Originally Posted by euphoria89
IO latency is really the only element that differs from interface-interface. VI/Samples/Native effects are all computed by the host CPU/RAM.
I don't agree, sorry.
Latency is only one factor that needs to be weighed up. The largest factor separating audio interfaces is the driver performance @ those respective latencies, VI Samples/Native effects may be dependent on CPU/memory resources, but without the foundation of the driver performance in the first instance, performance variables can be huge.
From all accounts, the Apollo has the same amount (if not slightly improved) latency from other similar devices. Someone posted values of around 5.3ms at 32 samples/buffer at 44.1Khz. If working at 96Khz, this would be less than half this value
Can you point me to the thread where that is posted please.
Is that Round Trip Latency or just for the Output ?
FWIW : Both RME and Lynx PCIe /ADDA deliver under 3 ms RTL @ 032 Buffers , the UFX delivers 5.056 under USB/ 5.546 under FW @ 064
The practice of doubling the sample rate simply to lower latency doesn't pay off performance wise due to the extensive increase in processing overhead to do so.
As I mentioned above but its worth repeating, RTL is only 1/2 the equation, the second half being how the drivers actually perform at those latency settings.