Releases: intel/vpl-gpu-rt
Releases · intel/vpl-gpu-rt
oneVPL GPU Runtime 2022Q4 Release - 22.6.5
This release of Intel® oneVPL GPU Runtime is part of open source release of media stack. Component revisions included into package:
- oneVPL GPU Runtime: https://github.com/oneapi-src/oneVPL-intel-gpu/releases/tag/intel-onevpl-22.6.5
- oneVPL Dispatcher and Samples: https://github.com/oneapi-src/oneVPL/releases/tag/v2023.1.0
- Driver: https://github.com/intel/media-driver/releases/tag/intel-media-22.6.6
- Gmmlib: https://github.com/intel/gmmlib/releases/tag/intel-gmmlib-22.3.3
- libva: https://github.com/intel/libva/releases/tag/2.17.0
- libva-utils: https://github.com/intel/libva-utils/releases/tag/2.17.1
What’s New in oneVPL GPU Runtime 22.6.5
- Decode:
- Fixed H.264 DPB management issue to avoid no available surface.
- Optimized AV1 film grain memory footprint.
- Improved JPEG height alignment from 16 to 8.
- AV1 Encode:
- Added reset functionality support.
- Fixed memory leak issue for Temporal Unit change.
- Added TCBRC functionality support.
- VP9 Encode:
- Fixed issue for segmentation map alignment.
oneVPL GPU Runtime 2022Q3 Release - 22.5.4
This release of Intel® oneVPL GPU Runtime is part of open source release of media stack. Component revisions included into package:
- oneVPL GPU Runtime: https://github.com/oneapi-src/oneVPL-intel-gpu/releases/tag/intel-onevpl-22.5.4
- oneVPL Dispatcher and Samples: https://github.com/oneapi-src/oneVPL/releases/tag/v2022.2.2
- Driver: https://github.com/intel/media-driver/releases/tag/intel-media-22.5.4
- Gmmlib: https://github.com/intel/gmmlib/releases/tag/intel-gmmlib-22.2.0
- libva: https://github.com/intel/libva/releases/tag/2.16.0
- libva-utils: https://github.com/intel/libva-utils/releases/tag/2.16.0
What’s New in oneVPL GPU Runtime 22.5.4
- Decode:
- Fixed AV1 decode hang issue in high throughput rate transcoding workload.
- Enabled HEVC decode HDR SEI parameters parsing.
- AV1 Encode:
- Enabled Hierarchical B frames and set Mini-Gop Size to 8 for better compression efficiency.
- VP9 Encode:
- Fixed fps calculation issue in IVF header.
- Video Processing
- Added Camera Pipe API for RAW format hard acceleration.
- Enhanced BGRP/I420/UYVY format support.
oneVPL GPU Runtime 2022Q2 Release - 22.4.4
This release of Intel® oneVPL GPU Runtime is part of open source release of media stack. Component revisions included into package:
- oneVPL GPU Runtime: https://github.com/oneapi-src/oneVPL-intel-gpu/releases/tag/intel-onevpl-22.4.4
- oneVPL Dispatcher and Samples: https://github.com/oneapi-src/oneVPL/releases/tag/v2022.1.5
- Driver: https://github.com/intel/media-driver/releases/tag/intel-media-22.4.4
- Gmmlib: https://github.com/intel/gmmlib/releases/tag/intel-gmmlib-22.1.4
- libva: https://github.com/intel/libva/releases/tag/2.15.0
- libva-utils: https://github.com/intel/libva-utils/releases/tag/2.15.0
What’s New in oneVPL GPU Runtime 22.4.4
- Decode:
- Fix VC1 Decode assertion failure when frame type is NONE_PICTURE
- Fix AV1 Decode hang issue in error clips that OBU info size is larger than bitstream size.
oneVPL GPU Runtime 22.3.2
This release of Intel® oneVPL GPU Runtime is part of open source release of media stack. Component revisions included into package:
- oneVPL GPU Runtime: https://github.com/oneapi-src/oneVPL-intel-gpu/releases/tag/intel-onevpl-22.3.2
- oneVPL Dispatcher and Samples: https://github.com/oneapi-src/oneVPL/releases/tag/v2022.0.3
- Driver: https://github.com/intel/media-driver/releases/tag/intel-media-22.3.1
- Gmmlib: https://github.com/intel/gmmlib/releases/tag/intel-gmmlib-22.1.2
- libva: https://github.com/intel/libva/releases/tag/2.14.0
- libva-utils: https://github.com/intel/libva-utils/releases/tag/2.14.0
Release validated with:
- OpenCL runtime: https://github.com/intel/compute-runtime/releases/tag/22.12.22749 (re-built against intel-gmmlib-22.1.2)
What’s New in oneVPL GPU Runtime 22.3.2
New features:
-
API 2.7:
- Enabled Encode Statistics (streamout).
