Unsupported SM: 0x601

AI & Data Science Deep Learning (Training & Inference) TensorRT
1

Description

I’m want to use tensorRT on my 1050Ti mobile laptop, when I’ve installed cuda12.1 & cudnn8.9.7 & TensorRT 10 on /usr/local folder and then compiled sample_onnx_mnist to validate TensorRT is usable, but there is an error says “Unsupported SM: 0x601”. Is mobile series GPU unsupport TensorRT?

English is not my native language, sorry for my poor English.

Environment

TensorRT Version: 10.0.0.6
GPU Type: Nvidia GeForce 1050Ti Mobile
Nvidia Driver Version: 545.29.06
CUDA Version: cuda_12.1.r12.1
CUDNN Version: 8.9.7
Operating System + Version: Ubuntu 22.04.4 LTS (x86_64)
Python Version (if applicable): Python 3.10.12
TensorFlow Version (if applicable): 2.16.1
PyTorch Version (if applicable): 2.2.2
Baremetal or Container (if container which image + tag):

image
image1130×467 103 KB

1 Like
2

having the same issue on GTX 1080Ti Desktop GPU

1 Like
3

Same issue here

1 Like
4

Hi @iervn6341
do you mean laptop GPUs or Jetson embedded type devices? The answer is yes to both. TRT supports almost every NVIDIA device except very small GPUs may have some trouble running models that are larger

Thanks

6

I have the same issue on 1080ti with cuda 12.4, cudnn 8.9.7 and trt 10.0.1
The same code and model compiles on older tensorrt python API (ver 8), but then installed TensorRT and python library don’t match.

7

NVIDIA Pascal (SM 6.x) devices are deprecated in TensorRT 8.6.

8

Thanks. I reinstall TensorRT-8.4.3 on my laptop and test sampleMNIST is working right now.

9

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.