· 2019
ExTensor
- 235citations
- 2019year
Short summary
ExTensor accelerates sparse tensor algebra by hierarchically eliminating computations where operands are zero, using rapid intersection finding to avoid memory latency.
AI-generated from the title and abstract; the full text is not read.
Key points
- Proposes ExTensor for accelerating sparse tensor algebra via ASICs.
- Employs hierarchical elimination of computations in the presence of sparsity.
- Relies on rapidly finding intersections of non-zero elements.
- Enables new data fetching mechanisms and avoids memory latency overheads.
AI-generated from the title and abstract; the full text is not read.
Abstract
Generalized tensor algebra is a prime candidate for acceleration via customized ASICs. Modern tensors feature a wide range of data sparsity, with the density of non-zero elements ranging from 10-6% to 50%. This paper proposes a novel approach to accelerate tensor kernels based on the principle of hierarchical elimination of computation in the presence of sparsity. This approach relies on rapidly finding intersections---situations where both operands of a multiplication are non-zero---enabling new data fetching mechanisms and avoiding memory latency overheads associated with sparse kernels implemented in software.
The authors' abstract, as published at the source., 2019 · DOI ↗
Continue with a free account
Ask the paper: 3 free questions a day about this paper; save it, get its citation, new summaries every day for your field. Takeaways are Premium.
Continue free on the webSign in with Google or Apple; no card needed. You come back to this paper.
On your phone:
Field: Computational Mathematics
Computational MathematicsMathematics