From fdddc26351a583a40745517a1156bc05e318fdda Mon Sep 17 00:00:00 2001
From: Saveliy Yusufov <sy2685@columbia.edu>
Date: Fri, 2 Feb 2024 19:16:46 -0500
Subject: [PATCH] Update benchmark plotting

---
 benches/benchmark_plots.py | 60 ++++++++++++++++++++++----------------
 1 file changed, 35 insertions(+), 25 deletions(-)

diff --git a/benches/benchmark_plots.py b/benches/benchmark_plots.py
index 68fcb8d..8d416b4 100644
--- a/benches/benchmark_plots.py
+++ b/benches/benchmark_plots.py
@@ -1,8 +1,6 @@
 """
 Plot benchmark results for FFTW3, RustFFT, and PhastFT
 """
-from collections import defaultdict
-
 import matplotlib.pyplot as plt
 import numpy as np
 import pandas as pd
@@ -20,13 +18,13 @@
 }
 
 
-def read_file(filepath: str) -> list[int]:
+def read_file(filepath: str) -> list[float]:
     y = []
 
     with open(filepath) as f:
         for line in f:
             line = line.strip()
-            y.append(int(line))
+            y.append(float(line))
 
     return y
 
@@ -35,31 +33,32 @@ def get_figure_of_interest(vals: list[int]) -> float:
     return np.median(vals)
 
 
-def build_and_clean_data(root_benchmark_dir: str, *names) -> defaultdict[str, list]:
-    libs = ("rustfft", "phastft")
-    n_range = range(12, 30)
-
-    data = defaultdict(list)
+def build_and_clean_data(root_benchmark_dir: str, n_range: range, lib_name: str) -> list[float]:
+    data = []
 
-    for lib in libs:
-        for n in n_range:
-            y = read_file(f"{root_benchmark_dir}/{lib}/size_{n}")
-            y_k = get_figure_of_interest(y)
-            data[lib].append(y_k)
+    for n in n_range:
+        y = read_file(f"{root_benchmark_dir}/{lib_name}/size_{n}")
+        y_k = get_figure_of_interest(y)
+        data.append(y_k)
 
     return data
 
-def plot_lines(data: defaultdict[str, list]) -> None:
-    index = [bytes2human(2**n * (128 / 8)) for n in range(12, 30)]
+
+def plot_lines(data: dict[str, list], n_range: range) -> None:
+    index = [bytes2human(2**n * (128 / 8)) for n in n_range]
     plt.figure()
 
-    y0 = np.asarray(data["phastft"])
-    y1 = np.asarray(data["rustfft"])
-    y0 = y1/y0
+    y0 = np.asarray(data["fftw3"])
+    y1 = np.asarray(data["phastft"])
+    y2 = np.asarray(data["rustfft"])
+
+    y0 /= y2
+    y1 /= y2
 
     df = pd.DataFrame(
         {
-            "PhastFT": y0,
+            "FFTW3": y0,
+            "PhastFT": y1,
             "RustFFT": np.ones(len(index)),
         },
         index=index,
@@ -70,10 +69,11 @@ def plot_lines(data: defaultdict[str, list]) -> None:
     plt.xticks(fontsize=9, rotation=-45)
     plt.yticks(fontsize=9)
     plt.xlabel("size of input")
-    plt.ylabel("speedup (relative to RustFFT)")
+    plt.ylabel("time taken (relative to RustFFT)")
     plt.legend(loc='best')
     plt.tight_layout()
     plt.savefig("benchmarks_bar_plot.png", dpi=600)
+    # plt.show()
 
 
 # Source: https://stackoverflow.com/a/1094933
@@ -97,9 +97,19 @@ def bytes2human(n, format='%(value).1f %(symbol)s', symbols='customary'):
     return format % dict(symbol=symbols[0], value=n)
 
 
+def main():
+    lib_names = ("rustfft", "phastft", "fftw3")
+    n_range = range(12, 30)
+
+    all_data = {}
+
+    for lib in lib_names:
+        data = build_and_clean_data("benchmark-data.2024.02.02.16-45-50", n_range, lib)
+        all_data[lib] = data
+
+    assert len(all_data["rustfft"]) == len(all_data["fftw3"]) == len(all_data["phastft"])
+    plot_lines(all_data, n_range)
+
 
 if __name__ == "__main__":
-    # y = read_file("benchmark-data.2024.02.02.11-02-07/phastft/size_16")
-    data = build_and_clean_data("benchmark-data.2024.02.02.11-43-10")
-    # print(data)
-    plot_lines(data)
\ No newline at end of file
+    main()
\ No newline at end of file