Excellenti bro =)

 

i can go down to 8 samples with no glitch with a Behringer UMC204HD , here with 16 samples 1.86ms in, 3.49ms out (safe mode)



at 8 samples without safe mode

 

Last edited: Nov 29, 2023

That's already 5.35ms, assuming that's the real RTL. Not great for the lowest setting.

I get the 3.9ms at the 16 setting but prefer to use 32 and it's still lower than that at 4.7ms.

I guess these Focusrite interfaces are not really that bad. Don't think there's better than this for just a little over 100 "currencies".

Since they are very popular, it's easy to find people with low computer skills blaming them for their problems.

 

The number of samples is irrelevant. There are likely many interfaces that have lower latency at 2-4x the number of samples, and can run a greater number of plugins as well.

There's not even a Focusrite interface in the top 10, but your post could still be 100% accurate if you've never used anything better.
https://dawbench.com/images/DAWbench LLP Database-December-2023.pdf

 

The Scarlett range is Focusrite's entry level, consumer-grade stuff; however you want to frame it. Their Clarett line are better interfaces, and more expensive. I had a lot of problems with a Focusrite 18i20 mk3 on Mac. Their software was awful. The version I was using was brand new software for their MK3, but it would keep dropping connection to the interface and other funny things.

But what are you referring to plugins for, when discussing Round Trip Latency specs of interfaces? If you are not running the plugins with some type of external hardware DSP processing, what is this greater plugin count you are referring to?

 

There's something worth noting. While certainly the goal is to get low latency in audio and recording, when exporting/rendering a large project that is either all audio or a convoluted combination in real time, a higher latency is often needed to avoid disk buffer/cache failure or CPU overload. Because I tend to use 48Khz generally, 256 satisfies most needs (third image), with the exception of very large projects.
That said, I can get around 3.5ms at 192Khz if the only goal was to approach zero as close as possible .

 

Generally the interface's driver panel reporting latency are the theoretical mathematical numbers given a host system that will incur 0 excess latency of its own. This is never the case however. ALL host systems will incur their own amounts of latency added to the figures presented by the drivers/setup panels for any given digital audio output to input path.



This is the Real Time Latency display for my RME Fireface UFX USB device. The latency figure on the right shows what the hardware buffers and Sample rates are set to which gives a base latency figure in a perfect world (i.e. a host machine that incurs 0 latency of its own). The readings on the left indicate the actual RTL of the host system AND the audio hardware configuration combined. In this case ~4ms of total real time latency from output to input @96Khz which in the real world is perfectly acceptable.

My point being is the interface is really inconsequential latency wise as long as it's made by a reputable and stable OEM who is known for good solid drivers and quality engineered hardware. The remaining latency bits lie with how well the host system that interface is connected to deals with digital audio.

 

Last edited: Jan 20, 2024

Agreed - There are a pile of additional dependencies. Add into the mix - The power of the host CPU, the amount of RAM, the often overlooked GPU grunt in the mix, the speed of the hard disk/SSD/NVME and the number of processes running at the time of need. Some DAW's also lower the latency a little. The less the computer overall has to work harder compensating in any area, the better it runs in a DAW..that's why the GPU was included.

Additionally, I have found that a 5-pin DIN MIDI cable provides unnoticeable MIDI latency tracking than any USB by far. If Audio tracking for minimal recording monitoring latency is required, using 96Khz gets rid of most of it (if several plugins on recording are required {rarely for me}).

 

Last edited by a moderator: Jan 20, 2024

https://gearspace.com/board/music-c...erface-low-latency-performance-data-base.html

 

Great table, thanks.
I'm surprised how little difference USB vs PCIx makes.

 

@Hazen why GS link? Here's the original site https://original.dawbench.com/index.html

 

Latency freaks be like, "mines' smaller".....

 

Inverse size matters

 

It seems that those are old results from a bunch of years ago. For instance the RME UCX is not the newer II version released July of 2021.

 

Absolutely Part I is from 2011.