completed thesisHEADmaster

1 files changed, 78 insertions, 0 deletions

diff --git a/python/openvino/runtime/python_demos/OpenVINO_benchmark_app/benchmark.patch b/python/openvino/runtime/python_demos/OpenVINO_benchmark_app/benchmark.patch
new file mode 100644
index 0000000..6696804
--- /dev/null
+++ b/python/openvino/runtime/python_demos/OpenVINO_benchmark_app/benchmark.patch
@@ -0,0 +1,78 @@
+--- /nfs/site/disks/swip_dla_1/resources/inference_engine/2023.3.0_with_dev_tools/1/linux64/suse12/python/openvino/tools/benchmark/benchmark.py	2024-03-01 14:01:50.443877000 -0500
++++ benchmark.py	2024-04-01 10:06:18.751566000 -0400
+@@ -1,14 +1,15 @@
+-# Copyright (C) 2018-2023 Intel Corporation
++# Copyright (C) 2018-2022 Intel Corporation
+ # SPDX-License-Identifier: Apache-2.0
+ 
+ import os
+ from datetime import datetime
+ from math import ceil
++import warnings
+ from openvino.runtime import Core, get_version, AsyncInferQueue
+ 
+-from .utils.constants import GPU_DEVICE_NAME, XML_EXTENSION, BIN_EXTENSION
+-from .utils.logging import logger
+-from .utils.utils import get_duration_seconds
++from openvino.tools.benchmark.utils.constants import GPU_DEVICE_NAME, XML_EXTENSION, BIN_EXTENSION
++from openvino.tools.benchmark.utils.logging import logger
++from openvino.tools.benchmark.utils.utils import get_duration_seconds
+ 
+ def percentile(values, percent):
+     return values[ceil(len(values) * percent / 100) - 1]
+@@ -17,7 +18,17 @@
+     def __init__(self, device: str, number_infer_requests: int = 0, number_iterations: int = None,
+                  duration_seconds: int = None, api_type: str = 'async', inference_only = None):
+         self.device = device
+-        self.core = Core()
++        dla_plugins = os.environ.get('DLA_PLUGINS', default='')
++        if dla_plugins == '':
++            # Backwards compatability for old DLA_PLUGINS_XML_FILE
++            warnings.warn("DLA_PLUGINS_XML_FILE option is deprecated as of 2024.1, Please use DLA_PLUGINS") 
++            dla_plugins = os.environ.get('DLA_PLUGINS_XML_FILE', default='')
++        self.core = Core(dla_plugins)
++        if "FPGA" in self.device:
++            dla_arch_file = os.environ.get('DLA_ARCH_FILE')
++            if dla_arch_file is None:
++                raise Exception(f"To use FPGA, you need to specify the path to an arch_file!")
++            self.core.set_property(device_name="FPGA", properties={"ARCH_PATH": dla_arch_file})
+         self.nireq = number_infer_requests if api_type == 'async' else 1
+         self.niter = number_iterations
+         self.duration_seconds = get_duration_seconds(duration_seconds, self.niter, self.device)
+@@ -59,6 +70,9 @@
+     def set_cache_dir(self, cache_dir: str):
+         self.core.set_property({'CACHE_DIR': cache_dir})
+ 
++    def set_allow_auto_batching(self, flag: bool):
++        self.core.set_property({'ALLOW_AUTO_BATCHING': flag})
++
+     def read_model(self, path_to_model: str):
+         model_filename = os.path.abspath(path_to_model)
+         head, ext = os.path.splitext(model_filename)
+@@ -110,7 +124,7 @@
+               (self.duration_seconds and exec_time < self.duration_seconds) or \
+               (iteration % self.nireq):
+             idle_id = infer_queue.get_idle_request_id()
+-            if idle_id in in_fly:
++            if idle_id in in_fly:       # Is this check neccessary?
+                 times.append(infer_queue[idle_id].latency)
+             else:
+                 in_fly.add(idle_id)
+@@ -162,7 +176,6 @@
+     def main_loop(self, requests, data_queue, batch_size, latency_percentile, pcseq):
+         if self.api_type == 'sync':
+             times, total_duration_sec, iteration = self.sync_inference(requests[0], data_queue)
+-            fps = len(batch_size) * iteration / total_duration_sec
+         elif self.inference_only:
+             times, total_duration_sec, iteration = self.async_inference_only(requests)
+             fps = len(batch_size) * iteration / total_duration_sec
+@@ -175,6 +188,9 @@
+         min_latency_ms = times[0]
+         max_latency_ms = times[-1]
+ 
++        if self.api_type == 'sync':
++            fps = len(batch_size) * 1000 / median_latency_ms
++
+         if pcseq:
+             for group in self.latency_groups:
+                 if group.times: