HTTomoLibGPU is a library of data processing methods and reconstruction tools that are tailored for the GPU-efficient computations using CuPy API. Most of the methods were ported from the CPU implementations in TomoPy and Savu.

The reconstruction part in HTTomoLibGPU relies on ToMoBAR package which wraps ASTRA toolbox modules for parallel-beam geometry. The reconstruction modules use CuPy to support device-to-device operations.

Installation

HTTomoLibGPU is easily available via conda for Linux, but will be changed to support all OS’s. It can be installed with the command:

conda install -c httomo httomolibgpu

Usage

It is worth checking some tests, but simply import the module in Python and run it. The API reference is to be added…

Benchmarks

To be added…