PR summary

Towards #5665

Currently we are calling _update_ticks() 3 times per axis every draw call, and _get_ticklabel_bboxes 2 times per axis. These calls take up about 35% of total draw time for an empty plot.

We can eliminate 25% of total draw time by caching the results of these calculations every draw cycle. This introduces some state, but I think it's decently well guarded and the performance boost is definitely worth it.

Before & After
The circled red areas show the before & after runtime of axis._update_label_position within Axis.draw(). It runs after the cache values have been set by other functions, so we nearly completely eliminate its runtime.

Before:

After:

Profiling script:

import time
import matplotlib.pyplot as plt

fig, ax = plt.subplots()

print("Timing...")
start_time = time.perf_counter()
for i in range(100):
    fig.canvas.draw()
end_time = time.perf_counter()

plt.close()
print(f"Time taken: {end_time - start_time:.4f} seconds")

PR checklist

• [n/a] "closes #0000" is in the body of the PR description to link the related issue
• new and changed code is tested
• [n/a] Plotting related features are demonstrated in an example
• [n/a] New Features and API Changes are noted with a directive and release note
• [n/a] Documentation complies with general and docstring guidelines