- Fixed naming issue in parsing
If
conditonal graphs
See the 8.2 EA release notes for new features added in TensorRT 8.2.
- Removed duplicate constant layer checks that caused some performance regressions
- Fixed expand dynamic shape calculations
- Added parser-side checks for Scatter layer support
- Added support for the following ONNX operators:
- Einsum
- IsNan
- GatherND
- Scatter
- ScatterElements
- ScatterND
- Sign
- Round
- Updated
Gather
andGatherElements
implementations to natively support negative indices - Updated
Pad
layer to support ND padding, along withedge
andreflect
padding mode support - Updated
If
layer with general performance improvements.
- Rehauled resize operator, now fully supporting the following modes:
- Coordinate Transformation modes:
half_pixel
,pytorch_half_pixel
,tf_half_pixel_for_nn
,asymmetric
, andalign_corners
- Modes:
nearest
,linear
- Nearest Modes:
floor
,ceil
,round_prefer_floor
,round_prefer_ceil
- Coordinate Transformation modes:
- QuantizeLinear/DequantizeLinear updates:
- Added support for tensor scales
- Added support for per-axis quantization
- Added support for multi-input ConvTranpose
- Added support for generic 2D padding
- Added experimental support for
NonMaxSuppression
- Moved
RefitMap
API to core TensorRT. - Added Datatype column to operators.md
- Added library only build target #659
- Added support for negative gather indices #681
- Added support for
DOUBLE
-typed inputs and weights through downcast to float #674 - Added support for optional plugin fields in FallbackPlugin path #676
- Updated license #657
- Fixed index offset calculation in GatherElements #675
- Clarified dynamic shape support for ReverseSequence
- Added opset13 support for
SoftMax
,LogSoftmax
,Squeeze
, andUnsqueeze
- Added support for the
EyeLike
operator - Added support for the
GatherElements
operator
- Added support for the
ReverseSequence
operator #590 - Updated
parse()
andsupportsModel()
API calls with an optionalmodel_path
parameter to support models with external weights #621 - Added support for the
Celu
operator - Added support for the
CumSum
operator - Added support for the
LessOrEqual
operator - Added support for the
LpNormalization
operator - Added support for the
LpPool
operator - Added support for the
GreaterOrEqual
operator - Added support for dynamic inputs in
onnx_tensorrt
python backend - Added FAQ section for commonly asked questions
- Fixed relative path imports for models with external weights [#619]onnx#619
- Fixed importing loops operators with no loop-carried depedencies #619
- Worked around unsupported BOOL concats through casting #620
- Fixed compilation error with GCC9 #568
- Removed
onnx_tensorrt/config.py
as it is no longer needed
- Added
setup.py
to properly installonnx_tensorrt
python backend - Added 4D transpose for ONNX weights #557
- Fixed slice computations for large slices #558
- Added support for parsing large models with external data
- Added API for interfacing with TensorRT's refit feature
- Updated
onnx_tensorrt
backend to support dynamic shapes - Added support for 3D instance normalizations #515
- Improved clarity on the resize modes TRT supports #512
- Added Changelog
- Unified docker usage between ONNX-TensorRT and TensorRT.
- Removed deprecated docker files.
- Removed deprecated
setup.py
.