hwloc parsing: add tests for opencl coprocessors

[SteVwonder]

We currently have code that supports opencl devices, but no tests for that code. We should grab the hwloc output from a machine with AMD (or maybe even Xeon Phi) coprocessors and add that to the test suite.

[dongahn]

Thanks! yes this topic came out today and this will be very important!

[dongahn]

A question is, though how we can get AMD GPU machine access so that we can collect some hwloc data.

[SteVwonder]

@trws: will your testbed, by chance, have AMD GPUs?

[trws]

Yup. My primary goal is for it to have a platform fully capable for both ROCM and CUDA simultaneously so we can test a unified install and compiles to both at once.

…

--- Sent from Workspace ONE Boxer<https://whatisworkspaceone.com/boxer> On December 14, 2018 at 6:23:52 PM PST, Stephen Herbein <notifications@github.com> wrote: @trws<https://github.com/trws>: will your testbed, by chance, have AMD GPUs? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub<#413 (comment)>, or mute the thread<https://github.com/notifications/unsubscribe-auth/AAoStSDE0NILaafi_wqJ2gmYX6rADtmpks5u5F0ogaJpZM4Y4k8e>.

[trws]

In the mean time IPA, ulna, crayz and soon corona are the options. Crayz is the most stable of these.

…

--- Sent from Workspace ONE Boxer<https://whatisworkspaceone.com/boxer> On December 14, 2018 at 6:23:52 PM PST, Stephen Herbein <notifications@github.com> wrote: @trws<https://github.com/trws>: will your testbed, by chance, have AMD GPUs? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub<#413 (comment)>, or mute the thread<https://github.com/notifications/unsubscribe-auth/AAoStSDE0NILaafi_wqJ2gmYX6rADtmpks5u5F0ogaJpZM4Y4k8e>.

[dongahn]

This was done as part of #464.