Back to the watchlist
Computer science / Milestone

Faster Matrix Multiplication

How low can the exponent for multiplying two square matrices go?

The source trail 10

Findings, with context
DiscussionFound Sep 20, 2026

A source discussing Matrix multiplication

scalable-ai.eecs.berkeley.edu contains the excerpt below in a discussion related to Matrix multiplication. Read the source for the surrounding context.

A large language model proposing edits to a piece of code, an evaluator scoring them, the best edits surviving and proliferating
This excerpt is evidence of what the linked page says, not verification of its claims. Publication date has not been established.
Source pageIndirect account · Found Sep 20
scalable-ai.eecs.berkeley.edu
DiscussionAug 19, 2026

A source discussing Matrix multiplication

ainews.imrenagi.com contains the excerpt below in a discussion related to Matrix multiplication. Read the source for the surrounding context.

AlphaEvolve extended an already successful human-designed and ML-enabled optimization system
This excerpt is evidence of what the linked page says, not verification of its claims. Source reports an update on 2026-08-19; this is not necessarily the original posting date.
Source pageIndirect account · Found Sep 20
ainews.imrenagi.com
DiscussionSep 19, 2026

A source discussing Matrix multiplication

traes writes the excerpt below in a discussion related to Matrix multiplication. Read the source for the surrounding context.

I guess they could make some minor improvements to matrix multiplication algorithms or whatever
This excerpt is evidence of what the linked page says, not verification of its claims.
traesOriginal source · Found Sep 20
Hacker News
DiscussionSep 12, 2026

A source discussing Matrix multiplication

eru writes the excerpt below in a discussion related to Matrix multiplication. Read the source for the surrounding context.

the new faster matrix multiplication algorithm was found and proven correct by a machine or a meatbag.
This excerpt is evidence of what the linked page says, not verification of its claims.
eruOriginal source · Found Sep 20
Hacker News
DiscussionFound Sep 20, 2026

A source discussing Matrix multiplication

sungsoo.github.io contains the excerpt below in a discussion related to Matrix multiplication. Read the source for the surrounding context.

그 결과 행렬곱 지수의 엄밀한 상한은 \(\omega<2.371339\)에서 \(\omega<2.371177\)로 내려간다.
This excerpt is evidence of what the linked page says, not verification of its claims. Publication date has not been established.
Source pageIndirect account · Found Sep 20
sungsoo.github.io
DiscussionAug 18, 2026

A source discussing Matrix multiplication

muhammad-ahmed.com contains the excerpt below in a discussion related to Matrix multiplication. Read the source for the surrounding context.

AlphaEvolve did not find an algorithm and did not produce a proof.
This excerpt is evidence of what the linked page says, not verification of its claims. Source reports an update on 2026-08-18; this is not necessarily the original posting date.
Source pageIndirect account · Found Sep 20
muhammad-ahmed.com
DiscussionSep 12, 2026

A source discussing Matrix multiplication

Dylan16807 writes the excerpt below in a discussion related to Matrix multiplication. Read the source for the surrounding context.

We're way into diminishing returns in matrix multiplication, and being clever about ALUs and cache layout is likely to dwarf any asymptotic improvements
This excerpt is evidence of what the linked page says, not verification of its claims.
Dylan16807Original source · Found Sep 19
Hacker News
DiscussionSep 12, 2026

A source discussing Matrix multiplication

Dylan16807 writes the excerpt below in a discussion related to Matrix multiplication. Read the source for the surrounding context.

If you tell someone they can't use a matrix multiplication with better asymptotic performance than n^2.3755 from 1990 they're going to shrug and not care.
This excerpt is evidence of what the linked page says, not verification of its claims.
Dylan16807Original source · Found Sep 19
Hacker News
DiscussionMay 14, 2025

Background: AlphaEvolve reports new matrix-multiplication algorithms

deepmind.google contains the excerpt below in a discussion related to Matrix multiplication. Read the source for the surrounding context.

AlphaEvolve designed many components of a novel gradient-based optimization procedure that discovered multiple new algorithms for matrix multiplication
Historical context published May 14, 2025, not a new September 2026 result. The source describes algorithms found by AlphaEvolve; it does not establish the optimal matrix-multiplication exponent.
Source pageOriginal source · Found Sep 19
Google DeepMind
PaperAug 17, 2026

AlphaEvolve helps improve a matrix multiplication bound

A collaboration combines optimization improvements with AlphaEvolve to report a lower upper bound for the matrix multiplication exponent.

Older background. A theoretical exponent improvement is not a measured speedup for every practical matrix multiplication.
Dupont and collaboratorsOriginal source · Found Sep 19
arXiv