Rindow Neural Networks installation

Operating environment
Installation instructions for Windows
Installation instructions for Linux
GPU/OpenCL support for Windows
GPU/OpenCL support for Linux

Operating environment

Rindow Neural Networks has been tested in the following operating environments:

PHP 8.1, 8.2, 8.3 (When using in PHP 7.x, 8.0 environment, please use Release 1.x.)
Windows 10 20H2 or later.
Ubuntu 20.04, 22.04 or Debian 12 or later
AMD/Intel CPU/APU 64bit (SSE2 or later)
OpenBLAS (>=0.3.20:Windows-x64, 0.3.20:Ubuntu-2204, 0.3.8:Ubuntu-2004, 0.3.21:Debian12)
CLBlast (>=1.5.2:Windows-x64, 1.5.2:Ubuntu-2204, >=1.5.2:Ubuntu-2004, 1.5.3:Debian12)

It also works with Intel/AMD CPU/APU and integrated graphics with OpenCL drivers.

Installation instructions for Windows

Installing PHP

For Windows 10/11, install PHP for Windows.

Download the PHP x64 version from https://windows.php.net/download/. Either Non Thread Safe or Thread Safe version is fine.
Extract to a location of your choice.
Create php.ini by copying php.ini-development.
Set execution PATH for PHP.EXE.
Make sure PHP works with PHP -v.

C:TEMP>COPY C:\php\php.ini-development C:\php\php.ini
Edit php.ini to your liking.

C:TEMP>PATH %PATH%;C:\php
C:TEMP>php -v
PHP 8.3.4 (cli) (built: Mar 13 2024 11:42:47) (NTS Visual C++ 2019 x64)
Copyright (c) The PHP Group
Zend Engine v4.3.4, Copyright (c) Zend Technologies
    with Zend OPcache v8.3.4, Copyright (c), by Zend Technologies
C:TEMP>

Install composer.

Download composer from https://getcomposer.org/download/.
Copy composer.phar to the directory where the execution PATH is set.
Create composer.bat in the same location.

C:TEMP>COPY composer.phar C:\bin
C:TEMP>CD \bin
C:bin>echo @php "%~dp0composer.phar" %*>composer.bat

Install the PHP extensions required by Rindow Neural Networks.

Download and unzip the corresponding pre-built binary file from https://github.com/OpenMathLib/OpenBLAS/releases.
Download and unzip the corresponding pre-built binary file from https://github.com/rindow/rindow-matlib/releases.
Set the OpenBLAS and Rindow-Matlib DLL paths to the execution path.
Make the necessary settings in php.ini. - memory_limit = 8G - extension = ffi - extension=gd - extension = mbstring - extension=openssl - extension=pdo_sqlite - extension=zip
Make sure PHP extensions are loaded with PHP -m.

C:TEMP>PATH %PATH%;C:\OpenBLAS\OpenBLAS-0.3.26-x64\bin
C:TEMP>PATH %PATH%;C:\Matlib\rindow-matlib-1.0.0-win64\bin

Edit php.ini

C:TEMP>php -m
[PHP Modules]
...
ffi
...
pdo_sqlite
...
C:TEMP>

Install Rindow Neural Networks.

Create your project directory.
Install rindow/rindow-neuralnetworks with composer.
Install rindow/rindow-math-matrlix-matlibffi with composer to speed up.
Install rindow/rindow-math-plot with composer for graph display.
Verify that the state of rindow-math-matrlix is Advanced or Accelerated.

C:TEMP>MKDIR \tutorials
C:TEMP>CD \tutorials
C:tutorials>composer require rindow/rindow-neuralnetworks
C:tutorials>composer require rindow/rindow-matrlix-matlibffi
C:tutorials>composer require rindow/rindow-math-plot
C:tutorials>vendor/bin/rindow-math-matrix
Service Level   : Advanced
Buffer Factory  : Rindow\Math\Buffer\FFI\BufferFactory
BLAS Driver     : Rindow\OpenBLAS\FFI\Blas
LAPACK Driver   : Rindow\OpenBLAS\FFI\Lapack
Math Driver     : Rindow\Matlib\FFI\Matlib

Run the sample program

Run the sample to see if it works.
The results are displayed graphically.

