I convert a onnx model to tensorrt.But found there is a “inf” result in a intermedia output from a Pow(2) operator. the dtype of the operator’s input and output tensor are all float32. Is seems precision reduced in operator Pow. How can I forbid the precision reduction?
convert log :
the result log:
the convert command line： trtexec --onnx="$onnx" \ --saveEngine="$engine_file" \ --verbose --explicitBatch --workspace=262144
tensorrt version : TensorRT v8204
Hi,
Could you please try on the latest TensorRT version 8.4 EA. If still face this issue, please share with us the issue repro model/script and steps to try from our end.
Thank you.
NVES
May 27, 2022, 12:06pm
3
Hi ,
These support matrices provide a look into the supported platforms, features, and hardware capabilities of the NVIDIA TensorRT 8.4.3 APIs, parsers, and layers.
You can refer below link for all the supported operators list.
<!--- SPDX-License-Identifier: Apache-2.0 -->
# Supported ONNX Operators
TensorRT 8.4 supports operators up to Opset 17. Latest information of ONNX operators can be found [here](https://github.com/onnx/onnx/blob/master/docs/Operators.md)
TensorRT supports the following ONNX data types: DOUBLE, FLOAT32, FLOAT16, INT8, and BOOL
> Note: There is limited support for INT32, INT64, and DOUBLE types. TensorRT will attempt to cast down INT64 to INT32 and DOUBLE down to FLOAT, clamping values to `+-INT_MAX` or `+-FLT_MAX` if necessary.
See below for the support matrix of ONNX operators in ONNX-TensorRT.
## Operator Support Matrix
| Operator | Supported | Supported Types | Restrictions |
|---------------------------|------------|-----------------|------------------------------------------------------------------------------------------------------------------------|
| Abs | Y | FP32, FP16, INT32 |
| Acos | Y | FP32, FP16 |
| Acosh | Y | FP32, FP16 |
| Add | Y | FP32, FP16, INT32 |
show original
Thanks!
