auto-optimize speeds up Bonsai ~6x; homer2 motif tools made 10-100x faster
jmschreiber91 · x · 2026-10-30
- A follow-up in the same auto-optimize series: a PR speeds up Bonsai (single-cell data representation) by 4.2x-7.9x on nearly every input, with byte-identical output. The 500-cell PBMC input (4 MPI processes) drops from 80.22s to 13.41s (5.98x); four full 10x Genomics experiments see a geometric-mean 5.99x speedup; pbmc10kv3 (11,769 cells) at 16 processes falls from 3,526s to 526s.
- Main gains come from numba kernels for per-gene likelihood loops and freezing the garbage collector after setup.
- Also released homer2-ao: speedups to homer2's motif discovery and scanning algorithms making them 10-100x faster with a fraction of the memory, self-published because the original authors were unresponsive, with credit explicitly given to them.
More from Research
- SCOPE workshop proposed for ICLR 2027 asks what counts as an AI scientific discovery — mariyaivasileva · 2026-10-31
- New MolmoSpaces-CommonSense benchmark: no robot policy tops 30% on manipulation tasks — RanjayKrishna · 2026-10-31
- Mischief launches video model trained from lab-grown brain cells at 0.5¢ per second — evilsocket · 2026-10-31
- Stanford CS336 notes: Percy Liang on scaling recipes and hyperparameter transfer — stanfordnlp · 2026-10-31
- Composable LoRA directions for texture, geometry and lighting make 3D fine-tuning cheaper — burkov · 2026-10-31
- Moondream releases RefCOCO-M with pixel-accurate masks and cleaned prompts for grounding evals — victormustar · 2026-10-31