This directory contains the R package for H2O4GPU, a collection of GPU solvers by H2O.ai with APIs in Python and R. The Python API builds upon the easy-to-use scikit-learn API. The h2o4gpu R package is a wrapper around the h2o4gpu Python module. The R package makes use of RStudio's reticulate package for facilitating access to Python libraries through R.
There are a few system requirements, including Ubuntu 16.04+, Python >=3.6, R >=3.1, CUDA 8 or 9, and a machine with Nvidia GPUs. The code should still run if you have CPUs, but it will fall back to scikit-learn CPU based versions of the algorithms.
The h2o4gpu Python module is a prerequisite for the R package. So first, follow the instructions here to install the h2o4gpu Python package (either at the system level or in a Python virtual envivonment). The easiest thing to do is to pip install the stable release whl file. To ensure compatibility, the Python package version number should match the R package version number.
The recomended way of installing the R package can is from CRAN using install.packages("h2o4gpu"). To install the development version of the h2o4gpu R package, you can install directly from GitHub as follows:
library(devtools)
devtools::install_github("h2oai/h2o4gpu", subdir = "src/interface_r")Using a Python virtual environment is a good solution if you don't want to upgrade your main Python installation to 3.6. If you installed the h2o4gpu Python module into a virtual environment, you will have to add a line of code to tell R which Python envivonment you want to use:
library(reticulate)
use_virtualenv("/home/username/venv/h2o4gpu") # set this to the path of your venvIf you have installed h2o4gpu Python module using Anaconda, then you can use the use_condaenv() function instead. More information about Python environment configuration is available in the reticulate user guide.
To test your installation, try the following example that builds a simple random forest classifier:
library(h2o4gpu)
# Prepare data
x <- iris[1:4]
y <- as.integer(iris$Species) # all columns, including the response, must be numeric
# Initialize and train the classifier
model <- h2o4gpu.random_forest_classifier() %>% fit(x, y)
# Make predictions
pred <- model %>% predict(x)For examples of how to use all of the functions in the package, please visit the vignettes section here.
If you have any issues, or have any recommendations to the installation instructions, please let us know by filing a GitHub issue. If there are installation issues, the first thing to check is the system requirements.
The first thing to check is that you actually have a machine with Nvidia GPUs. If you can run the nvidia-smi command in the shell and get an output that looks similar to this, it means you indeed have GPUs and working drivers on your machine:
username@gpubox:~$ nvidia-smi
Tue Mar 27 11:38:14 2018
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 387.34 Driver Version: 387.34 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
|===============================+======================+======================|
| 0 GeForce GTX 1080 On | 00000000:02:00.0 Off | N/A |
| 27% 39C P8 10W / 180W | 12MiB / 8112MiB | 0% Default |
+-------------------------------+----------------------+----------------------+
| 1 GeForce GTX 1080 On | 00000000:81:00.0 Off | N/A |
| 27% 39C P8 11W / 180W | 12MiB / 8114MiB | 0% Default |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: GPU Memory |
| GPU PID Type Process name Usage |
|=============================================================================|
| No running processes found |
+-----------------------------------------------------------------------------+
To check if CUDA 8 or 9 is installed, run the nvcc --version command. If you see this:
-bash: nvcc: command not found
That means that CUDA is not installed. If you do have CUDA installed, you will see something like this:
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2016 NVIDIA Corporation
Built on Tue_Jan_10_13:22:03_CST_2017
Cuda compilation tools, release 8.0, V8.0.61
If you try to train a model and you get a non-descript error like this:
> # Initialize and train the classifier
> model <- h2o4gpu.random_forest_classifier() %>% fit(x, y)
Error:
Or if you receive a full error like this:
Error: Python module h2o4gpu was not found.
Detected Python configuration:
python: /usr/local/bin/python
libpython: /usr/local/Cellar/python/2.7.10/Frameworks/Python.framework/Versions/2.7/lib/python2.7/config/libpython2.7.dylib
pythonhome: /usr/local/Cellar/python/2.7.10/Frameworks/Python.framework/Versions/2.7:/usr/local/Cellar/python/2.7.10/Frameworks/Python.framework/Versions/2.7
version: 2.7.10 (default, Jun 1 2015, 09:44:56) [GCC 4.2.1 Compatible Apple LLVM 6.1.0 (clang-602.0.53)]
numpy: /usr/local/lib/python2.7/site-packages/numpy
numpy_version: 1.12.0
h2o4gpu: [NOT FOUND]
python versions found:
/usr/local/bin/python
/usr/bin/python
/usr/local/bin/python3
That means that R package cannot locate the h2o4gpu Python module. To fix this, make sure you have installed the h2o4gpu Python module, and that you are using one of the reticulate functions (e.g. use_python(), use_virtualenv(), use_condaenv()) to specify which Python environment you want to use.
If you have multiple versions of Python installed on your machine and don't want to use the primary version (the one you get when you type python at the command line), then you may consider using reticulate::use_python() function to explicitly specify which one to use:
library(reticulate)
use_python("/usr/local/bin/python")If you encounter an issue that is not documented here, please file a GitHub issue to tell us about it.