Merge pull request #57 from johnramsden/format-plots

johnramsden · web-flow · commit 6f67c27106dc · 2026-01-11T01:22:59.000-08:00
Format plots
diff --git a/eval/boxplot_wt.py b/eval/boxplot_wt.py
@@ -9,10 +9,11 @@
 import matplotlib.pyplot as plt
 from matplotlib import patches as mpatches
 from matplotlib import rcParams
+from matplotlib import ticker
 import data_cache
 
-# Increase all font sizes by 8 points from their defaults
-rcParams.update({key: rcParams[key] + 8 for key in rcParams if "size" in key and isinstance(rcParams[key], (int, float))})
+# Increase all font sizes by 16 points from their defaults
+rcParams.update({key: rcParams[key] + 16 for key in rcParams if "size" in key and isinstance(rcParams[key], (int, float))})
 
 # ============================================================================
 # CONFIGURATION SECTION
@@ -297,16 +298,22 @@ def generate_boxplot(block_dir, zns_dir, output_file, sample_size=None, show_out
 
     # Create boxplot (identical to distribution_comparison_boxplots.py)
     if current_data:
+        # Adjust box width based on number of boxes so they're consistent across subplots
+        # Base width is 0.8 for 4 boxes, scale proportionally for fewer boxes
+        num_boxes = len(current_data)
+        box_width = 0.8 * (num_boxes / 4) if num_boxes > 0 else 0.8
+
         bp = ax.boxplot(current_data,
                         showfliers=show_outliers,
-                        widths=0.8,
+                        widths=box_width,
                         medianprops=dict(linewidth=2, color='black'),
                         patch_artist=True)
 
         # Apply colors and hatches
         for i, (box, color, hatch) in enumerate(zip(bp['boxes'], colors, hatches)):
             box.set_facecolor(color)
             box.set_hatch(hatch)
+            box.set_hatch_linewidth(3.0)
             box.set_alpha(0.7)
 
         # Set x-axis labels (empty for cleaner look)
@@ -319,6 +326,9 @@ def generate_boxplot(block_dir, zns_dir, output_file, sample_size=None, show_out
         # Set y-axis label
         ax.set_ylabel(metric_label, fontsize=22, weight='bold')
 
+        # Use scalar formatter without scientific notation
+        ax.yaxis.set_major_formatter(ticker.ScalarFormatter(useOffset=False, useMathText=False))
+
         # Rotate y-axis labels
         for label in ax.get_yticklabels():
             label.set_rotation(45)
diff --git a/eval/distribution_comparison_boxplots.py b/eval/distribution_comparison_boxplots.py
@@ -12,12 +12,14 @@
 from matplotlib import patches as mpatches
 from matplotlib.patches import Rectangle
 from matplotlib import rcParams
+from matplotlib import ticker
+from matplotlib.gridspec import GridSpec
 import numpy as np
 from datetime import datetime, timedelta
 import data_cache
 
-# Increase all font sizes by 8 points from their defaults
-rcParams.update({key: rcParams[key] + 8 for key in rcParams if "size" in key and isinstance(rcParams[key], (int, float))})
+# Increase all font sizes by 16 points from their defaults
+rcParams.update({key: rcParams[key] + 16 for key in rcParams if "size" in key and isinstance(rcParams[key], (int, float))})
 
 # ============================================================================
 # CONFIGURATION SECTION
@@ -247,14 +249,16 @@ def generate_distribution_comparison(block_dir, zns_dir, distribution, metric, o
 
     print(f"Found {len(block_runs)} block runs and {len(zns_runs)} ZNS runs")
 
-    # Create figure with 6 subplots (1 row x 6 columns)
+    # Create figure with 6 subplots (1 row x 6 columns) using GridSpec for custom widths
     # Subplots are 2/5 original height, but all spacing preserved
     num_subplots = len(RATIOS) * len(CHUNK_SIZES)  # 2 ratios * 3 chunk sizes = 6
-    fig, axes = plt.subplots(1, num_subplots, figsize=(5 * num_subplots, 5.38))
 
