Do direct drive trainers simulate gradient resistance more accurately?

neeb

I have a Tacx Genius Smart (i.e. a high-end smart but non-direct drive trainer) I've been using for three or four years now. It's fine and just keeps going - I use it with a P2M crank based PM, so all the trainer needs to do is vary the resistance according to the gradient, which it does very well.

What would I gain by going to a state-of-the art direct drive trainer? Can I assume that the resistance simulation would be more accurate? With the Tacx I have the trainer difficulty setting in Zwift set at 85% which seems close to real resistance for a given gradient. I assume that the discrepancy here (~ 85% vs. 100%) is due to the relative inaccuracy of the Genius's power calculation (not that I'm using it for my Zwift power numbers), and maybe the unpredictability added by stuff like tyre pressure and roller pressure? With a good direct drive trainer, could I be confident that 100% trainer difficulty on Zwift would be a very accurate simulation of real world resistance for a given gradient?

Is "road feel" / pedalling feel better on direct drive trainers?

They make less noise, right? But some of them have pretty heavy flywheels - does this mean they are better or worse for transmitting noise and/or vibrations through floors and walls?

StillGoing

|The heavier the flywheel, the more realistic the feel. With a low end turbo, once you stop pedaling, you tend to grind to a halt almost immediately. With the higher end turbos and the heavier flywheel, you'll continue rolling and have less difficulty going over the top with the pedal stroke. The higher end turbos are seriously quiet. There's more noise from the chain and mechs than the turbos.

I doubt the heavier flywheel increases vibration.

dannbodge

Direct drive will generally simulate higher % gradients as they don't need to worry about tyre slip