One maintainer of Claude Code made a humorous comparison between their work and a major video game, which sparked a deeper look into the performance of Claude Code relative to classic games like Super Mario 64. The author acknowledges that comparing a modern game engine to a 1996 title isn't straightforward, but they argue that both operate under constraints that make the comparison interesting. Super Mario 64 ran on an N64 with a CPU clocked at 93.75 MHz and only 4 MB of RAM, yet managed to deliver engaging graphics and gameplay. The author questions whether Claude Code is doing more work to generate its frames compared to Super Mario 64, hinting at potential inefficiencies in Claude Code's operation. Using performance metrics, the analysis reveals that Claude Code utilizes the CPU about 26% of the time and experiences significant delays due to a high volume of system calls, particularly related to futexes. Specifically, it spent 70% of its time waiting on these locks, which suggests that the code isn't efficiently managing its resources. The author contrasts this with another application, amp, which demonstrates more efficient syscall usage. While Claude Code appears to be bogged down by spinning threads and frequent yielding, the main thread in amp correctly waits for I/O without unnecessary context switching. This raises questions about the architecture of Claude Code and whether its issues stem from the design of the underlying technology it relies on, such as Node.js or Bun.