Great post. Some interesting food for thought though: I have been finding myself on the edge of containers and VMs actually being useful. The issues usually arise when having to access graphics hardware. I have typically been in the games, film production, and robotics arenas and they all have a massive need for 3D graphics. ML also has the same needs these days. The support has been mediocre at best. I am looking forward to the day when I can virtualize and still get full access to the GPU.
The truth is that today, containers and VMs serve a large contingent of the developers out there but miss another huge faction. The one I am in. I try to keep abreast of the new developments in containers and VMs but each time I have tried to use them for something useful in my space I am turned away by disappointment after disappointment.
So to those in a position to help those like myself, please keep us in mind. The GPU is one of the most powerful and game changing pieces of hardware we have right now (along with mass amounts of cores and better multi-threaded program paradigms in lower level languages like C++ 11/14/17/20).
I will surely check it out. Does this offer 3D support for user interfaces as well?
To add a bit of context, I'm working on a 3D digital content application. I'd like to be able to build an Ubuntu environment with 3D graphics support for the user interface (and 3D rendering) on a Windows 10 machine. It happens to have significantly more (and faster) cores than my development Linux machine so compile times are much better. I can currently build there but I cannot test when using a VM because the graphics support is not up to par with hardware on the machine. The machine has an NVIDIA 2080 RTX in it.
Great post. Some interesting food for thought though: I have been finding myself on the edge of containers and VMs actually being useful. The issues usually arise when having to access graphics hardware. I have typically been in the games, film production, and robotics arenas and they all have a massive need for 3D graphics. ML also has the same needs these days. The support has been mediocre at best. I am looking forward to the day when I can virtualize and still get full access to the GPU.
The truth is that today, containers and VMs serve a large contingent of the developers out there but miss another huge faction. The one I am in. I try to keep abreast of the new developments in containers and VMs but each time I have tried to use them for something useful in my space I am turned away by disappointment after disappointment.
So to those in a position to help those like myself, please keep us in mind. The GPU is one of the most powerful and game changing pieces of hardware we have right now (along with mass amounts of cores and better multi-threaded program paradigms in lower level languages like C++ 11/14/17/20).
Not sure if you know about it, but NVIDIA Docker solves this issue for us:github.com/NVIDIA/nvidia-docker
We are playing in the ML area and have no issues using the GPU in Docker.
I will surely check it out. Does this offer 3D support for user interfaces as well?
To add a bit of context, I'm working on a 3D digital content application. I'd like to be able to build an Ubuntu environment with 3D graphics support for the user interface (and 3D rendering) on a Windows 10 machine. It happens to have significantly more (and faster) cores than my development Linux machine so compile times are much better. I can currently build there but I cannot test when using a VM because the graphics support is not up to par with hardware on the machine. The machine has an NVIDIA 2080 RTX in it.
Sorry, I don't know. Haven't used it for anything else than executing Tensorflow models.
Fair enough. Thank you for responding back.