C:tutorials>MKDIR samples
C:tutorials>CD samples
C:tutorials\samples>COPY ..\vendor\rindow\rindow-neuralnetworks\samples\* .
C:tutorials\samples>php mnist-basic-clasification.php
Downloading train-images-idx3-ubyte.gz ...Done
....
Epoch 4/5 [.........................] 1 sec. remaining:00:00  - 2 sec.
 loss:0.1264 accuracy:0.9640 val_loss:0.1246 val_accuracy:0.9604
Epoch 5/5 [.........................] 1 sec. remaining:00:00  - 2 sec.
 loss:0.1054 accuracy:0.9698 val_loss:0.1129 val_accuracy:0.9675

The graph is displayed

Installation instructions for Linux

Install php.

Install php-cli, php-mbstring, and unzip using the apt command.

$ sudo apt install php8.3-cli php8.3-mbstring php8.3-curl php8.3-sqlite3 php8.3-gd php8.3-xml php8.3-opcache unzip
$ php -v
PHP 8.3.4 (cli) (built: Mar 16 2024 08:40:08) (NTS)
Copyright (c) The PHP Group
Zend Engine v4.3.4, Copyright (c) Zend Technologies
    with Zend OPcache v8.3.4, Copyright (c), by Zend Technologies

Install composer.

Download composer from https://getcomposer.org/download/.
Copy composer.phar to the directory where the execution PATH is set.
Make composer in the same place.

$ cp composer.phar ~/.local/bin
$ cd ~/.local/bin
$ cat > composer
#!/usr/bin/env sh
dir=$(cd "${HOME}/.local/bin" && pwd)
php "${dir}/composer.phar" "$@"
^D
$ chmod +x composer
$ composer -V
Composer version 2.6.6 2023-12-08 18:32:26

Install the libraries required by Rindow NeuralNetworks.

Install OpenBLAS with apt command
Download the latest version of Rindow-Matlib’s pre-built binary files from https://github.com/rindow/rindow-matlib/releases.
Install the downloaded deb file using the apt command.
Set Rindow-Matlib to serial mode for use with PHP.

$ sudo apt install libopenblas0-openmp liblapacke
$ wget https://github.com/rindow/rindow-matlib/releases/download/X.X.X/rindow-matlib_X.X.X_amd64.deb
$ sudo apt install ./rindow-matlib_X.X.X_amd64.deb

But if you are not allowed to use the openmp version of openblas, there is another way. Click here for more information.

Install Rindow Neural Networks.

Set the image display command for rindow-math-plot.
Create your project directory.
Install rindow/rindow-neuralnetworks with composer.
Install rindow/rindow-math-matrix-matlibffi with composer to speed up.
Install rindow/rindow-math-plot with composer for graph display.
Verify that the state of rindow-math-matrlix is Advanced or Accelerated.

$ RINDOW_MATH_PLOT_VIEWER=/some/bin/dir/png-file-viewer
$ export RINDOW_MATH_PLOT_VIEWER
$ mkdir ~/tutorials
$ cd ~/tutorials
$ composer require rindow/rindow-neuralnetworks
$ composer require rindow/rindow-math-plot
$ vendor/bin/rindow-math-matrix
Service Level   : Advanced
Buffer Factory  : Rindow\Math\Buffer\FFI\BufferFactory
BLAS Driver     : Rindow\OpenBLAS\FFI\Blas
LAPACK Driver   : Rindow\OpenBLAS\FFI\Lapack
Math Driver     : Rindow\Matlib\FFI\Matlib

Run the sample program

Run the sample to see if it works.
The results are displayed graphically.

