bash script to compile and run using iree

run_mul_2.sh

out=notes/mlir/mul_2
file=mul_2

iree-compile ${out}/${file}.mlir \
    --iree-hal-target-backends=llvm-cpu \
    --iree-hal-dump-executable-files-to=${out} \
    -o ${out}/${file}.vmfb
    # --iree-hal-cuda-llvm-target-arch=sm_80 \
    # --dump-compilation-phases-to=${out} \

# nsys nvprof --print-gpu-trace \
iree-run-module \
    --device=local-task \
    --module=${out}/${file}.vmfb \
    --function=main \
    --input="4xf32=@${out}/input.npy" \
    # --output="@${out}/output_cuda.npy"
    # --input="4xf32=[0,1,2,3]" \