AOCL-Sparse contains basic linear algebra subroutines for sparse matrices and vectors optimized for AMD processors. It is designed to be used with C and C++.

Highlights of AOCL-Sparse 3.1

  • API for Sparse Matrix Dense Matrix Multiply (SPMM) giving dense matrix output
    • Supports single and double precision data types
    • Supports general sparse matrices in CSR format with zero-based indexing and no transpose
  • Conversion routine “CSR to Dense” that converts sparse matrix in CSR format to dense matrix format

Highlights of AOCL-Sparse 3.0

  • Supports CSR, Ellpack, Diagonal, Blocked-CSR data formats for SPMV function
  • New API, Sparse Triangular Solve (TRSV) for Single and Double Precision data types
  • Supports General matrices with zero-based indexing and no transpose.
  • New Sparse data format conversion routines:
    • CSR to Ellpack
    • CSR to Diagonal
    • CSR to Blocked-CSR
    • CSR to CSC

The packages containing AOCL-Sparse binaries, examples and documentation are available in the Downloads section below.

Source code for AOCL-Sparse will be available shortly on GitHub https://github.com/amd/aocl-sparse

Refer here for prior versions of AOCL-Sparse documentation and downloads.

Download:

File Name Version Size Launch Date OS Bitness Description
Binary Packages Compiled with AOCC 3.2

File Name

Version

3.1

Size

424 KB

Launch Date

12/10/2021

OS

Ubuntu, SLES, CentOS, RHEL

Bitness

64-bit

Description

AOCC compiled AOCL Sparse library binary package with example code sha256 Checksum: 24184c1b515e21ca058222b4952649e38ba26671d2f8aee044d3205ece0d1fba
Binary Packages Compiled with GCC 11.1

File Name

Version

3.1

Size

420 KB

Launch Date

12/10/2021

OS

Ubuntu, SLES, CentOS, RHEL

Bitness

64-bit

Description

GCC Compiled AOCL sparse library binary package with example code sha256 Checksum: cb5760fc793105c659efe16681fe940be188a8c9385e8a1aefbd6748bbe6aede