Can you suggest a PC machine with better cpu, equivalent GPU and similar potential with number of pcie lanes and 2x 10gb ethernet + thunderbolt i/o’s? Genuinely interested to see an equivalent and the price it would be from the pc world.
Hey, I don't think all of these criteria are even a consideration for 3D so I would not factor all of them in personally for a purchase. I'm also not trying to say PC is generally better (I like macOS - and I even bought a Trashcan) or go buy cheapest components and self-assemble - just that for 3D the MP seems rather uninteresting at that price.
Thunderbolt in particular seems entirely uninteresting for this case - I don't think many in the PC world even know what it is.
I'd go with a prebuilt system from a domestic retailer I had good experiences with and base it off one of the top contenders from this list:
https://www.cpubenchmark.net/singleThread.html
That's the single most important criteria to speed up interaction with the machine. Number of cores is pretty uninteresting outside rendering and perhaps simulation (and I'd want a separate box for that if its a regular part of the work and can't be performed overnight), I think anything beyond 6 cores is plenty, you'll be lucky to find apps that really go beyond two cores or so for most things.
3D apps don't usually stream large datasets in or out - assets get loaded into the scene and stay in memory and as I found out the hard way when making the jump from spinners to SSDs: these apps spend far more time processing data internally when im/exporting than they do accessing the disk so having the most bandwidth for quick loading isn't all that interesting. Standard NVMe is already overkill.
The machine I'd spec would come in somewhere between 4 and 5000 Euros (base MP is 6499 Euros here) with either Intel i9 (proven platform without upgrade path) or AMD Ryzen (new but potential for future upgrades apparently there).
128 GB of RAM is the sore point here - plenty nowadays though but wouldn't 256 be nice.... and as I mentioned previously - if you approach that then your 3D app is most likely already creaking and crashing with datasets too demanding to be handled reliably or fast enough.
Two smallish SSDs each for the OS/apps and two large ones for workfiles (mirrored/cloned) and a spinner for backup/versioning of files internally.
One 2080 TI GPU should do for now - if VRAM is the sore point then no choice but to spend around 2.7k for the Titan though. At any rate it would have to be Nvidia for best compatibility and RTX is taking off as an interesting feature.
All that wrapped in a silent case, adequate power supply and case fans provided by beQuiet. CPU cooler one of the huge ones from Noctua, mainboard either Asus or ASrock, I had good experiences with their overclocker ranges for build quality and edge case features (I'd not overclock personally).
But as said - I let other people handle the build and warranty, I just pick and choose from their part lists.
3D apps where you might be better off with the kind of bandwidth, memory limits and number of cores you can find in systems sold explicitly as workstations are few and far between if we are talking interactive work, not rendering.
Benchmarks indicate that ZBrush can utilize lots of cores well since it relies on the CPU for its viewport/canvas rendering. Other than that Mari might be a candidate for a workstation type setup where disk bandwidth could be more of a factor on large projects than raw CPU or GPU speed. This is a niche texturing app that seems to be losing ground to a very GPU-reliant and RTX-enhanced competitor though.