Can anyone shed some light into why specifically AI developers prefer using Linux over MacOS? I realize the tradeoffs and preferences between Mac and Linux, but this article calls out AI devs specifically, and I wasn't sure the reason for that.
Granted, I think if you're building native mobile apps and web front-ends, your life is easier on a Mac just by virtue of having access to the same tooling as your design team. But for everything else, it just feels easier to get stuff done on Linux.
Complicating matters is the declining quality of Apple Macbook Pros and their propensity to thermally throttle. They are quite beautiful though! Still, as a 2018 15" MBP user at work, I often regret not getting a T or P-series Lenovo which carry with them their own flaws.
There is also the X1 Extreme.
Can in a pinch do everything on the laptop I can do on my dev desktops.
With iOS, Apple has it so that you more or less need a MacOS running XCode. While the Mac I use has Android Studio on it, I find myself switching back to Ubuntu Linux to program Android. It is just an easier workflow for me.
In the context it was written it sounds like “this other shadow IT usage can be under the table because it’s deemed unimportant to the business but now that our business is all-in on AI we can use that momentum to get better servers”.
Another major factor is CUDA - all the frameworks use CUDA and Nvidia cards, which MacOS doesn't support. OpenCL is way slower and undersupported. AMD has been working on ROCm but it doesn't feel like it's going anywhere fast enough. Everyone uses CUDA.
In our case I convinced that it's a design flaw in the MacBook Pro.
What we'll often do is use a small dataset locally on our desktop to confirm that something works, and then ship it to a multi-GPU setup in our cluster to train things for real-world uses. It's much easier to do this when you're developing on the same OS that is running in the cluster, hence Ubuntu desktop.
I guess I rarely run into ubuntu vs macos compatibility issues while I do notice the lack of QOL features on ubuntu. I would rather stay on macos