-    # Ensure axes is always a list
-    if num_subplots == 1:
-        axes = [axes]
+    # Width ratios: 1077MiB subplots (indices 2 and 5) are half as wide since they have 2 boxes instead of 4
+    width_ratios = [1, 1, 0.5, 1, 1, 0.5]
+
+    fig = plt.figure(figsize=(5 * num_subplots, 5.38))
+    gs = GridSpec(1, num_subplots, figure=fig, width_ratios=width_ratios)
+    axes = [fig.add_subplot(gs[0, i]) for i in range(num_subplots)]
 
     # First pass: collect all data to find global maximum for y-axis
     all_subplot_data = []
@@ -391,24 +395,37 @@ def generate_distribution_comparison(block_dir, zns_dir, distribution, metric, o
 
         # Create boxplot for this subplot
         if current_data:
+                # Adjust box width based on number of boxes so they're consistent across subplots
+                # Base width is 0.8 for 4 boxes, scale proportionally for fewer boxes
+                # Exception: 1077MiB boxes are made thicker since they have fewer boxes
+                num_boxes = len(current_data)
+                if chunk_size == 1129316352:  # 1077MiB
+                    box_width = 0.8  # Full width like regular boxplots
+                else:
+                    box_width = 0.8 * (num_boxes / 4) if num_boxes > 0 else 0.8
+
                 bp = axes[idx].boxplot(current_data,
                                       showfliers=show_outliers,
-                                      widths=0.8,
+                                      widths=box_width,
                                       medianprops=dict(linewidth=2, color='black'),
                                       patch_artist=True)
 
                 # Apply colors and hatches
                 for i, (box, color, hatch) in enumerate(zip(bp['boxes'], colors, hatches)):
                     box.set_facecolor(color)
                     box.set_hatch(hatch)
+                    box.set_hatch_linewidth(3.0)
                     box.set_alpha(0.7)
 
                 # Set x-axis labels (empty for cleaner look, or could add device labels)
                 axes[idx].set_xticks(range(1, len(labels) + 1))
                 axes[idx].set_xticklabels([], rotation=45, fontsize=10)
 
                 # Add chunk size label below subplot
-                axes[idx].set_xlabel(CHUNK_SIZE_LABELS[chunk_size], fontsize=16, weight='bold')
+                axes[idx].set_xlabel(CHUNK_SIZE_LABELS[chunk_size], fontsize=28, weight='bold')
+
+                # Use scalar formatter without scientific notation
+                axes[idx].yaxis.set_major_formatter(ticker.ScalarFormatter(useOffset=False, useMathText=False))
 
                 # Rotate y-axis labels
                 for label in axes[idx].get_yticklabels():
@@ -425,10 +442,10 @@ def generate_distribution_comparison(block_dir, zns_dir, distribution, metric, o
         idx += 1
 
     # Add y-axis label on the far left
-    fig.text(0.02, 0.5, metric_label, va='center', rotation='vertical', fontsize=22, weight='bold')
+    fig.text(-0.005, 0.5, metric_label, va='center', rotation='vertical', fontsize=22, weight='bold')
 
     # Adjust layout (do these BEFORE computing positions) - subplots at 2/5 height with proportional spacing
-    plt.subplots_adjust(wspace=0.05, hspace=0.0)
+    plt.subplots_adjust(wspace=0.2, hspace=0.0)
     plt.tight_layout(pad=0.0)
     plt.subplots_adjust(top=0.851, bottom=0.279, left=0.05)
 
@@ -452,8 +469,8 @@ def generate_distribution_comparison(block_dir, zns_dir, distribution, metric, o
     g2_width = g2_right - g2_left
 
     # Vertical placement of the grey boxes in figure coords
-    box_y = 0.93
-    box_h = 0.06
+    box_y = 0.85
+    box_h = 0.10
 
     # Grey box for Ratio 1:2
     fig.add_artist(
@@ -492,7 +509,7 @@ def generate_distribution_comparison(block_dir, zns_dir, distribution, metric, o
         "Ratio: 1:2",
         ha='center',
         va='center',
-        fontsize=20,
+        fontsize=26,
         weight='bold',
         zorder=2,
     )
@@ -502,7 +519,7 @@ def generate_distribution_comparison(block_dir, zns_dir, distribution, metric, o
         "Ratio: 1:10",
         ha='center',
         va='center',
-        fontsize=20,
+        fontsize=26,
         weight='bold',
         zorder=2,
     )
diff --git a/eval/distribution_comparison_ecdfs.py b/eval/distribution_comparison_ecdfs.py
@@ -18,8 +18,8 @@
 from datetime import datetime, timedelta
 import data_cache
 
-# Increase all font sizes by 8 points from their defaults
-rcParams.update({key: rcParams[key] + 8 for key in rcParams if "size" in key and isinstance(rcParams[key], (int, float))})
+# Increase all font sizes by 16 points from their defaults
+rcParams.update({key: rcParams[key] + 16 for key in rcParams if "size" in key and isinstance(rcParams[key], (int, float))})
 
 # ============================================================================
 # CONFIGURATION SECTION
@@ -426,8 +426,10 @@ def generate_distribution_comparison(block_dir, zns_dir, distribution, output_fi
                         print(f"Warning: No run found for {device} {eviction_type} chunk={chunk_size} dist={distribution} ratio={ratio}")
 
             # Configure subplot
-            current_ax.set_xlabel(CHUNK_SIZE_LABELS[chunk_size], fontsize=16, weight='bold')
-            current_ax.set_ylabel('Cumulative Probability (%)', fontsize=18)
+            current_ax.set_xlabel(CHUNK_SIZE_LABELS[chunk_size], fontsize=28, weight='bold')
+            if idx == 0:
+                ylabel = current_ax.set_ylabel('Cumulative Probability (%)', fontsize=25)
+                ylabel.set_position((-0.1, 0.3))
             current_ax.set_ylim(0, 100)
 
             # Configure x-axis scale
@@ -457,7 +459,6 @@ def exp_formatter(x, pos):
 
                 current_ax.xaxis.set_major_formatter(FuncFormatter(exp_formatter))
             else:
-                current_ax.set_xlim(left=0)
                 # Use MaxNLocator to ensure nice, evenly-spaced tick intervals
                 current_ax.xaxis.set_major_locator(MaxNLocator(nbins=6, integer=False, prune=None))
 
@@ -496,13 +497,13 @@ def exp_formatter(x, pos):
         Line2D([0], [0], color='#f781bf', linestyle='--', linewidth=LINE_WIDTH,
                label='Block (Chunk LRU)', alpha=0.8),
     ]
-    fig.legend(ncols=4, handles=legend_lines, bbox_to_anchor=(subplot_center, 0.02),
+    fig.legend(ncols=4, handles=legend_lines, bbox_to_anchor=(subplot_center, -0.09),
                loc='center', fontsize="large", columnspacing=2.0, frameon=False)
 
     # Add a background box for the x-axis label to make it stand out
-    label_y = 0.08
-    label_width = 0.12
-    label_height = 0.04
+    label_y = 0.01
+    label_width = 0.18
+    label_height = 0.08
     fig.add_artist(
         Rectangle(
             (subplot_center - label_width/2, label_y - label_height/2),
@@ -518,7 +519,7 @@ def exp_formatter(x, pos):
 
     # Add x-axis label at the bottom, centered over the subplots
     fig.text(subplot_center, label_y, 'Latency (ms)', ha='center', va='center',
-             fontsize=18, weight='bold', zorder=11)
+             fontsize=30, weight='bold', zorder=11)
 
