pykeio / ort

Fast ML inference & training for Rust with ONNX Runtime

Home Page:https://ort.pyke.io/

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

takes 10 minutes to load model

NofAssiri opened this issue · comments

i have 150MB onnx model that takes about 10 minutes to load with onnxruntime.dll 1.16.3, but when using onnxruntime.dll version 1.10.2201 in onnxruntime crate it loads in 4,5 seconds

platform:
windows 11 pro 64-bit build 22621
processor:
12th Gen Intel Core i7-12800HX
chip Type:
AMD Software: Pro Edition 0x73E1 VRAM 8GB
Ram:
32GB

code used
let _session = Session::builder().expect("builder error ")
.with_optimization_level(ort::GraphOptimizationLevel::Disable).expect("opt level error 14")
.with_optimization_level(GraphOptimizationLevel::Level1).unwrap()
.with_model_from_file("C:\models\model.onnx").expect("model file error 15");