- Upstreamed AV1 Encode.
-
AV1 Decode:
- Large Scale Tile support.
-
HEVC Decode:
- Enabled HDR SEI parsing.
-
VP9 Decode:
- Added PXP Support (Secure Decode).
-
HEVC Encode:
- Enabled LPLA.
-
VPP:
- Added Video Signal Info.
- Implemented new Scaling modes (Compute, VDBox, VEBox).
-
EncTools:
- MBQP Integration.
- Enabled PAQ (Persistence Adaptive QP) for AVC.
-
Misc:
- Enabled CM copy by default for Tiger Lake and Alchemist.
- Upstreamed Raptor Lake S and Alder Lake N platforms support.
- Improved tracing for Encoders and Decoders.
oneVPL GPU Runtime 22.1.0
This release of Intel® oneVPL GPU Runtime is part of open source release of media stack. Component revisions included into package:
- oneVPL GPU Runtime: https://github.com/oneapi-src/oneVPL-intel-gpu/releases/tag/intel-onevpl-22.1.0
- oneVPL Dispatcher and Samples: https://github.com/oneapi-src/oneVPL/releases/tag/v2022.0.2
- Driver: https://github.com/intel/media-driver/releases/tag/intel-media-22.1.1
- Gmmlib: https://github.com/intel/gmmlib/releases/tag/intel-gmmlib-22.0.1
- libva: https://github.com/intel/libva/releases/tag/2.13.0
- libva-utils: https://github.com/intel/libva-utils/releases/tag/2.13.0
Release validated with:
- OpenCL runtime: https://github.com/intel/compute-runtime/releases/tag/22.01.22131 (re-built against intel-gmmlib-22.0.1)
What’s New in oneVPL GPU Runtime 22.1.0
New features:
-
API 2.6
- Added Alchemist (DG2/ATS-M) base support
- Added SurfacePool interface
-
HEVC Encode
- Added maximum and minimum QP settings
-
AVC Encode
- Enabled adaptive CQM
-
AVC Decode
- Removed AVC_HIGH_422, AVC_HIGH10 profiles, that not supported by HW.
-
VP9 Decode
- Enabled gpucopy support
- Enabled VP9 CSC through SFC pipeline
-
Jpeg Decode
- Added missed color format support (yuv400, rgb44)
- improved error message reporting
-
Misc
- Enable layered logging utilities
- Refactored CM integration
- Added issue templates to speed up request filling on github issue page.
oneVPL GPU Runtime 21.3.4
This release of Intel® oneVPL GPU Runtime is part of open source release of media stack. Component revisions included into package:
- oneVPL GPU Runtime: https://github.com/oneapi-src/oneVPL-intel-gpu/releases/tag/intel-onevpl-21.3.4
- oneVPL Dispatcher and Samples: https://github.com/oneapi-src/oneVPL/releases/tag/v2021.6.0
- Driver: https://github.com/intel/media-driver/releases/tag/intel-media-21.3.4
- Gmmlib: https://github.com/intel/gmmlib/releases/tag/intel-gmmlib-21.3.1
- libva: https://github.com/intel/libva/releases/tag/2.13.0
- libva-utils: https://github.com/intel/libva-utils/releases/tag/2.12.0
Release validated with:
- OpenCL runtime: https://github.com/intel/compute-runtime/releases/tag/21.38.21026 (re-built against intel-gmmlib-21.3.1)
What’s New in oneVPL GPU Runtime 21.3.4
New features:
-
VP9 Encode
- Added WebRTC mode.
-
HEVC Encode
- Enable GPU weighted prediction.
- Added BGR color format.
- Switched HEVCe LPLA to EncTools usage.
-
AVC Encode
- Integrated HVS denoise into AVC encoder.
-
EncTools
- Enabled EncTools Library
- Added ET BRC for HEVCe, including LPLA-assisted ET BRC and Persistence Adaptive Quantization.
- Implemented ET control via EXTBRC and LADepth.
- Added mechanism to disable AdaptiveB/ACQM via CO.AdaptiveB/AdaptiveCQM.
- Added AGOP for HEVCe.
- Added closed GOP support for LPLA.
-
VPP
- Added helper that should facilitate the use of VP features.
- Added mfxExtVideoSignalInfo(VideoFullRange, ColourPrimaries) support.
- Enabled HVS Denoise Filter.
-
Misc
- Added clang-12 for github actions.
- Adeed MFX traces to CM.
- Removed opaque memory.