NVIDIA
Suitable trt environment can't serialize model?
Description
I'm trying to deserialize model on device A and serialize it on device B, but failed。Except for the graphics driver, the environment of A and B is the same. Does the graphics driver version affect the model serialization? ERROR_LOG: CUDA_ERROR_INVALID_IMAGE(200): device kernel image is invalid
Environment
TensorRT Version: 7.2.2.3
NVIDIA GPU: 3080
NVIDIA Driver Version: device A: 536.23(max cuda version: 12.2) device B: 516.94(max cuda version: 11.7)
CUDA Version: 11.1
CUDNN Version: 8.0.4
When I update device B's graphics driver to 536.23(device A's version), serialization is ok. But i can't understand this，my cuda version is 11.1, I think both versions of the driver should be able to serialize the model
root cause : ERROR_LOG: CUDA_ERROR_INVALID_IMAGE(200): device kernel image is invalid
Driver version not compatible with your cuda, I suggest you build and run on the same device.
REF https://docs.nvidia.com/deeplearning/tensorrt/release-notes/index.html#rel_7-2-3 to see the right Compatibility set.
Driver version not incompatible with your cuda, I suggest you build and run on the same device.
REF https://docs.nvidia.com/deeplearning/tensorrt/release-notes/index.html#rel_7-2-3 to see the right Compatibility set.
But it should work, cause two device have same CUDA,CUDNN,TENSORRT version, only Graphic driver diff. Both drivers are compatible with the current CUDA version 11.0, one up to 11.7 and the other up to 12.2.