use onnx with tensorrt

#1
by Abdorifaat - opened

can you provide the code to use it with tensorrt because I use tensorrt on the model.onnx and it takes 470 ms which is too much on 4090 gpu

when I use tensorrt with model_fp16.onnx model it takes 260 ms which is also too much

when I use tensorrt with the int8 model I got the following error
notimplemented: [onnxruntimeerror] : 9 : not_implemented : could not find an implementation for convinteger(10) node with name '/owlv2/vision_model/embeddings/patch_embedding/conv_quant'

I think the tensorrt doesnot work with me like that or there something I am doing wrongly

ONNX Community org

Hi, @Abdorifaat ! I don't have an answer for it, but I came to suggest you create an issue on NVIDIA/TensorRT repository, explaining it.

Sign up or log in to comment