How to Build NVIDIA's RAPIDS on Arch Linux on CUDA11

PKGBUILD's for building NVIDIA's RAPIDS (cuML/cuDF) on Arch Linux + CUDA11 a procedure of building NVIDIA's RAPIDS under Arch Linux with CUDA environment on '''Single Nvidia's GPU'''.

Prerequisites

• Hardware

** Intel or AMD's CPU

** NVIDIA's Single GPU which has architecture whose type is Pascal / Volta / Turing / Ampere. (eg GeForce / TITAN / Tesla / Quadro)

• Software

** Arch Linux or its derivatives.

** AUR / yay required

** Build on CUDA11.1 (+CUDNN8 + NCCL)

• Disclaimer

Operation is not necessarily guaranteed. The author are not responsible for any damage of your environment by any of the operations described here.

Order in build

1. protobuf-static

* Set ver 3.12.4 (do not build ver 3.13 or higher because its incompatible with cuDF).
* static library required for building treelite

2. arrow-cuda / python-pyarrow-cuda

* Set ver 0.17.1 (do not build ver 1.0.0 or higher because its incompatible with cuDF).
* Build with "CUDA" (differencet from AUR).
* Apache Orc is required located in AUR.
* During compilation in Apache Flight, there is a patch so that gRPC-c++-plugin cab be easily integrated in Higher Version.

3. python-cmake-setuptools

* Build  Python Library for enabling to run cmake during "python setup.py build"

4. Cupy

* An implementation of NumPy-compatible  array on CUDA presented by Preferred Networks.
* AUR provides the same package. However,  AUR only contains older ver (7.2.0). which does not support CUDA 11.0 , (current ver. is 8.0.0) 
*  See [github](https://github.com/cupy/cupy)

5. Openuxc / python-ucx-py

* an optimized communication layer for Message Passing (MPI), PGAS/OpenSHMEM libraries and RPC/data-centric applications. and wrapper for python.
* See [OpenUCX's https://github.com/openucx/ucx)
* See [ucx-py's github](https://github.com/rapidsai/ucx-py)

6. XGBoost

* an optimized distributed gradient boosting library 
* Builds Library and its Python-Wrapper
* See [github](https://github.com/dmlc/xgboost)

7. treelite

* a model compiler for efficient deployment of decision tree ensembles 
* Builds Library and its Python-Wrapper
* See [github](https://github.com/dmlc/treelite)

8 RMM

* RAPIDS Memory Manager provided by  Rapids.
* See [github](https://github.com/rapidsai/rmm)
* amend PKGBUILD and setup.py.patch to avoid missing libraries (cudart fmt...) (2020.8.21. 2300GMT)

9 dask-cuda

* Utilities for CUDA and DASK.
* See [github](https://github.com/rapidsai/dask-cuda)

10 nvtx, python-nvtx

* NVIDIA Tool Extension Library 

11 cuDF

*  a GPU DataFrame taking place of pandas
* See [github](https://github.com/rapidsai/cudf)

12 cuML

*  A suite of libraries that implement machine learning algorithms presented by Rapids.
* See [github](https://github.com/rapidsai/cuml)
* amend PKGBUILD and setup.py.patch to avoid missing libraries (cublas cusuparse cusolver ...) (2020.9.19. 1400GMT)

13. libcypher-parse

* required for building cuGraph

14. cuSignal

* GPU accelerated signal processing which may replace scipy signal?
* See [github](https://github.com/rapidsai/cusignal)

15. gdal, python-gdal

* a bug-fix patch required to build gdal under poppler upgraded to 20.08.1

16. rapids-cuSpatial

* a library for handling spatial and trajectory data.
* gdal and python-gdal are needed
* See [github](https://github.com/rapidsai/cuspatial)

17. rapids-cuGraph

* a library for Graph Analytics
* See [github](https://github.com/rapidsai/cugraph)

18. rapids-cuxfilter

* cuxfilter ( ku-cross-filter ) is a RAPIDS framework to connect web visualizations to GPU accelerated crossfiltering.
* See [github](https://github.com/rapidsai/cuxfilter)