Update benchmark util for example using (#11027)

* mv benchmark_util.py to utils/

* remove

* update

python/llm/dev/benchmark/README.md

@@ -1,6 +1,6 @@
 # Benchmark tool for transformers int4 (separate 1st token and rest)
 
-`benchmark_util.py` is used to provide a simple benchmark tool for transformer int4 model to calculate 1st token performance and the rest on CPU and GPU.
+[benchmark_util.py](https://github.com/intel-analytics/ipex-llm/tree/main/python/llm/src/ipex_llm/utils/benchmark_util.py) is used to provide a simple benchmark tool for transformer int4 model to calculate 1st token performance and the rest on CPU and GPU.
 
 ## CPU Usage
 Just put this file into your benchmark directory, and then wrap your transformer int4 model with `BenchmarkWrapper` (`model = BenchmarkWrapper(model)`).
@@ -9,7 +9,7 @@ Take `chatglm-6b` as an example:
 import torch
 from ipex_llm.transformers import AutoModel
 from transformers import AutoTokenizer
-from benchmark_util import BenchmarkWrapper
+from ipex_llm.utils.benchmark_util import BenchmarkWrapper
 
 model_path ='THUDM/chatglm-6b'
 model = AutoModel.from_pretrained(model_path, trust_remote_code=True, load_in_4bit=True)
@@ -37,7 +37,7 @@ import torch
 import intel_extension_for_pytorch as ipex
 from ipex_llm.transformers import AutoModel
 from transformers import AutoTokenizer
-from benchmark_util import BenchmarkWrapper
+from ipex_llm.utils.benchmark_util import BenchmarkWrapper
 
 model_path ='THUDM/chatglm-6b'
 model = AutoModel.from_pretrained(model_path, trust_remote_code=True, load_in_4bit=True)
@@ -66,7 +66,7 @@ For example, just need to apply following code patch on [Deepspeed Autotp exampl
  import torch
  import transformers
  import deepspeed
-+from benchmark_util import BenchmarkWrapper
++from ipex_llm.utils.benchmark_util import BenchmarkWrapper
  
  def get_int_from_env(env_keys, default):
      """Returns the first positive env value found in the `env_keys` list or the default."""

python/llm/dev/benchmark/all-in-one/run-stress-test.py

@@ -27,10 +27,8 @@ from datetime import date
 
 import os
 current_dir = os.path.dirname(os.path.realpath(__file__))
-benchmark_util_path = os.path.join(current_dir, '..')
 import sys
-sys.path.append(benchmark_util_path)
+from ipex_llm.utils.benchmark_util import BenchmarkWrapper
-from benchmark_util import BenchmarkWrapper
 from ipex_llm.utils.common.log4Error import invalidInputError
 
 LLAMA_IDS = ['meta-llama/Llama-2-7b-chat-hf','meta-llama/Llama-2-13b-chat-hf',

python/llm/dev/benchmark/all-in-one/run.py

@@ -28,10 +28,8 @@ from datetime import date
 
 import os
 current_dir = os.path.dirname(os.path.realpath(__file__))
-benchmark_util_path = os.path.join(current_dir, '..')
 import sys
-sys.path.append(benchmark_util_path)
+from ipex_llm.utils.benchmark_util import BenchmarkWrapper
-from benchmark_util import BenchmarkWrapper
 from ipex_llm.utils.common.log4Error import invalidInputError
 from ipex_llm.utils.common import invalidInputError
 

python/llm/dev/test/lint-python

@@ -21,7 +21,7 @@ SCRIPT_DIR="$( cd "$( dirname "$0" )" && pwd )"
 PYTHON_ROOT_DIR="$SCRIPT_DIR/.."
 echo $PYTHON_ROOT_DIR
 PATHS_TO_CHECK="$SCRIPT_DIR/../../src"
-PATTERNS_TO_EXCLUDE="__init__.py,log4Error.py,$SCRIPT_DIR/../../src/ipex_llm/langchain/*,$SCRIPT_DIR/../../src/ipex_llm/transformers/gguf/models/model_implement/yuan2/*"
+PATTERNS_TO_EXCLUDE="__init__.py,log4Error.py,$SCRIPT_DIR/../../src/ipex_llm/langchain/*,$SCRIPT_DIR/../../src/ipex_llm/transformers/gguf/models/model_implement/yuan2/*,benchmark_util.py"
 PEP8_REPORT_PATH="$PYTHON_ROOT_DIR/test/pep8-report.txt"
 PYLINT_REPORT_PATH="$PYTHON_ROOT_DIR/test/pylint-report.txt"
 PYLINT_INSTALL_INFO="$PYTHON_ROOT_DIR/test/pylint-info.txt"

python/llm/example/GPU/Deepspeed-AutoTP-FastAPI/serving.py

@@ -31,7 +31,7 @@ from typing import Dict, List, Optional
 from transformers.utils import logging
 logger = logging.get_logger(__name__)
 
-from benchmark_util import BenchmarkWrapper
+from ipex_llm.utils.benchmark_util import BenchmarkWrapper
 
 
 def get_int_from_env(env_keys, default):