pylib: add run_simulator runner

Fixes #106

pylib/run_simulator.py

@@ -0,0 +1,55 @@
+#!/usr/bin/env python3
+
+# Python module to run programs natively.
+
+# Copyright (C) 2019 Clemson University
+# Copyright (C) 2023 University of Bremen
+#
+# Contributor: Ola Jeppsson <[email protected]>
+# Contributor Sören Tempel <[email protected]>
+#
+# This file is part of Embench.
+
+# SPDX-License-Identifier: GPL-3.0-or-later
+
+"""
+Embench module to run benchmark programs.
+
+This version is suitable for running programs within a simulator.
+"""
+
+__all__ = [
+    'get_target_args',
+    'build_benchmark_cmd',
+    'decode_results',
+]
+
+import argparse
+import re
+
+from run_native import decode_results
+from embench_core import log
+
+
+def get_target_args(remnant):
+    """Parse left over arguments"""
+    parser = argparse.ArgumentParser(description='Get target specific args')
+    parser.add_argument(
+        '--simulator',
+        type=str,
+        help='Simulator command to run benchmarks with'
+    )
+
+    # No target arguments
+    return parser.parse_args(remnant)
+
+
+def build_benchmark_cmd(bench, args):
+    """Construct the command to run the benchmark.  "args" is a
+       namespace with target specific arguments"""
+    simulator = args.simulator
+
+    # Due to way the target interface currently works we need to construct
+    # a command that records both the return value and execution time to
+    # stdin/stdout. Obviously using time will not be very precise.
+    return ['sh', '-c', f'time -p {simulator} ./' + bench + '; echo RET=$?']