Can I upgrade tensorRT 7.0.x in deepstream5.0 to 7.1.x in dGPU？
Because I learned that only Jetson could match tensorRT 7.1.x and CUDnn 8.
I tried to deploy my own model to DeepStream, but got an error：
ERROR: …/nvdsinfer/nvdsinfer_func_utils.cpp:31 [TRT]: (Unnamed Layer* 176) [Constant]: invalid weights type of Bool
ERROR: …/nvdsinfer/nvdsinfer_func_utils.cpp:31 [TRT]: (Unnamed Layer* 176) [Constant]: invalid weights type of Bool
Could it be the tensorRT version? If it is a version issue, how do I upgrade in dGPU?
Hi,
Deepstream do support TensorRT 7.1.
https://docs.nvidia.com/metropolis/deepstream/dev-guide/index.html
Based on the error, it should be a compatibility issue.
Please removed the TensorRT engine created by previous TensorRT package first.
Thanks.