llama 3.3 70b q4_k_m - perplexity on code with async/await vs callbacks at same token count

1. tested on 40 node.js files, half async/await half callbacks, normalized to 22k tokens each 2. async/await degrades around 20.1k, callbacks hold to 22.8k 3. perplexity delta is 2.4 at the 21k mark honestly think async/await creates more implicit control flow tokens that eat attention budget faster than explicit callback chains