Challenge #2: Energy-efficient matmul
Minimum-energy 16x16 matrix multiply, expressed as an intermediate representation on a 2D grid.
Minimize the data-movement energy of matrix multiplication, expressed as an intermediate representation (explicit load and store ops) on Bill Dally's 2D grid. The focus size is 16x16.
Where it lives
sutro-problems, the matmul/ directory. The scorer (matmul.py) is locked; submissions must not modify it.
Cost metric
The Dally 2D-grid cost model.
State
Active hill-climbing. The 16x16 record stands at 67,821 (mid-May), down from 68,452 (Sung Jae Bae, May 5) and 68,392 (Cosmin Negruseri, May 13). Cosmin ran a Codex-driven auto-research loop that produced a steady stream of small PRs. Lower bounds are an open hard problem: an agent-generated bound was wrong on inspection, and AlphaTensor could not bound 4x4 either.
Who is active
Cosmin Negruseri, Sung Jae Bae, Anastasiia Zhiboedova. See the May recap.