Evaluating the 2D Vector and Raster Graphics Processing Proficiency of CMU CS Academy's IDE

Introduction

This sets out to evaluate the computational capabilities of CMU Computer Science (CS) Academy's Integrated Development Environment (IDE) in processing 2D Vector and Raster Graphics.

Methodology

The evaluation framework centers on the continuous creation and rendering of 2D objects while closely observing the total count of objects effectively processed by the CMU CS Academy's IDE. The selection of test objects comprised vector shapes, along with raster text, that I found to fetch fonts from fonts.google.com based on network requests and debugging observations.

Evaluation Method

Performance testing was carried out using a system equipped with a 12th generation Intel Core i7 processor, 16GB RAM, and an Nvidia T1000 with 4GB of GDDR6, running on Windows 10. The testing spans were deployed on Firefox and Google Chrome.

Firefox demonstrated performance of an average count of 5,423 shapes before crash over ten runs, with each stress test lasting near five minutes before the page became unresponsive, prompting either the page to crash, the browser to crash or the Windows desktop to crash.

The variability of the data on Chromium-based browsers shows that this test may not be optimal for Chromium-based browsers. For a performance log recorded via Firefox's built-in profiler, click here.

Some assumptions on CMU CS Academy's Graphics Rendering

CMU CS Academy leverages Brython, which is an implementation of Python 3 for the web, to execute code in a Python 3 environment and uses Ace Editor as the code editor. While Brython supports outputting to a display, specific JavaScript (JS) frameworks are required. However, due to CMU's use of Ace, it seems that none of these frameworks would be compatible. To solve this problem, they wrote a graphics library that renders directly to the webpage, but this library does not yet seem to be equipped to manage multiple shapes simultaneously. To handle the creation of shapes or text labels, an instance of their JS graphics library is spawned on-demand, creating massive inefficiency in memory and CPU usage.

Code available here