For mainstream stuff, not really that by much. 3900X will only show its muscles if you throw VMs, numbering-crunching and content creation stuff at it.
I think you severely underestimate the difference between a 3770k core and a 3900x...
Nope, it is pretty close under mainstream stuff. Zen2 isn't that much faster than Sandy Bridge/Ivy Bridge at low-threaded stuff assuming clockspeeds are equal. 3770K easily boosts to its maximum of 3.9Ghz and can OC'ed easily to 4.2Ghz. 3900X is lucky to boost 4.5Ghz sustained. The difference between the isn't that dramatic unless you throw workloads that scale well beyond one or two-threads. That throws it squarely under number-crunching, content creation, VMs and power multi-tasking.
3900X is a very powerful workhorse CPU that has utterly disrupted the HEDT-scene, but frankly it is overkill for non-power users.
I think you vastly overestimate the demands of mainstream software and workloads. FYI, I have done comparisons between my current 9700K and 3570K systems. The 9700K only pulls ahead by significant margins under number-crunching, content creation and VMs (Thanks to proper VT-D support which 3570K lacks). Under mainstream stuff you would be hard-pressed to find a noticeable difference.
You aren't wrong about day to day work being pretty comparable.But I dunno... I've been struck by the speed gap between my Ryzen 1700 and the 8 core Sandy Bridge EP Xeon E5-4640 I just built. For well-threaded jobs the clock speed gap isn't gigantic - 2.5 GHz all core turbo on the Sandy, 3.15 GHz on Zen 1 - and the 1700 still walks away from it. In a Handbrake 1080p30 Fast encode the Xeon managed 35.8 fps, the Ryzen 59.1; in Cinebench R20, the Ryzen produced 3190, the Xeon 1750. And this is with the Ryzen running dual-channel DDR4-2400 versus the Xeon's quad-channel DDR3-1600. If I normalize those back, compensating for the Ryzen's 26% speed advantage, the Handbrake number would be 46.9 fps and 2532 for Cinebench R20. That's still an observed difference of ~30% and 44% to the aggregate IPC respectively.
The Xeon doesn't make me unhappy, but it hasn't put up the fight I expected. For weakly threaded apps there would be a full 1 GHz speed gap, so I'm not even going there, but there's more to the per-thread performance advantage than the clock speed delta alone makes up. I might manually underclock the Ryzen 1700 to a constant 2.5 GHz and re-run the benchmarks just for a closer apples to apples comparison... I'd grab an upgrade kit to nudge it to a matching 32 gigs of DDR4-3200 while I'm at it, but it's not hurting for that right now anyway.
Props to Starfalcon, by the way, for building a hyper rig. I'm planning to assemble a 3900x for a friend early next year to replace his iMac for multimedia work because he can't rationalize dropping six grand for an "entry-level" Mac Pro that would be beaten by a system costing about half as much. Every year Apple snorts more powdered money, I swear.
Work: Core i9 7940x, 32 gigs RAM, RX Vega 56 + GTX Titan Xm, Win10 Pro + Xubuntu 18.04
Media: Ryzen 7 1700, 16 gigs RAM, Radeon R9 Fury, Win10 Pro
Tinker: Xeon E5-4640, 32 gigs ECC RAM, GTX 750 Ti, Slackware 14.2
Read me at https://www.wallabyjones.com/