github/SHARK-Studio
1
1
Fork 0
mirror of https://github.com/nod-ai/SHARK-Studio.git synced 2026-01-10 06:17:55 -05:00
Code Issues Packages Projects Releases Wiki Activity

Update benchmark command to ToM and Add to test (#125)

Browse Source 
-Update benchmark_cl to latest benchmark_module API
-tensorflow OOM issue.
This commit is contained in:
Stanley Winata
2022-06-13 23:57:05 -07:00
committed by GitHub
parent 61b19ce091
commit e48642f8a5
8 changed files with 37 additions and 30 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
.github/workflows/test-models.yml vendored
View File

@@ -52,7 +52,7 @@ jobs:
cd $GITHUB_WORKSPACE
IMPORTER=1 ./setup_venv.sh
source shark.venv/bin/activate
pytest -k 'not benchmark' --ignore=benchmarks/tests/test_hf_benchmark.py --ignore=benchmarks/tests/test_benchmark.py --ignore=tank/tf/ --ignore=shark/tests/test_shark_importer.py
pytest -k 'not benchmark' --ignore=tank/tf/ --ignore=shark/tests/test_shark_importer.py
perf-macOS:
runs-on: MacStudio

2
.gitignore vendored
View File

@@ -161,7 +161,7 @@ cython_debug/
#.idea/
# Shark related artefacts
shark.venv/
*venv/
# ORT related artefacts
cache_models/

24
benchmarks/tests/test_benchmark.py
View File

@@ -11,6 +11,9 @@ import pytest
import unittest
torch.manual_seed(0)
gpus = tf.config.experimental.list_physical_devices('GPU')
for gpu in gpus:
tf.config.experimental.set_memory_growth(gpu, True)
##################### Tensorflow Hugging Face LM Models ###################################
MAX_SEQUENCE_LENGTH = 512
@@ -42,8 +45,7 @@ class TFHuggingFaceLanguage(tf.Module):
def get_TFhf_model(name):
model = TFHuggingFaceLanguage(name)
tokenizer = BertTokenizer.from_pretrained(
"microsoft/MiniLM-L12-H384-uncased")
tokenizer = BertTokenizer.from_pretrained(name)
text = "Replace me by any text you'd like."
encoded_input = tokenizer(text,
padding='max_length',
@@ -124,20 +126,19 @@ pytest_benchmark_param = pytest.mark.parametrize(
'gpu',
marks=pytest.mark.skipif(check_device_drivers("gpu"),
reason="nvidia-smi not found")),
pytest.param(True,
'gpu',
marks=pytest.mark.skipif(check_device_drivers("gpu"),
reason="nvidia-smi not found")),
pytest.param(True,
'gpu',
marks=pytest.mark.skip),
pytest.param(
False,
'vulkan',
False,
'vulkan',
marks=pytest.mark.skipif(
check_device_drivers("vulkan"),
reason="vulkaninfo not found, install from https://github.com/KhronosGroup/MoltenVK/releases"
)),
pytest.param(
True,
'vulkan',
True,
'vulkan',
marks=pytest.mark.skipif(
check_device_drivers("vulkan"),
reason="vulkaninfo not found, install from https://github.com/KhronosGroup/MoltenVK/releases"
@@ -188,8 +189,7 @@ def test_bench_distilbert(dynamic, device):
assert False
@pytest.mark.skipif(importlib.util.find_spec("iree.tools") is None,
reason="Cannot find tools to import TF")
@pytest.mark.skip(reason="XLM Roberta too large to test.")
@pytest_benchmark_param
def test_bench_xlm_roberta(dynamic, device):
model, test_input, act_out = get_TFhf_model("xlm-roberta-base")

2
pyproject.toml
View File

@@ -4,7 +4,7 @@ requires = [
"wheel",
"packaging",
"numpy",
"numpy==1.22.4",
"torch-mlir>=20220428.420",
"iree-compiler>=20220427.13",
"iree-runtime>=20220427.13",

2
setup_venv.sh
View File

@@ -72,7 +72,7 @@ fi
# Upgrade pip and install requirements.
$PYTHON -m pip install --upgrade pip || die "Could not upgrade pip"
$PYTHON -m pip install --upgrade -r "$TD/requirements.txt" --extra-index-url https://download.pytorch.org/whl/nightly/cpu -f https://github.com/llvm/torch-mlir/releases
$PYTHON -m pip install --upgrade -r "$TD/requirements.txt"
if [ "$torch_mlir_bin" = true ]; then
$PYTHON -m pip install --find-links https://github.com/llvm/torch-mlir/releases torch-mlir --extra-index-url https://download.pytorch.org/whl/nightly/cpu
if [ $? -eq 0 ];then

