From 8bc0d50e8de8ebbdd41a6312713730f218eefcf3 Mon Sep 17 00:00:00 2001 From: Johnnie Birch Date: Wed, 4 Oct 2023 12:19:58 -0700 Subject: [PATCH] Update instructions to run image-classify benchmark --- benchmarks/image-classification/README.md | 2 +- benchmarks/image-classification/setup.sh | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/benchmarks/image-classification/README.md b/benchmarks/image-classification/README.md index 11c87eae..92d518b1 100644 --- a/benchmarks/image-classification/README.md +++ b/benchmarks/image-classification/README.md @@ -6,7 +6,7 @@ learning (ML). It uses the OpenVINO backend to identify the subject of the test ## Steps to run this benchmark from the top sightglass directory - `./benchmarks/image-classification/setup.sh` - `source benchmarks/image-classification/openvino/setupvars.sh` -- `cargo run -- benchmark --engine engines/wasmtime/libengine.so --engine-flags=--wasi-modules=experimental-wasi-nn -- benchmarks/image-classification/image-classification-benchmark.wasm` +- `cargo run -- benchmark --engine engines/wasmtime/libengine.so --engine-flags=--wasi nn -- benchmarks/image-classification/image-classification-benchmark.wasm` ## Notes: - You must install and setup OpenVINO to run this benchmark. You can install OpenVINO and the other required files by running `setup.sh`. You'll also need to call `source benchmarks/image-classification/openvino/setupvars.sh` to set up the OpenVINO enviromnent variables each time you start a new terminal. - You must use `--engine-flags=--wasi-modules=experimental-wasi-nn` when running this benchmark. Otherwise Wasmtime won't link the required wasi-nn functions. \ No newline at end of file diff --git a/benchmarks/image-classification/setup.sh b/benchmarks/image-classification/setup.sh index 3e85fa8c..74b6f400 100755 --- a/benchmarks/image-classification/setup.sh +++ b/benchmarks/image-classification/setup.sh @@ -1,7 +1,7 @@ #!/bin/bash # This will download a copy of OpenVINO and extract it locally, as well as the needed model files and test image. Once this script is run, you can use this command to run the benchmark: -# cargo run -- benchmark benchmarks/image-classification/image-classification-benchmark.wasm --engine-flags="--wasi-modules=experimental-wasi-nn" --engine engines/wasmtime/libengine.so +# cargo run -- benchmark benchmarks/image-classification/image-classification-benchmark.wasm --engine-flags="--wasi nn" --engine engines/wasmtime/libengine.so WASI_NN_DIR=$( cd -- "$( dirname -- "${BASH_SOURCE[0]}" )" &> /dev/null && pwd ) FILENAME=l_openvino_toolkit_ubuntu20_2022.2.0.7713.af16ea1d79a_x86_64