Merge pull request #1211 from Pinata-Consulting/plot-running-times

Plot running times

Author: maliberty

flow/designs/src/mock-array/plot-running-times.py

@@ -0,0 +1,106 @@
+import subprocess
+import os
+import re
+from matplotlib import pyplot as plt
+import numpy as np
+from matplotlib import cm
+import itertools
+
+def run():
+    test_design = 'make DESIGN_CONFIG=designs/asap7/mock-array/config.mk'
+
+    times = []
+
+    # measure three variables at the time.
+    measurements = {'datawidth': (('MOCK_ARRAY_DATAWIDTH',), (8,)),
+                    'arraysize': (('MOCK_ARRAY_WIDTH', 'MOCK_ARRAY_HEIGHT'), (8,)),
+                    'pitches': (('MOCK_ARRAY_TABLE',), ('8 8 4 4 5 5',)),
+                    'elementsize': (('MOCK_ARRAY_SCALE',), (45, 80, 160, 320, 640))}
+    measure_ids = sorted(measurements.keys())
+
+    for measurement in itertools.product(*map(lambda key: measurements[key][1], measure_ids)):
+        variant = '-'.join(map(str, measurement)).replace(' ', '-')
+        print(f'testiong {variant}')
+        env_change = {'FLOW_VARIANT': variant}
+        for e in itertools.chain(*map(lambda measure: map(lambda var: {var: measure[1]}, measurements[measure[0]][0]),
+            zip(measure_ids, measurement))):
+            u = dict(map(lambda item: (item[0], str(item[1])), e.items()))
+            env_change.update(u)
+        env = os.environ.copy()
+        env.update(env_change)
+
+        if not os.path.exists(f'results/asap7/mock-array_Element/{variant}/'):
+            print(f"Measuring {variant}")
+            for cmd in (test_design + ' verilog', test_design,):
+                returncode = subprocess.call(cmd,
+                env=env,
+                shell=True)
+                if returncode != 0:
+                    print("Skipping variant, doesn't compile: " + variant)
+        result = subprocess.check_output(test_design + " elapsed", shell=True, env=env).decode('utf-8')
+
+        # Modify pattern to capture the stage of interest
+        # Log                       Elapsed seconds
+        # logname                            40
+        pattern = r'^5_2_route\s+(\d+)$'
+        match = re.search(pattern, result, re.MULTILINE)
+        if match is None:
+            print("Variant skipped: " + variant)
+            continue
+        value = int(match.group(1))
+        sample = list(measurement) + [value]
+        print(' '.join(map(str, sample)))
+        times.append(sample)
+
+    dimensions = sum(map(lambda id: len(measurements[id][1]) > 1, measure_ids))
+    if dimensions == 3:
+        # 4 dimensional plot
+        # plt.rcParams["figure.figsize"] = [7.00, 3.50]
+        plt.rcParams["figure.autolayout"] = True
+        fig = plt.figure()
+        ax = fig.add_subplot(111, projection='3d')
+
+        transposed = list(map(list, zip(*times)))
+
+        norm = plt.Normalize(min(transposed[-1]), max(transposed[-1]))
+        colors = cm.viridis(norm(transposed[-1]))
+
+        cbar = fig.colorbar(cm.ScalarMappable(norm=norm, cmap=cm.viridis), ax=ax)
+        cbar.set_label("running time/seconds")
+
+        img = ax.scatter(*transposed,  c=colors, alpha=1)
+
+        ax.set_xlabel(measure_ids[0])
+        ax.set_ylabel(measure_ids[1])
+        ax.set_zlabel(measure_ids[2])
+        plt.show()
+    elif dimensions == 1:
+        list(enumerate(map(lambda id: len(measurements[id][1]) > 1, measure_ids)))
+        measure = next((i for i, e in enumerate(map(lambda id: len(measurements[id][1]) > 1, measure_ids)) if e), -1)
+
+        fig, ax = plt.subplots()
+        x = np.array(list(map(lambda m: m[measure], times)))
+        y = np.array(list(map(lambda m: m[-1], times)))
+
+        # Calculate the best-fit line
+        # slope, intercept = np.polyfit(x, y, 1)
+        # trendline = slope * x + intercept
+
+        # Create the plot
+        ax.plot(x, y, marker='o', color='red', label='detailed route/seconds')  # Plot the data points
+        # ax.plot(x, trendline, '-', label='Trendline')  # Plot the trendline
+
+        # Customize the plot
+        ax.set_xlabel(measure_ids[measure])
+        ax.set_ylabel("Running time / seconds")
+        #ax.set_yscale('log')
+        ax.set_title("Detailed routing time")
+        ax.legend()
+
+        # Display the plot
+        plt.show()
+
+
+
+if __name__ == '__main__':
+    run()