Quote:
Originally Posted by
xmein
No you are guessing. "Their latency should be" is not what it is.
I just checked and yet again wasted my precious time trying to find to what you are referring to with that link to their FAQ. It's not your fault for them not having info but why do you post a link to something which doesn't tell me anything?
In their FAQ question they absolutely have ZERO information about latency. ZERO. I being naive idiot opened all questions. No info. But i browsed up trying to find at least something.
What you think about "they should" is not what it could be on top of that you do realize that it's practically impossible to have sub 1ms processing when you are sending from DAW to processing unit then back to DAW. And ASIO buffer DEFINITELY has impact on latency monitoring.
You see exactly that marketing **** which they intentionally put out - is making people post nonsense in forums. It's FPGA so it must be below 1ms. It's FPGA it's superior. It's not DSP it's FPGA...
Why do you think a bridge plugin
which is sending sound signal from your daw to their interface, processing there and then getting back to your DAW wouldn't be affected with your processing buffer - assuming they deliver it to windows which can only be practical under ASIO environment.
Actually I am betting that their marketing nonsense about near zero latency with DAW plugin is actually sole reason why they can't keep up to deliver Windows version. Because it's impossible to have ASIO buffers and at the same time process and monitor below 1ms latency.
Because it's FPGA? Oh really. Give me some reference. How and where did you started to believe this has to be truth? I want some link.
They say at the top of a page "near zero latency" but that can be anything between zero and as much as 8ms if you ask me.
This is really getting hilarious (i mean whole concept of unreliable informations).
Finally some representative finally added up some chart which again is a joke.
NOT ALL LISTED FX ARE AVAILABLE FOR EVERY INTERFACE - but they don't tell you which ones?

What session rate? Is this chart valid for 44khz? 96khz? What?
Do i read this correctly. On all interfaces you can run only 1 like Single 1 reverb in total. What the hell? So that's superiority of FPGA. 1 reverb.
Ok then. I see i am turning to be negative dickhead then so be it. I can't believe that after so much time and negative critics they still refuse to provide exact values any professional even amateur would like to know.
It's like...ahh they want some answers. Ok throw them some bones and let them dig something. We never provided them real info so later on they can't blame us...which is actually what already happened to some people.
Look another nonsense. They proudly headlined "99 instances total" in that chart for what reason? Which when you think twice is a lie! By reading that chart someone could believe that assuming you buy all their plugins you can run 99 instances total on single interface but number 99 is actually number of listed effects in that list. But they can always tell you something like: hey...NOT ALL LISTED FX ARE AVAILABLE FOR EVERY INTERFACE we made that clear.
Why on earth they can't copy UA or even now deceased Powercore DSP chart concept. Where you knew each plugin consumption on every frequency rate. UA make it perfectly clear abut processing delay for EVERY product they have.
Hi!
Thank you very much for your post as it poses a lot of important questions, which I would be more than glad to answer.
Firstly, let me explain why we have no articles and detailed information about latency. With interfaces of this caliber which support both Windows and MAC OS, both via USB and Thunderbolt, you get a lot of variables. Different USB chipsets on different motherboards, different Thunderbolt cards, with each BIOS firmware and motherboard altering a lot of the dependencies, as well as different cables, a wild variety of processors, etc. - all of these things affect the latency and different builds will have different results. If we were doing only USB or only Thunderbolt just on MAC OS for example, we would be able to give estimates for different computers, as the configurations are not that varied. However,
providing estimates would be misleading in our scenario as it might not be achievable on each and every configuration.
Sub-1ms round-trip is advertised for our new Windows Thunderbolt driver. An external tester has achieved it and it is practically possible, I have been able to achieve it as well in my home studio on 192kHz with a 32 sample buffer. Other users have achieved it too and it is pretty realistic on most modern computer configurations -
Antelope Audio Announces Thunderbolt™ Beta Driver for Windows
The FPGA chip introduces less latency when processing effects because of the way it is constructed and the logical operations are processed. With DSP processors you get task after task where with the FPGA you can do multiple tasks at the same time. Let me explain with an example how our FPGA handles the effects processing and an estimate for the latency that is introduced. The internal processing of the signal from the USB/Thunderbolt takes up to 7 samples one way, so 14 samples maximum to get into the FPGA and back to the USB/Thunderbolt (in the longest possible scenario, meaning that most of the time it takes less). One of our compressors will take 2 samples to be processed. If we add another compressor we’ll get 2 more samples introduced, so 4 samples latency is introduced for 2 compressors. If we include the trip to and from the FPGA, we get 18 samples in total. The FPGA processes the signal sample by sample, so when one sample is processed it directly goes out of the FPGA, this is the fastest processing possible with the lowest latency possible on chips built in audio interfaces. Overall, such low latency figures are achievable only with the FPGA when it comes down to processing effects, because of the sample by sample processing. A DSP however can hardly achieve such numbers, because it cannot process sample by sample due to its nature, it takes task after task, and in most scenario it pre-buffers 16/32/64 samples of information to process, this and also when running a lot of things simultaneously is a weak spot which introduces more latency.
As for the effects table - when I posted the table here I said that - “The instance count for the effects doesn't change with the sample rate, so you always get this amount.*“
And yes, you can run a single reverb, since we only have 1 reverb for now and it is designed
only for monitoring purposes via the built in virtual mixer, it is not used as the other effects are and we could run more of it, but again - this reverb is not designed to be used for anything else besides monitoring.
As for the 99 instances total, interfaces like the Zen Tour have all of these effects and can run all of the 99 instances. For some of our interfaces, not all of the effects are released, that’s why there is a notice in the table. You can check which effects are available for which unit here -
https://support.antelopeaudio.com/su...lability-table
This table will be updated with the new interfaces until the end of the week.
A DSP chart is not possible when using FPGA chips, a specific effect doesn’t take a specific amount of resources as it is with the DSP and that’s why we made the table this way, this is the most clear way to understand how many effects you can run, just because you cannot visualise this with percentages, the way these processors work and the logical operations they use are completely different when compared to a DSP and the possibilities of visualising them.
I hope that I have answered most of your questions - if something is left unclear, please message me or reply, I will be glad to answer.