Latency measurements when recording 16 inputs with Tascam 16x08 USB 2.0 with VST plugins

I've been experimenting with using the Tascam 16x08 USB 2.0 interface for recording live performance with Reaper with only software plugins for all FX and a relatively cheap but new Acer Swift 3 i5-1135G7 laptop. The setup seems to have sufficiently low latency for real time tracking.

I did some measurements of the latency. I really just want a set up where I can record guitar and mic playing along to backing tracks (e.g., see: https://www.tascamforums.com/threads/beach-boys-cover-sloop-john-b-with-us-16x08-interface.9636/)

But since the interface has 16 channels I was curious to test recording all 16. I recorded a click track and then played it out and then back into an input channel and recorded it, and measured the delay. This is with a Win10 laptop with everything possible uninstalled or turned off (e.g., no antivirus, no wifi). Using 96 kHz/24 bit and a 64 sample buffer (I think about the lowest before you start to get glitches), I measured ~6-7 ms. Turning on/off hardware (DSP) EQ and compressor didn't change this. If there is a 64 sample buffer on both input and output then ~1.2 ms of would be the buffers (2*64/96000), so ~5-6 ms must be the USB and/or computer speed. Interestingly recording all 16 channels simultaneously doesn't slow it much, so maybe it is mostly "overhead" in the drivers.

I then added EQ and compressor input VST plugins (the basic ones free in Reaper) to every channel and the master track, and a track with a reverb VST (free Voxengo "old school verb") with all 16 channels routed to it, and those don't slow it much. Then I put an Amplitube 5 "American Tube Amp (with spring reverb)" VST on one channel and that alone added about 2 ms, so that VST runs much slower than the others. But adding it to two other channels did not increase this, probably because the laptop has 4 processors/8 threads? So in the end, regardless of whether you record 1 vocal + 1 guitar with amp sim, or 16 channels with 3 amp sims, the overall latency is ~8-9 ms. And that seems tolerable for real-time monitoring (though I read that some people claim to be bothered by latency as little as ~2 ms). (* I'm assuming that the VSTs run at the same speed regardless of whether or not significant signals are going into them, because I only have the click signal going into one channel while the rest are recording nothing/just noise...)

Regarding nudging the items, that's true for recording, but I'm also trying to use the setup for live performance with "real time processing". But I'm finding the ~8-9 ms to be "fast enough" for me to not notice serious latency. I'm also experimenting with using xvive wireless mic and guitar relays, which may add up to ~5 ms, and those seem ok too. So I'm pretty happy. Though I wonder if there are some other interfaces that are faster, maybe due to better drivers, like maybe Focusrite (people seem to praise their drivers more often than many other brands)? I also don't actually need 16 inputs, just 2, so would consider maybe buying a cheaper 2 input interface it it was substantially lower latency. I have actually ordered a US-2x2HR for non-live recording and am going to test that, but imagine that since it is also USB 2, it will probably be similar to the 16x08?