# Asymptotically faster matrix-multiplication algorithms

**URL:** <https://discourse.julialang.org/t/asymptotically-faster-matrix-multiplication-algorithms/96710>\
**Category:** Offtopic\
**Tags:** linearalgebra\
**Created:** [March 27, 2023, 2:57pm UTC](https://discourse.julialang.org/t/asymptotically-faster-matrix-multiplication-algorithms/96710 "2023-03-27T14:57:11Z")\
**Posts on this page:** 1\
**Showing post:** 7

<div class="post-metadata">

**Author:** ![stevengj](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/stevengj/32/71_2.png) [@stevengj](https://discourse.julialang.org/u/stevengj)\
**Post date:** [March 28, 2023, 12:23pm UTC](https://discourse.julialang.org/t/asymptotically-faster-matrix-multiplication-algorithms/96710/7 "2023-03-28T12:23:27Z")

</div>

> [@Oscar\_Smith](#):
>
> the strassen reloaded papers put the cutoff closer to 1000 and they were benchmarking vs worse regular implementions. I would be interested in someone trying strassen on Octavian

To have more than marginal (\> 10%) speedup they need a few thousand, and to have \> 20% speedup they need sizes \gtrsim 10^4. This seems pretty consistent with other experiments I’ve seen over the years, e.g. with [ATLAS BLAS in 2006](https://link.springer.com/chapter/10.1007/978-3-540-77704-5_12).

See also this discussion: [Matrix multiply breakthrough, AlphaTensor (could also do for other algorithms): "AlphaTensor discovered algorithms that are more efficient than the state of the art for many matrix sizes." - #7 by stevengj](https://discourse.julialang.org/t/matrix-multiply-breakthrough-alphatensor-could-also-do-for-other-algorithms-alphatensor-discovered-algorithms-that-are-more-efficient-than-the-state-of-the-art-for-many-matrix-sizes/88455/7)

---

_[View the full topic](https://discourse.julialang.org/t/asymptotically-faster-matrix-multiplication-algorithms/96710)._