Installation#
Prerequisites#
GPU: NVIDIA Volta or newer. The CUDA 13 builds require Turing or newer.
NVIDIA driver: r525 or newer. The CUDA 13 builds require r580 or newer.
CUDA Toolkit: 12.0 or newer.
Optional libraries#
nvTIFF loads these libraries at runtime. They are needed only for the compression types that use them.
nvCOMP 6 or newer: Deflate and Zstandard.
nvJPEG2000 0.8.1 or newer: JPEG 2000.
Supported Platforms#
Linux#
nvTIFF requires glibc 2.17 or newer. The following distributions are supported.
x86_64#
RHEL 8, 9, 10
Rocky Linux 8, 9, 10
Amazon Linux 2023
Microsoft Azure Linux 3
Ubuntu 22.04, 24.04, 26.04
OpenSUSE Leap 15, 16
SUSE SLES 15, 16
Debian 12, 13
Fedora 44
KylinOS V11
arm64 (sbsa)#
RHEL 8, 9, 10
Amazon Linux 2023
Microsoft Azure Linux 3
Ubuntu 22.04, 24.04, 26.04
OpenSUSE Leap 16
SUSE SLES 15, 16
Debian 12, 13
KylinOS V11
arm64 (sbsa) Jetson#
dGPU and iGPU with OpenRM/nvgpu.
Ubuntu 24.04 (JP7.x)
aarch64 Jetson#
Ubuntu 22.04
Windows#
Windows 10
Windows 11
Windows Server 2022
Tegra#
arm64-sbsa builds are supported on Thor starting with CUDA 13.0.
arm64-sbsa builds are supported on Orin starting with CUDA 13.2.
aarch64-jetson binaries based on CUDA 12 are available for Orin.
Installing nvTIFF#
NVIDIA DevZone#
nvTIFF binaries and download instructions are available on the NVIDIA DevZone.
For Linux, the binaries are available as an archive (.tar.xz) and in package manager formats (.deb, .rpm).
For Windows, the binaries are available as an archive (.zip) and as an installer (.exe).
PyPI#
nvTIFF binaries are available on PyPI and can be installed with pip, which makes it easy to use the library in Python environments.
To install nvTIFF for CUDA 13.x:
python3 -m pip install nvidia-nvtiff-cu13
To install nvTIFF for CUDA 12.x:
python3 -m pip install nvidia-nvtiff-cu12
To install nvTIFF for CUDA 12.x on Tegra platforms:
python3 -m pip install nvidia-nvtiff-tegra-cu12