yan@yandesbiens

yan@yandesbiens:~$ cat citations.bib

Citations

Every output here is meant to be a permanent, citable scientific asset. If a result helped your work, please cite it — that's how a small lab compounds. Outputs are reproducible or marked speculative; none are opinions.

Author: Yan Desbiens · Affiliation: Éthiqueia Québec inc., Québec, Canada

@misc{ethiqueia,
  title = {Éthiqueia — independent AI research},
  author = {Yan Desbiens},
  institution = {Éthiqueia Québec inc.},
  howpublished = {\url{https://yandesbiens.com}},
  note = {Reproducible local-first AI research}
}

// citable outputs (1)

Unified Fractal Memory: running a routed model larger than VRAM on a single consumer GPU Memory Systems

Yan Desbiens · 2026-06-27 · benchmark · v0.1.1

We benchmark Unified Fractal Memory (UFM), a residency manager that treats GPU VRAM and CPU RAM as one elastic pool, on a routed Mixture-of-Experts whose expert bank exceeds device memory. On a 23.5 GB RTX 4090, the standard all-on-GPU placement OOMs at a 24 GB bank; UFM runs the same model holding VRAM at 19.6 GB. When the active working set fits the VRAM budget, UFM matches baseline throughput within ~1% and is ~240x faster than naive per-call CPU offloading; when every expert is touched every step (no locality), UFM is transfer-bound and offers no speedup over naive streaming. UFM is a bet on routing locality, not unbounded memory.

Code + one-command repro · Benchmark report · Figures

// cite this

reproducible DOI pending

Desbiens, Y. (2026). Unified Fractal Memory: running a routed model larger than VRAM on a single consumer GPU (v0.1.1). Éthiqueia Québec inc.. https://yandesbiens.com/blog/ufm-benchmark/

@misc{desbiens2026ufm,
  title = {Unified Fractal Memory: running a routed model larger than VRAM on a single consumer GPU},
  author = {Yan Desbiens},
  year = {2026},
  howpublished = {\url{https://yandesbiens.com/blog/ufm-benchmark/}},
  institution = {Éthiqueia Québec inc.},
  version = {v0.1.1},
  note = {Reproducible benchmark, Éthiqueia Québec inc.},
  url = {https://yandesbiens.com/blog/ufm-benchmark/},
}

Full citation registry on the citations page. · repository ↗

DOIs are minted per release via Zenodo as outputs stabilize; until then, cite the canonical URL + version. Machine-readable CITATION.cff ships in each repository.