Skip to content

Arm backend: Make it easier to generate non delegated/quantized PTEs #10387

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Apr 23, 2025
Merged

Arm backend: Make it easier to generate non delegated/quantized PTEs #10387

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 9 additions & 3 deletions examples/arm/run.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,7 +20,9 @@ et_root_dir=$(realpath ${et_root_dir})
model_name=""
model_input_set=false
model_input=""
aot_arm_compiler_flags="--delegate --quantize"
aot_arm_compiler_flag_delegate="--delegate"
aot_arm_compiler_flag_quantize="--quantize"
aot_arm_compiler_flags=""
portable_kernels="aten::_softmax.out"
target="ethos-u55-128"
output_folder_set=false
Expand All @@ -41,7 +43,9 @@ function help() {
echo " --model_name=<MODEL> Model file .py/.pth/.pt, builtin model or a model from examples/models. Passed to aot_arm_compiler"
echo " --model_input=<INPUT> Provide model input .pt file to override the input in the model file. Passed to aot_arm_compiler"
echo " NOTE: Inference in FVP is done with a dummy input full of ones. Use bundleio flag to run the model in FVP with the custom input or the input from the model file."
echo " --aot_arm_compiler_flags=<FLAGS> Only used if --model_name is used Default: ${aot_arm_compiler_flags}"
echo " --aot_arm_compiler_flags=<FLAGS> Extra flags to pass to aot compiler"
echo " --no_delegate Do not delegate the model (can't override builtin models)"
echo " --no_quantize Do not quantize the model (can't override builtin models)"
echo " --portable_kernels=<OPS> Comma separated list of portable (non delagated) kernels to include Default: ${portable_kernels}"
echo " --target=<TARGET> Target to build and run for Default: ${target}"
echo " --output=<FOLDER> Target build output folder Default: ${output_folder}"
Expand All @@ -64,6 +68,8 @@ for arg in "$@"; do
--model_name=*) model_name="${arg#*=}";;
--model_input=*) model_input="${arg#*=}" ; model_input_set=true ;;
--aot_arm_compiler_flags=*) aot_arm_compiler_flags="${arg#*=}";;
--no_delegate) aot_arm_compiler_flag_delegate="" ;;
--no_quantize) aot_arm_compiler_flag_quantize="" ;;
--portable_kernels=*) portable_kernels="${arg#*=}";;
--target=*) target="${arg#*=}";;
--output=*) output_folder="${arg#*=}" ; output_folder_set=true ;;
Expand Down Expand Up @@ -160,7 +166,7 @@ if [[ -z "$model_name" ]]; then
model_compiler_flags=( "" "--delegate" "--delegate" "--delegate --quantize" )
else
test_model=( "$model_name" )
model_compiler_flags=( "$aot_arm_compiler_flags" )
model_compiler_flags=( "$aot_arm_compiler_flag_delegate $aot_arm_compiler_flag_quantize $aot_arm_compiler_flags" )
fi

# loop over running the AoT flow and executing the model on device
Expand Down
Loading