v2022.2.0

New in this release

Depreciated

• Support for Microsoft Visual Studio* 2017

Issues and Limitations

• CPU implementation is currently for reference only and not intended to be competitive in terms of performance or features.
• CPU implementation default colorspaces are I420 and I010. This does not match GPU default colorspaces of NV12/P010. For sample_* tools, colorspace flags must be added to command lines to work around this issue. For example code and application code, additional logic must be added to handle default colorspace differences.
  sample_encode does not support preprocess commands for scaling or color conversion for CPU
• CPU compilation on WSL2 is not currently supported, but precompiled CPU implementation will work on WSL2 systems.
• Decode input bitstream buffer size must be large enough to hold several frames. Buffer sizes that are too small may cause issues. The necessary minimum size is stream dependent but enough for 10 frames is a conservative estimate.
• CPU implementation decode error recovery is limited. Decode of corrupted bitstreams may fail.
• Only a small subset of parameters and functions from the specification are implemented. For more information on mandatory and optional APIs and features, see the oneVPL specification.
  MFX_BITSTREAM_EOS from the input bitstream dataFlag is ignored by MFXVideoDECODE_DecodeFrameAsync().
• Windows* OS app verifier detects memory leak from CPU implementation when openH264 is enabled.