Molecular Dynamics on Apple M4
A developer implemented 15 molecular dynamics kernels on the Apple M4 chip to explore performance across different hardware units, including CPU, GPU, and Neural Engine. The project achieved up to 810 GFLOPS on the Metal GPU and demonstrated significant speedups using optimization techniques like cell lists and tiling. The rapid iteration allowed real-time exploration of hardware-specific optimizations for N-body simulations.
- ▪The M4 Metal GPU achieved a peak of 810 GFLOPS, reaching ~19% of its theoretical 4.26 TFLOPS.
- ▪Cell list optimizations reduced complexity from O(N²) to O(N), making Metal+CL 45× faster than all-pairs at 70K particles.
- ▪An ANE direct kernel used reverse-engineered private APIs to run Lennard-Jones forces on the Neural Engine with FP16 precision.
- ▪OpenMP on 4 P-cores outperformed the Metal GPU in cell list performance above 32,000 particles.
- ▪Double-precision NEON (f64) was 2.2× slower than single-precision (f32), with negligible accuracy gains for simulation stability.
Hacker News (Newest) files mainly under programming. We currently carry 5,306 of its stories.
Story provenance
Source · retrieval · rights · ranking — open for full record
inspect →
Story provenance
Attribution is not the same as permission. This drawer separates discovery metadata, excerpts, WeSearch-generated summaries, reuse status, and whether the publisher receives the visit. Nothing here claims a legal grant the publisher has not made.
Record
| Original publisher | GitHub |
| Canonical URL | https://github.com/vyasgiridhar/moleqular |
| Publication time | Sun, 17 May 2026 05:03:58 +0000 |
| Retrieval time | 2026-05-17T05:33:58.675Z |
| Last seen | 2026-05-17T05:33:58.675Z |
| Headline source | Publisher (no WeSearch rewrite) |
| Excerpt source | publisher body |
| Excerpt method | First ~120 words (~800 chars) of extracted publisher body, fair-use limited. |
| Summary | WeSearch · cerebras-chat (WeSearch summarizer) |
| Summary source text | contentText |
| Citation coverage | Summary is a WeSearch-generated derivative; primary citation is the original publisher URL. |
| Cluster | ciQO6w7JAo9X |
| Cluster logic | Grouped by semantic title/content similarity across sources within a rolling window. Same-publisher template collisions are excluded from coverage comparison. |
| Ranking reason | Story pages are not engagement-ranked. Hub feeds use recency, with optional source-diversified chronological ordering (cap consecutive stories per source). No personalized ranking. |
| Publisher visit | Yes — open original |
| Substitutes article? | No — link-out required for full text |
Rights status (four layers)
WeSearch handling by dimension
| Indexing | May the item be indexed (stored, ranked, made findable)? | Allowed |
| Snippet | May a short excerpt of the publisher's text be shown? | Allowed |
| AI summary | May WeSearch generate its own short summary of the article? | Limited |
| Retrieval / RAG | May the content be exposed for third-party retrieval-augmented generation? | Not asserted |
| Model training | May the content be used to train AI models? | Not asserted |
| Commercial reuse | May the content be reused commercially? | Not permitted |
Basis: Derived from the published RSS/Atom feed. Contact: [email protected]. Reviewed: 2026-07-24.
Opening excerpt (first ~120 words) tap to expand
moleqular Molecular dynamics on Apple M4 — pushing every compute path to its limits. LJ (Lennard-Jones) N-body simulation with 15 force kernels targeting different hardware units on Apple Silicon. Same physics, same particles, wildly different performance characteristics. Built in 2 days. 15 kernels across 5 architectures (M4 NEON, Metal GPU, M4 Neural Engine, NVIDIA CUDA, GCP Axion SVE2). A real-time Metal particle renderer. A quantized BVH. A GROMACS-style NBNXM cluster pair kernel. A direct ANE kernel bypassing CoreML via reverse-engineered private APIs. Cross-compiled and benchmarked on cloud GPUs.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at GitHub.