9 encoders.
Among the changes in the new SVT-AV1 release:
- Optimizations for quality/speed trade-offs have been made for presets that set an average and high level of quality. In 'fast-decode 0' mode for presets M1-M5, speed has increased by 15-25% while maintaining quality levels. In fast-decode modes 1 and 2 for presets M3-M7, speed has improved by 15-20% while preserving quality. For presets M0 and MR, the quality level has increased by approximately 1-1.5% according to the BD-Rate (Bjontegaard-Delta) assessment method.
- Significant improvements have been made to the low latency (Low Delay) encoding mode.
- Performance in RTC mode has increased by 5-10%.
- On ARM systems, optimizations based on the Neon and SVE2 extensions have been implemented, allowing for a 5% increase in encoding performance of streams at low resolution with high color depth.
- On ARM systems, video encoding performance has been improved in landscape (1080p +4%, 720p +6%, 480p: +6%, 360p: +3%, 240p: +4%) and portrait (1080p +8%, 720p +4%, 480p +3%, 360p +7%, 240p +4%) modes.
- Random access mode now supports S-frames.
- The implementation of the flags "—chroma-qm-min" and "—chroma-qm-max" has been transferred from the SVT-AV1-SPY fork, along with improvements related to Variance Boost (the adaptive quantization VAQ algorithm implementation).
- RTC mode now includes presets M11 and M12.
Source: opennet.ru
