lix19937

So please use -pthread as a flag and not -lpthread. This behavior is also mentioned in the following, https://gcc.gnu.org/bugzilla/show_bug.cgi?id=59830 There's a difference between -pthread and -lpthread. Looking at the man...

I think should see **compatibility** https://onnxruntime.ai/docs/reference/compatibility.html

install poppler https://github.com/oschwartz10612/poppler-windows/tags , and then ```py images = convert_from_path('cuda-graph-introduction.pdf', poppler_path=r'E:\python_workspace_transformer\Python-Remove-Watermark-master\poppler-24.07.0\Library\bin') # images = np.array(images) index = 0 ```

+1

I think it overoptimized.

RTX 2080 sm is 75, not supported by libspconv.so which need sm>=80, so it supports Ampere, like RTX3070, 3080, or ORIN.

@lynettez https://github.com/NVIDIA/Deep-Learning-Accelerator-SW/issues/27 how to view DLA utilization rate ?