@krzyzanowskim: Swift is swift, then I rewrite class to Objective-C and change is: 21.3 ms→95 µs at 1M units, 384 µs→6.4 µs at 67K ther…
Summary
The article highlights a performance benchmark where rewriting code from Swift to Objective-C drastically improved execution times, from milliseconds to microseconds for large datasets.
View Cached Full Text
Cached at: 09/24/26, 06:19 AM
Swift is swift, then I rewrite class to Objective-C and change is: 21.3 ms→95 µs at 1M units, 384 µs→6.4 µs at 67K
there is no comparison. rewrite 1:1.
Similar Articles
Swift at Apple: Migrating the TrueType Hinting Interpreter
Apple migrated the TrueType hinting interpreter from C to Swift, achieving memory safety and a 13% performance improvement. The source code has been open-sourced.
@no_stp_on_snek: anyone still talking about mlx-swift-lm? said i was taking the day off... cleaned the chicken coop, got a workout in, f…
The author describes implementing TurboQuant KV-cache compression into Apple's mlx-swift-lm, achieving 2.7x compression with quality on par with 8-bit, and 3-4x decode speed improvements via a fused Metal kernel.
Apple Internals: Swift in the Kernel
Apple has begun integrating Swift into the kernel via a new effort called KernelKit, with an Embedded Swift runtime appearing in macOS and iOS, marking a step toward memory-safe kernel extensions.
@no_stp_on_snek: busy evening in custom rust kernel and swift dispatch land. 3 of the 9 PRs are genuine new or changed metal kernels. th…
在自定义Rust内核和Swift调度逻辑的优化下，Qwen3.6-35b-A3B模型的预填速度在2k提示下从255 tok/s提升到1058 tok/s，实现了约4倍的加速，解码和困惑度未受影响。
Training an LLM in Swift, Part 1: Taking matrix mult from Gflop/s to Tflop/s
The author details the process of optimizing custom matrix multiplication kernels in Swift to train a Large Language Model on Apple Silicon, aiming to outperform C implementations by leveraging CPU, SIMD, AMX, and GPU capabilities.