$ mkdir samples
$ cd samples
$ cp ../vendor/rindow/rindow-neuralnetworks/samples/* .
$ php mnist-basic-clasification.php
Downloading train-images-idx3-ubyte.gz ...Done
....
Epoch 4/5 ........................ - 10 sec.
 loss:0.1276 accuracy:0.9641 val_loss:0.1162 val_accuracy:0.9649
Epoch 5/5 ........................ - 11 sec.
 loss:0.1063 accuracy:0.9703 val_loss:0.1059 val_accuracy:0.9688

Note: Specify “viewnior” etc. for RINDOW_MATH_PLOT_VIEWER

The result is displayed as a graph.

Result

GPU/OpenCL support for Windows

OpenCL can be used by default on Windows.

Please download the CLBlast library and set the execution path.

CLBlast library

C:TEMP>PATH %PATH%;C:\CLBlast\CLBlast-1.6.2-Windows-x64\bin

Configure the rindow-neuralnetworks backend to use OpenCL.

Verify that rindow-math-matrix is Accelerated and the OpenCL driver is recognized.
Set environment variables to use GPU in rindow-neuralnetworks backend.
Run the sample program.

C:tutorials>vendor\bin\rindow-math-matrix
Service Level   : Accelerated
Buffer Factory  : Rindow\Math\Buffer\FFI\BufferFactory
BLAS Driver     : Rindow\OpenBLAS\FFI\Blas
LAPACK Driver   : Rindow\OpenBLAS\FFI\Lapack
Math Driver     : Rindow\Matlib\FFI\Matlib
OpenCL Factory  : Rindow\OpenCL\FFI\OpenCLFactory
CLBlast Factory : Rindow\CLBlast\FFI\CLBlastFactory

C:tutorials>SET RINDOW_NEURALNETWORKS_BACKEND=rindowclblast::GPU
C:tutorials>cd samples
C:samples>php basic-image-clasification.php

Note: RINDOW_NEURALNETWORKS_BACKEND can specify an OpenCL device type or a set of platform ID and device ID in addition to a name such as rindowclblast. If you have two or more GPUs, you can use this specification method to identify them. for example;

rindowclblast => platform #0, device #0
rindowclblast::GPU => GPU type device: Integrated GPU, etc.
rindowclblast::CPU => CPU type device: pocl-opencl-icd, etc.
rindowclblast::0,0 => platform #0, device #0
rindowclblast::0,1 => platform #0, device #1
rindowclblast::1,0 => platform #1, device #0

If you are unable to successfully set the target GPU, please check the OpenCL device status using the clinfo command.

C:tutorials>vendor\bin\clinfo
Number of platforms(1)
Platform(0)
    CL_PLATFORM_NAME=Intel(R) OpenCL
    CL_PLATFORM_PROFILE=FULL_PROFILE
....
...
..

GPU/OpenCL support for Linux

It is essential that OpenCL works properly in the Linux environment. (That’s quite difficult)

Install the OpenCL environment.

$ sudo apt install clinfo
$ sudo apt install intel-opencl-icd

Linux standard OpenCL drivers include:

mesa-opencl-icd
beignet-opencl-icd
intel-opencl-icd
nvidia-opencl-icd-xxx
pocl-opencl-icd

The standard Linux OpenCL driver does not work properly, so we deal with it on a case-by-case basis to make it work somehow for each driver and version.

Check that OpenCL is running using the clinfo command.

$ clinfo
Number of platforms                               1
  Platform Name                                   Intel Gen OCL Driver
  Platform Vendor                                 Intel
....
...
..

Install the CLBlast library.

$ sudo apt install libclblast1

If you are using Ubuntu 20.04, your distribution does not include clblast. Download and install the clblast pre-built binaries.

$ cp vendor/rindow/rindow-clblast-ffi/clblast-packdeb.sh .
$ vi clblast-packdeb.sh
CLBLASTVERSION=1.6.2   <===== change
$ sh clblast-packdeb.sh
$ sudo apt install ./clblast_X.X.X-1_amd64.deb