DESCRIPTION: I know Canonical has put a lot of effort into Mir and at this stage it is probably "too big to fail". But for various reasons my bet is that it will fail. I think this is Canonical repeating Microsoft's Metro mistake. I have a $12K dollar desktop and I don't want an OS optimized for phones !!! I will be able to avoid it but I would rather your engineering effort was better placed.
ROLE/AFFILIATION: Software Engineer / Data Scientist
I think most people are unaware of Mir vs Wayland and the issues that Ubuntu using Mir will cause. I only know about it because I was thinking about building a tiling window manager (because I'm weird like that).
I think there is a good chance that Mir will be technically inferior to Wayland and a better chance that Ubuntu UI designers will reinvent the user experience from a phone perspective.
Pretty much. It's probably worth $6K at replacement cost today. It's for deep learning and I included what I spent on it and I've refreshed the 4 GPUs once. The current config is probably still good for a couple of years. Now if I need more GPUs I'll just get a second box. The irony is that I graphics power coming out of my ears yet I spend most of my time in vim and tmux.
Just out of curiosity - are the GPUs really the best bet for you at that point? I believe you can buy dedicated FPGA cards, some of which support OpenCL workloads even which might offer you better performance - though they might cost a shitpile too.
HEADLINE: Dump Mir!!!!!
DESCRIPTION: I know Canonical has put a lot of effort into Mir and at this stage it is probably "too big to fail". But for various reasons my bet is that it will fail. I think this is Canonical repeating Microsoft's Metro mistake. I have a $12K dollar desktop and I don't want an OS optimized for phones !!! I will be able to avoid it but I would rather your engineering effort was better placed.
ROLE/AFFILIATION: Software Engineer / Data Scientist