Some of my thoughts and observations here might have been covered in previous threads, so apologies in advance for the redundancy. But, I had some recent observations to share to hopefully get to the bottom of my current dilemma.

I've been considering ditching the hardware console for some time and moving things completely "in the box." (With the obvious exception of mic preamps, and maybe a few favorite outboard processors...) So, I've been researching a few different options as far as interfaces/sound cards. It just so happens that a friend of mine is looking to part with his brand new Focusrite Scarlett 18i20 (it wasn't really what he was after for what he does...), so he's letting me demo it for a while. I also got my hands on a Digigram VX882e 8-channel sound card, so I've been fooling with that too.

I've been trying to set up a situation where I have a multitrack session, about 16 tracks recorded, and I'm overdubbing some vocals, as this is the most likely situation I'd find myself in aside from just mixing. I've got the session up in SAW, and I've got the mic going into, say, the Focusrite's Input 1. I've got headphones plugged into the Focusrite's headphone out. I'm trying to set up a route where I get the mix in the phones (using Aux 6, duplicating the main mix, and sending the Aux 6 Master to the Focusrite output assigned to the phones...), and then adding reverb to the phones mix, following the suggestions in the SAW manual.

The signals are all going where they're supposed to, but no matter what buffer settings I use, I still get noticeable latency delay, even when I'm just monitoring and not actually recording or playing back. The amusing bit is that this Focusrite 18i20 is advertised as 2ND GENERATION(!!), meaning they supposedly redesigned it for "super low" latency and more stable drivers. The stable drivers, I believe. The "super low" latency, not so much. I ran a round-trip latency test on it, and it clocked in at around 9ms. The Digigram card was around 10ms. Just for grins, I put the buffer settings in the cellar, 2x32, and things fell apart. So the lowest I can go is 64, and even then there's an occasional dropout if I have fewer than 4 buffers.

I could just use the Focusrite's own software monitor mixer to set up a headphone mix between the mic and the rest of the mix, but then I wouldn't be able to add mic effects to that mix without introducing a noticeable delay.

Unless I'm just not setting things up correctly.

I was a bit confused as to where the reverb plugin for the headphone mix should be patched in SAW.
Would the upcoming 64-bit version of SAW handle this any better?
The system I'm running this on is an i5 with 8GB of RAM. Is that enough?
Does SAC do a better job with this sort of thing?
Is there a way to monitor the clean mic signal in the Focusrite (or other interface), but still run a reverb return for it through SAW? (One can live with a slight delay on just the reverb...)