Niklas Hörnblad | Searcharxiv

arXiv2026

Tensor Algebra Processing Primitives (TAPP): Towards a Standard for Tensor Operations

To address the absence of a universal standard interface for tensor operations, we introduce the Tensor Algebra Processing Primitives (TAPP), a C-based interface designed to decouple the application layer from hardware-specific implementations. We provide a mathematical formulation of tensor contractions and a reference implementation to ensure correctness and facilitate the validation of optimized kernels. Developed through community consensus involving academic and industrial stakeholders, TAPP aims to enable performance portability and resolving dependency challenges. The viability of the standard is demonstrated through successful integrations with the TBLIS and cuTENSOR libraries, as well as the DIRAC quantum chemistry package.

cs.MS↗

arXiv2024

BLAS-like Interface for Binary Tensor Contractions

In the world of linear algebra computation, a well-established standard exists called BLAS(Basic Linear Algebra Subprograms). This standard has been crucial for the development of software using linear algebra operations. Its benefits include portability with efficiency and mitigation of suboptimal re-implementations of linear algebra operations. Multilinear algebra is an extension of linear algebra in which the central objects are tensors, which are generalizations of vectors and matrices. Though tensor operations are becoming more common, they do not have a standard like BLAS. Such standardization would be beneficial and decrease the now-visible replication of work, as many libraries nowadays use their own implementations. This master thesis aims to work towards such a standard by discovering whether or not a BLAS-like interface is possible for the operation binary tensor contraction. To answer this, an interface has been developed in the programming language C together with an implementation and tested to see if it would be sufficient. The interface developed is: xGETT(RANKA, EXTA, INCA, A, RANKB, EXTB, INCB, B, CONTS, CONTA, CONTB, PERM, INCC, C) with the implementation and tests, it has been deemed sufficient as a BLAS-like interface for binary tensor contractions and possible to use in a BLAS-like standardization for tensor operations.

cs.MS↗