8
shark/iree_utils.py
View File

@@ -42,7 +42,6 @@ IREE_TARGET_MAP = {
"rocm": "rocm"
}
UNIT_TO_SECOND_MAP = {"ms": 0.001, "s": 1}
@@ -112,7 +111,7 @@ def get_vulkan_triple_flag():
return "-iree-vulkan-target-triple=ampere-rtx3080-linux"
else:
print(
"Optimized kernel for your target device is not added yet. Contact SHARK Admin on discord or pull up an issue."
"Optimized kernel for your target device is not added yet. Contact SHARK Admin on discord[https://discord.com/invite/RUqY2h2s9u] or pull up an issue."
)
return None
@@ -178,8 +177,7 @@ def compile_module_to_flatbuffer(module, device, frontend, func_name,
module = str(module)
# Compile according to the input type, else just try compiling.
print(type(module))
if input_type not in ["mhlo","tosa"]:
if input_type not in ["mhlo", "tosa"]:
module = str(module)
if input_type != "":
# Currently for MHLO/TOSA.
@@ -318,7 +316,7 @@ def build_benchmark_args(input_file: str,
# TODO: Replace name of train with actual train fn name.
fn_name = "train"
benchmark_cl.append(f"--entry_function={fn_name}")
benchmark_cl.append(f"--driver={IREE_DEVICE_MAP[device]}")
benchmark_cl.append(f"--device={IREE_DEVICE_MAP[device]}")
mlir_input_types = tensor_to_type_str(input_tensors, frontend)
for mlir_input in mlir_input_types:
benchmark_cl.append(f"--function_input={mlir_input}")

22
shark/shark_importer.py
View File

@@ -12,6 +12,7 @@ from shark.shark_inference import SharkInference
class SharkImporter:
def __init__(self,
model_path,
model_type: str = "tflite",
@@ -41,22 +42,28 @@ class SharkImporter:
if self.model_type == "tflite":
print("Setting up for TMP_DIR")
exe_basename = os.path.basename(sys.argv[0])
self.workdir = os.path.join(os.path.dirname(__file__), "tmp", exe_basename)
self.workdir = os.path.join(os.path.dirname(__file__), "tmp",
exe_basename)
print(f"TMP_DIR = {self.workdir}")
os.makedirs(self.workdir, exist_ok=True)
self.tflite_file = '/'.join([self.workdir, 'model.tflite'])
print("Setting up local address for tflite model file: ", self.tflite_file)
print("Setting up local address for tflite model file: ",
self.tflite_file)
if os.path.exists(self.model_path):
self.tflite_file = self.model_path
else:
print("Download tflite model")
urllib.request.urlretrieve(self.model_path, self.tflite_file)
urllib.request.urlretrieve(self.model_path,
self.tflite_file)
print("Setting up tflite interpreter")
self.tflite_interpreter = tf.lite.Interpreter(model_path=self.tflite_file)
self.tflite_interpreter = tf.lite.Interpreter(
model_path=self.tflite_file)
self.tflite_interpreter.allocate_tensors()
# default input initialization
self.input_details, self.output_details = self.get_model_details()
inputs = self.generate_inputs(self.input_details) # device_inputs
self.input_details, self.output_details = self.get_model_details(
)
inputs = self.generate_inputs(
self.input_details) # device_inputs
self.setup_inputs(inputs)
def generate_inputs(self, input_details):
@@ -85,7 +92,8 @@ class SharkImporter:
if self.model_source_hub == "tfhub":
# compile and run tfhub tflite
print("Inference tfhub model")
self.shark_module = SharkInference(self.tflite_file, self.inputs,
self.shark_module = SharkInference(self.tflite_file,
self.inputs,
device=self.device,
dynamic=self.dynamic,
jit_trace=self.jit_trace)

5
shark/shark_inference.py
View File

@@ -51,7 +51,8 @@ class SharkInference:
# Sets the frontend i.e `pytorch` or `tensorflow`.
def set_frontend(self, frontend: str):
if frontend not in [
"pytorch", "torch", "tensorflow", "tf", "mhlo", "linalg", "tosa", "tflite"
"pytorch", "torch", "tensorflow", "tf", "mhlo", "linalg",
"tosa", "tflite"
]:
print_err("frontend not supported.")
else:
@@ -94,7 +95,7 @@ class SharkInference:
@benchmark_mode
def benchmark_all(self, inputs):
self.shark_runner.benchmark_all(inputs)
@benchmark_mode
def benchmark_frontend(self, inputs):
self.shark_runner.benchmark_frontend(inputs)