hiptensor: A C++ library for accelerating tensor primitives using GPU matrix cores¹

hiptensor is AMD's C++ library for accelerating tensor primitives using GPU matrix cores on AMD's latest discrete GPUs. hipTensor currently supports the following AMDGPU architectures: CDNA class GPU featuring matrix core support: gfx908, gfx90a, gfx940, gfx941, gfx942 as 'gfx9' :::{note} Double precision FP64 datatype support requires gfx90a, gfx940, gfx941 or gfx942.

... part of T2, get it here

URL: https://github.com/ROCm/hipTensor

Author: Advanced Micro Devices, Inc.
Maintainer: The T2 Project <t2 [at] t2-project [dot] org>

License: MIT
Status: Stable
Version: 6.4.0

Remark: Does cross compile (as setup and patched in T2).

Download: https://github.com/ROCm/hipTensor/ hipTensor-rocm-6.4.0.tar.gz

T2 source: hiptensor.cache
T2 source: hiptensor.desc

Build time (on reference hardware): 3300% (relative to binutils)²

Installed size (on reference hardware): 314.04 MB, 87 files

Dependencies (build time detected): bash cmake composable_kernel coreutils curl diffutils gawk gettext grep gzip hip-rocclr libedit libxml linux-header make openssl pkgconfig python rocm-cmake rocm-device-libs rocm-llvm sed sysfiles tar tbb zlib

Installed files (on reference hardware): n.a.

¹) This page was automatically generated from the T2 package source. Corrections, such as dead links, URL changes or typos need to be performed directly on that source.

²) Compatible with Linux From Scratch's "Standard Build Unit" (SBU).