Crowd Sourced Testing

I have taken measures to test as much of this package as possible and will continue to add more tests as the package develops. However, the fact that this package is intended to be used on different hardware inherently limits the extent of my testing (e.g different OS, GPU's, and OpenCL platforms).

As such, I am depending on the R user community to provide feedback on their experience installing and using this package. There are several levels of involvement a given user can choose.

Miminum/Automation

# Current Stable Release
wget https://cran.r-project.org/src/contrib/gpuR_1.0.1.tar.gz
R CMD check gpuR_1.0.1.tar.gz

Unit Tester

# Clone git repo (development)
git clone -b develop https://github.com/cdeterman/gpuR.git
devtools::test()

Naturally, you can just report whatever errors you find. If so, please be sure to also include the following information:

1. Operating System
2. GPU architecture (Vendor (e.g. Intel, AMD, NVIDIA) and Device Name (e.g. GeForce GTX 970)).
3. OpenCL version (eg. 1.1, 1.2, 2.0)

Involvement

However, troubleshooting these types of errors will be difficult for me to do alone. I can spin up virtual machines to troubleshoot OS problems and potentially OpenCL versions but the GPU hardware requires me to interact directly with the user. This can dramatically decrease turnaround time.

Ideally a user would be willing to at least troubleshoot the problem down to a specific error. I therefore provide the following recommendations to try and pinpoint errors.

1. If unit testing fails - try running individual test files e.g. devtools::test_file("tests/testthat/test_cpu_deepcopy.R"). Note - assumes in gpuR directory

Community

I have a gitter account setup for the package here. Designated channels for development and testing discussions are also provided. I am hoping that is will prove valuable so I can communicate with users (and with each other) more easily without cluttering the 'Issues' until a formal Issue is necessary.