Joel S. Emer

Toluwanimi O. Odemuyiwa, Serban D. Porumbescu, Joel S. Emer et al.

In this work, we propose a unified abstraction for graph algorithms: the Extended General Einsums language, or EDGE. The EDGE language expresses graph algorithms in the language of tensor algebra, providing a rigorous, succinct, and expressive mathematical framework. EDGE leverages two ideas: (1) the well-known foundations provided by the graph-matrix duality, where a graph is simply a 2D tensor, and (2) the power and expressivity of Einsum notation in the tensor algebra world. In this work, we describe our design goals for EDGE and walk through the extensions we add to Einsums to support more complex operations common in graph algorithms. Additionally, we provide a few examples of how to express graph algorithms in our proposed notation. We hope that a single, mathematical notation for graph algorithms will (1) allow researchers to more easily compare different algorithms and different implementations of a graph algorithm; (2) enable developers to factor complexity by separating the concerns of what to compute (described with the extended Einsum notation) from the lower level details of how to compute; and (3) enable the discovery of different algorithmic variants of a problem through algebraic manipulations and transformations on a given EDGE expression.

Yannan Nellie Wu, Po-An Tsai, Angshuman Parashar et al.

In recent years, many accelerators have been proposed to efficiently process sparse tensor algebra applications (e.g., sparse neural networks). However, these proposals are single points in a large and diverse design space. The lack of systematic description and modeling support for these sparse tensor accelerators impedes hardware designers from efficient and effective design space exploration. This paper first presents a unified taxonomy to systematically describe the diverse sparse tensor accelerator design space. Based on the proposed taxonomy, it then introduces Sparseloop, the first fast, accurate, and flexible analytical modeling framework to enable early-stage evaluation and exploration of sparse tensor accelerators. Sparseloop comprehends a large set of architecture specifications, including various dataflows and sparse acceleration features (e.g., elimination of zero-based compute). Using these specifications, Sparseloop evaluates a design's processing speed and energy efficiency while accounting for data movement and compute incurred by the employed dataflow as well as the savings and overhead introduced by the sparse acceleration features using stochastic tensor density models. Across representative accelerators and workloads, Sparseloop achieves over 2000 times faster modeling speed than cycle-level simulations, maintains relative performance trends, and achieves 0.1% to 8% average error. With a case study, we demonstrate Sparseloop's ability to help reveal important insights for designing sparse tensor accelerators (e.g., it is important to co-design orthogonal design aspects).

Michael Gilbert, Tanner Andrulis, Vivienne Sze et al.

The energy and latency of an accelerator running a deep neural network (DNN) depend on how the computation and data movement are scheduled in the accelerator (i.e., mapping), and picking an optimal mapping is essential to achieve high-performance accelerators. However, it is challenging to find mappings that maximize accelerator performance. The space of mappings is large, and prior works cannot guarantee finding optimal mappings because they use heuristics or metaheuristics to narrow the search space. To address this challenge, we propose the Turbo-Charged Mapper (TCM), a fast mapper that finds optimal mappings. The key to our approach is that we define a new mapping concept called dataplacement, which, like the prior concept of dataflow, allows for clear analysis and comparison of mappings. Through it, we identify opportunities to prune redundant and suboptimal mappings, reducing search space by up to 32 orders of magnitude ($10^{37}\rightarrow10^5$). TCM leverages these insights to perform full mapspace searches, making it the first mapper that can find optimal mappings in feasible runtime. Compared to prior mappers, TCM improves accelerator energy-delay-product by $1.2-6.5\times$ while simultaneously reducing mapping search time by $1000\times$ (5 hours $\rightarrow$ 17 seconds).

Tanner Andrulis, Michael Gilbert, Vivienne Sze et al.

A low-latency and energy-efficient tensor algebra accelerator design must optimize how data movement and operations are scheduled (i.e., mapped) in the accelerator architecture. A key mapping optimization is fusion, meaning holding data on-chip between computation steps in the workload, which has been shown to reduce energy and latency by reducing expensive off-chip data movement. However, the optimal fusion choice depends on the workload and workload shape, and a mapper, which searches for the optimal mapping, can improve energy and latency significantly. However, prior mappers cannot find optimal mappings with fusion (i.e., fused mappings) in a feasible runtime because the number of fused mappings to search increases exponentially with the number of computation steps in the workload. In this paper, we introduce the Fast and Fusiest Mapper (FFM), a mapper to quickly find optimal mappings in a comprehensive fused mapspace for tensor algebra workloads. FFM shrinks the search space by pruning subsets of mappings (i.e., partial mappings) that are shown to never be a part of optimal mappings, quickly eliminating all suboptimal mappings containing those partial mappings. Then FFM joins partial mappings to construct optimal fused mappings. Using FFM, we demonstrate an energy-delay-product (EDP) reduction by up to $1.8\times$ compared to TransFusion, a state-of-the-art accelerator with hand-optimized fusion. Moreover, we show that FFM finds mappings orders of magnitude faster ($>10,000\times$) than prior automated mappers TileFlow and SET, and given the same runtime, reduces EDP by $>2\times$.

Toluwanimi O. Odemuyiwa, John D. Owens, Joel S. Emer et al.

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$\times$ for prefill and 1.9$\times$ for generation over MARCA. In prefill-dominated scenarios, it achieves up to 1.5$\times$ over a recent fine-grained, memory-aware fusion accelerator for Mamba.

Yannan Nellie Wu, Po-An Tsai, Saurav Muralidharan et al.

Due to complex interactions among various deep neural network (DNN) optimization techniques, modern DNNs can have weights and activations that are dense or sparse with diverse sparsity degrees. To offer a good trade-off between accuracy and hardware performance, an ideal DNN accelerator should have high flexibility to efficiently translate DNN sparsity into reductions in energy and/or latency without incurring significant complexity overhead. This paper introduces hierarchical structured sparsity (HSS), with the key insight that we can systematically represent diverse sparsity degrees by having them hierarchically composed from multiple simple sparsity patterns. As a result, HSS simplifies the underlying hardware since it only needs to support simple sparsity patterns; this significantly reduces the sparsity acceleration overhead, which improves efficiency. Motivated by such opportunities, we propose a simultaneously efficient and flexible accelerator, named HighLight, to accelerate DNNs that have diverse sparsity degrees (including dense). Due to the flexibility of HSS, different HSS patterns can be introduced to DNNs to meet different applications' accuracy requirements. Compared to existing works, HighLight achieves a geomean of up to 6.4x better energy-delay product (EDP) across workloads with diverse sparsity degrees, and always sits on the EDP-accuracy Pareto frontier for representative DNNs