macpro build - what and why

May 12, 2020

In March of 2020, I went looking for a project. I was looking for something that I could focus some extra time and energy on (that wasn’t just frantically refreshing news sites). I was working from home full time for the first time in my life, a change which necessitated some alterations to my workspace. Graduate school was finished, so I could afford some instability on my personal machine, and the 5k iMac that had served as the anchor of my home computing life was now in the way (I couldn’t use it for work, so it ended up awkwardly shunted aside most days, and moving it back into place every night was just annoying enough to be untenable. It was easier to just leave my work machine plugged in and running, but that lead to the temptation to do juuuust a little more work whenever I sat down at my desk; you can imagine the impact that had on work-life balance).

Enter the 5,1 Mac Pro.

the cheese grater #

The 5,1 Mac Pro, released in 2010, with a minor spec bump shipped in 2012 and ultimately replaced by late 2013’s trash can, is a very special machine. Due to the design choices (and expense) of the models that replaced it, it’s had a long and vibrant life as an expandable, flexible, workstation that can be kitted out for a variety of use cases. It holds a special place in my heart as the most powerful machine Apple was shipping during my time as a Genius; it was the most complicated machine to troubleshoot, given the flexibility and complexity of its internals, but it was always a thrill to see one sidle up to the bar. It was a machine that did work (or at least, purported to. I fully recognize the myth of the Mac Pro, which was always more costly than it had any right to be).

Could it be my 2020 computer?

what’s the goal? #

What am I trying to do here, exactly? I’m looking to wrangle up a Mac desktop, responsive enough for day to day use, with enough compute and memory to handle my polyglot programming (virtualization and containerization, some Go, a smattering of C++/Clojure/Swift/Python, depending on what I’m picking at on a particular day), and the graphical power to run the handful of games (some Blizzard titles, Total War: Warhammer and its sequels, Tabletop Simulator).

I’m loosely describing my requirements as: a desktop, running macOS, built by Apple.

Why a desktop?

In my experience, laptops add a thin layer of unreliability when being used permanently docked at a desk. Peripheral negotiation is often fussy, cooling can be a problem, and ultimately it feels to me like a misuse of the object. Look at a laptop; it’s fundamentally designed for portability. Using it permanently tethered feels like hammering nails in with the back end of a screwdriver.

Why macOS?

Windows is a tire fire. The software ecosystem is a Hieronymus Bosch style rhizome of misery and suffering, and software development on Windows outside of the Microsoft ecosystem just sucks. That’s all a deliberately inflammatory description, but it captures how I feel (and the bulk of my experience trying to develop for Linux systems on Windows in my previous job). Window management is remedial (whoever thought full screen and half screen splits were a good idea, and parasitically infected other operating systems with that idea should be tried at the Hague), keyboard shortcuts across the OS for text wrangling suck, and Emacs on Windows suffers from all sorts of painful compromises.

A more measured answer to “Why not Windows?” is that I don’t have room in my brain at the moment for a detailed enough mental model of the foibles and pit traps of Windows 10. I’m not particularly interested in building that model, to be totally honest, since I find the essential primitives of Windows as an operating system (the registry? really?) and the user punishing choices (adware in the start menu, user hostile updates) alien and off putting. It’s the only way to play PC games, so I’ll always have it installed somewhere in the house, but I’d like it as cordoned off as possible.

The desktop experiences of the non-macOS *nixs are unpleasant for me. Window management and keyboard shortcuts tend to ape Windows out of the box (yes, there are distros and customization paths to mimic macOS, but they’re never quite 100% reliable in my experience). I’ll continue to happily run Linux and BSD servers, both in the house on a handful of headless machines and in VPS’s, but for a desktop machine macOS is the best choice for me.

Why not a Hackintosh?

Given the decision to run macOS and the computational/GPU requirements, an obvious question would be “Why not build a Hackintosh?” OpenCore has come a long way, the community is active and communicative, AMD has some rad chip offerings bringing high core counts way down in price - there are a lot of compelling reasons to build a Hackintosh. I’ve done it before, almost half a decade ago, and found Clover pretty straightforward to configure, and the resulting machine was powerful and flexible. But honestly, I’ve built more than enough PCs. It’s boring, in many ways, and I don’t think building a generic PC and installing macOS on it would be the engaging project that I’m looking for amidst all of this chaos.

I have a great deal of affection for the 5,1 Mac Pro. The high core count, high memory configuration is surprisingly effective in 2020, and fits my use cases especially well. Most intriguingly, the vibrant Mac Pro community has made huge leaps in recent months, bringing Catalina support, hardware acceleration and, most importantly to me, Thunderbolt 3. Thunderbolt support would mean one cable to plug in my work laptop during the day, and a single cable moving over to switch to my main machine outside of work.