     # First 3 subplots -> Ratio 1:2, next 3 -> Ratio 1:10
     group1 = axes_bboxes[0:3]
@@ -534,8 +535,8 @@ def exp_formatter(x, pos):
     g2_width = g2_right - g2_left
 
     # Vertical placement of the grey boxes in figure coords
-    box_y = 0.93
-    box_h = 0.06
+    box_y = 0.85
+    box_h = 0.10
 
     # Grey box for Ratio 1:2
     fig.add_artist(
@@ -574,7 +575,7 @@ def exp_formatter(x, pos):
         "Ratio: 1:2",
         ha='center',
         va='center',
-        fontsize=20,
+        fontsize=26,
         weight='bold',
         zorder=2,
     )
@@ -584,7 +585,7 @@ def exp_formatter(x, pos):
         "Ratio: 1:10",
         ha='center',
         va='center',
-        fontsize=20,
+        fontsize=26,
         weight='bold',
         zorder=2,
     )
diff --git a/eval/ecdf_wt.py b/eval/ecdf_wt.py
@@ -14,8 +14,8 @@
 import numpy as np
 import data_cache
 
-# Increase all font sizes by 8 points from their defaults
-rcParams.update({key: rcParams[key] + 8 for key in rcParams if "size" in key and isinstance(rcParams[key], (int, float))})
+# Increase all font sizes by 16 points from their defaults
+rcParams.update({key: rcParams[key] + 16 for key in rcParams if "size" in key and isinstance(rcParams[key], (int, float))})
 
 # ============================================================================
 # CONFIGURATION SECTION
@@ -402,8 +402,8 @@ def generate_ecdf(block_dir, zns_dir, output_file, metric_type="get_total", samp
                 print(f"Warning: No run found for {device} {eviction_type}")
 
     # Configure subplot
-    # ax.set_xlabel("64KiB", fontsize=16, weight='bold')
-    ax.set_ylabel('Cumulative Probability (%)', fontsize=18)
+    # ax.set_xlabel("64KiB", fontsize=28, weight='bold')
+    ax.set_ylabel('Cumulative Probability (%)', fontsize=25)
     ax.set_ylim(0, 100)
 
     # Configure x-axis scale
@@ -433,7 +433,6 @@ def exp_formatter(x, pos):
 
         ax.xaxis.set_major_formatter(FuncFormatter(exp_formatter))
     else:
-        ax.set_xlim(left=0)
         # Use MaxNLocator to ensure nice, evenly-spaced tick intervals
         ax.xaxis.set_major_locator(MaxNLocator(nbins=6, integer=False, prune=None))
 
@@ -465,14 +464,26 @@ def exp_formatter(x, pos):
         Line2D([0], [0], color='#f781bf', linestyle='--', linewidth=LINE_WIDTH,
                label='Block (Chunk LRU)', alpha=0.8),
     ]
-    fig.legend(ncols=4, handles=legend_lines, bbox_to_anchor=(subplot_center, 0.02),
+    fig.legend(ncols=4, handles=legend_lines, bbox_to_anchor=(subplot_center, -0.09),
                loc='center', fontsize="large", columnspacing=2.0, frameon=False)
 
     # Add a background box for the x-axis label to make it stand out
-    label_y = 0.08
+    label_y = 0.01
+    label_width = 0.18
+    label_height = 0.08
+    fig.add_artist(
+        Rectangle(
+            (subplot_center - label_width/2, label_y - label_height/2),
+            label_width,
+            label_height,
+            facecolor='white',
+            edgecolor='none',
+            zorder=10,
+        )
+    )
     # Add x-axis label at the bottom, centered over the subplot
     fig.text(subplot_center, label_y, 'Latency (ms)', ha='center', va='center',
-             fontsize=18, weight='bold', zorder=11)
+             fontsize=30, weight='bold', zorder=11)
 
     # Save figure
     plt.savefig(output_file, bbox_inches='tight', dpi=100)
