Mamba is an emerging, complex workload with various short-range and long-range dependencies, nonlinearities, and elementwise computations that are unable to run at near-peak speeds on modern hardware. Specifically, Mamba's complex dependency graph makes fusion across its full operator cascade difficult, leaving substantial inter-operator memory traffic on the table. To address these challenges, we propose Mambalaya, a novel reconfigurable accelerator that leverages fusion to overcome the limitations of Mamba. We use the recently proposed cascade-of-Einsums abstraction to characterize Mamba's full computational structure, then apply the extended Einsum framework to systematically explore inter-Einsum fusion opportunities. This principled approach yields a series of fusion mappings that reduce off-chip inter-Einsum traffic. These mappings are supported by the underlying Mambalaya architecture. Mambalaya achieves a layer performance speedup of 4.9× for prefill and 1.9× for generation over MARCA. In prefill-dominated scenarios, it achieves up to 1.5× over a recent fine-grained, memory-aware fusion accelerator for Mamba.
@article{arxiv.2604.03829,
title = {Mambalaya: Einsum-Based Fusion Optimizations on State-Space Models},
author = {Toluwanimi O. Odemuyiwa and John D. Owens and Joel S. Emer and Michael Pellauer},
journal= {arXiv preprint arXiv:2604.03829},
year = {2026}
}