Best Sample Rate To Reduce Latency

Greetings. The Waves plugins latency chart reports the sample latency of Waves plugins at different sample rates. Interestingly, the chart reveals that latency does not always diminish with a higher sampling rate, as indicated by the following examples. Note, I ignored the “Venue TDM” column, as this does not apply to me. Instead, I am interested in how latency tracks both natively and on a SoundGrid server. Also, I believe this chart depicts round-trip latency values (not sure).

Abbey Road Chambers (Native / SoundGrid)
101 samples @ 44.1 kHz and @ 48 kHz
33 samples @ 88.kHz and @ 96 kHz

API 550 & API 2500 (Native / SoundGrid)
0 samples @ all sample rates

API 560 (Native / SoundGrid)
65 samples @ all sample rates

CLA MixHub (Native / SoundGrid)
59 samples @ 44.1 kHz and @ 48 kHz
20 samples @ 88.kHz and @ 96 kHz
(This plugin is particular noteworthy, as it is designed to be used across all channels & busses.)

MaxxVolume (Native / SoundGrid)
64 samples @ 44.1 kHz and @ 48 kHz
128 samples @ 88.kHz and @ 96 kHz

MV2 (Native / SoundGrid)
64 samples @ 44.1 kHz and @ 48 kHz
128 samples @ 88.kHz and @ 96 kHz

Torque (Native / SoundGrid)
32 samples @ 44.1 kHz and @ 48 kHz
64 samples @ 88.kHz and @ 96 kHz

I have two questions. First, why do latency values not change consistently across the Waves plugins when run natively or on SoundGrid? For some plugins the latency decreases as the sample rate increases, and for others the inverse is true.

Second, given this inconsistency and acknowledging both the increased recording quality and file sizes at a higher sample rate, what has been your practical experience with respect to latency using Waves plugins processed by a SoundGrid Extreme Server-C both during recording and mixing? I realize this is considerably less of an issue during mixing, as one can increase the SoundGrid server buffer size. Still, I would appreciate your feedback for both recording and mixing.

Thank you.

I’m not a dev, but I’ll hazard an educated guess at this…

I think latencies vary because of what is required by the plugin’s algorithm, which might sound obvious. But I would say that some algorithms require only a certain number of samples in order for it to function correctly, whereas other algorithms probably require a slice of “time” to do their job correctly.

So 1ms would ether be 48 samples at 48k or 96 samples at 96k. At least in this case the latency remains the same 1ms no matter what sample rate you work at.

At least, this is what I can best make of it. I’d be interested in hearing from a real dev though. :smirk: