Yes. For me, stopping the transport, interrupting the sound in a linear DAW is the most counterproductive, the worst, most destructive thing that could happen to my creativity. There are a few workarounds (e.g. Live’s session view), but nothing really comes close to the modular approach.
The other big thing which annihilates my workflow is sitting in front of a DAW and a browser at the same time 
Interesting, I wondered why in the DX7 they called it an Algorithm instead of a patch.
So the only things that could lead to a perceptible differences would be bit depths and sample rates?
The Dx7 was 49K and 12bit plus a couple: In DX7 the actual sound was made with custom LSIs. I.e. there is no CPU involved… | Hacker News (ycombinator.com)
It kind of makes sense with the complex envelopes, I have begun to associate with digital rather than analog.
I know it’s pointless to argue, but it is not bit depth! It’s probably not sr either. It how much care you take, how much oversampling you use, how many bugs you have , if there is zipper noise in the envelopes. There are a ton of ways to mess these things up, and the two you mention are rarely the culprits (with modern software).
The way Chowning FM worked was there was a precomputed sine table, & FM synthesis involved varying the rate of scanning through that table. Wave table synthesis also uses scanning through waveforms but original Chowning FM was just sines.
Algorithms in FM synthesis is an imprecise use of the term. FM algorithms are configurations of the modulations between “operators” which are actually sine oscillators. It’s confusing because Chowning FM is software, and those words mean different things in the programming world.
Oh, sure. So by wavetable you mean a table of sin(x). Sure. I thought OP was implying wavetable for everything! Like sin(a + b* sin(x)). Which maybe someone was done, but would be kind of nuts.
In Chowning FM each operator is continuously scanning through the same sine table. It’s base frequency comes from midi notes. The output of other operators can modify its frequency, as can feedback. But it’s really a simple formula, that adds other operators to the base frequency to get the current frequency, then uses the current table position & that summed frequency to choose the next sample. Low frequency, short distance to next sample, high frequency longer distance. The resulting sample stream is smoothed so there are no big discontinuities.
Sure, that’s what they all do, including mine. OP implied/stated that some others used a different method.
Vital does that I’m sure they even show a 3d rendering of the wave table so you can see how the a and b parameters vary.
Wavetable Synthesis: A Complete Guide +150 Free Wavetables (producerhive.com)
Its a good idea, but IMO if you don’t have something that Transends the single cycle they sound thin to me. You know how sometimes out of sync oscillators are preferred over sync’d, the sound develops more.