The No.1 Website for Pro Audio
 All  This Thread  Reviews  Gear Database  Gear for sale     Latest  Trending
Audio Interface - Low Latency Performance Data Base
Old 3rd July 2019
  #3301
Gear Maniac
 
Dewdman42's Avatar
 

Did you measure rtl through loopback on all major daws? What they report is not always accurate
Old 3rd July 2019
  #3302
Lives for gear
 
Gomjab's Avatar
 

I received my MOTU 828es and as promised here are latency numbers on my iMac Pro at 44.1, 48, and 96
Attached Thumbnails
Audio Interface - Low Latency Performance Data Base-2dab7997-81e7-49c6-ada5-75fcba2f9798.jpg   Audio Interface - Low Latency Performance Data Base-3909d8d6-9e34-437b-a62d-0b5b73bdd037.jpg   Audio Interface - Low Latency Performance Data Base-1ade6424-4704-419b-bc58-386deb38d676.jpg  
Old 3rd July 2019
  #3303
Lives for gear
 
Gomjab's Avatar
 

And here is my old Quantum on my iMac Pro at 44.1, 48, and 96.

No doubt the Quantum is the winner in the latency department but I knew that going in. The features on the MOTU 828es are fantastic. I paired it with a 24ai over Ethernet using AVB to give me 32 balanced TRS inputs into the internal mixer/routing matrix.

It is interesting that the MOTU is not accurately reporting the RTL. The Quantum’s reported RTL was always pretty close to measured.
Attached Thumbnails
Audio Interface - Low Latency Performance Data Base-3da76bb6-4aec-4750-b843-8f1851a2e0e6.jpg   Audio Interface - Low Latency Performance Data Base-8e760c94-a5be-4128-ae9b-5b2a652f2bfa.jpg   Audio Interface - Low Latency Performance Data Base-26c30850-c510-442e-a1ee-250afe000956.jpg  
Old 3rd July 2019
  #3304
Gear Maniac
 
Dewdman42's Avatar
 

They're all great results, frankly I think that is about as good as it gets, you might get marginally better results from an RME or Lynx PCIe solution, but its splitting hairs. Enjoy your new setup!!
Old 3rd July 2019
  #3305
Gear Maniac
 

Quote:
Originally Posted by Dewdman42 View Post
Did you measure rtl through loopback on all major daws? What they report is not always accurate
Logic for instance seems to report a higher latency than pretty much any other DAW, and the latency that it's telling me I'm getting never feels like it actually is that number. 5ms in reaper feels like 5ms, 5ms in logic feels more like 12ms. I bet the new focusrite units can go lower than with other DAW's.
Old 3rd July 2019
  #3306
Gear Maniac
 
Dewdman42's Avatar
 

Here are screenshots from LogicPro 10.4.5, Cubase 10.0.3, StudioOne 4.5.1 and AULab. They all report the same thing.

Generally DAW's report whatever the audio driver reports. In this case the audio driver is reporting 144 samples of latency in each direction.

Now the question is whether that reported latency matches reality... You can test the driver itself by using the RTL utility. It is not a given that the audio driver is reporting the correct latency, especially for more complex setups.

On top of that, If you want to see whether each DAW is producing the actual latency that is reported, then you have to create your own loopback test in the DAW itself.

Can you tell us a little bit about the loopback test you are using with LPX to determine that its not achieving the reported latency?
Attached Thumbnails
Audio Interface - Low Latency Performance Data Base-cubase.jpg   Audio Interface - Low Latency Performance Data Base-s1.jpg   Audio Interface - Low Latency Performance Data Base-lpx.jpg   Audio Interface - Low Latency Performance Data Base-aulab.jpg  
Old 3rd July 2019
  #3307
Lives for gear
 
Gomjab's Avatar
 

One more set of measurements for the new setup. For those that are interested in MOTU’s AVB implementation. I repeated my RTL tests but this time took the output from the 828es and patched it into an input on the 24ai module which is connected to the 828es via Ethernet.

Using AVB to connect to a computer is worse than using thunderbolt due to the networking stack on the computer however the latency between two MOTU modules connected with AVB is very good.

Adding the 24ai AVB connection only added 0.363 ms of latency at 44.1 buffer size 64.

Here are the results.

iMac Pro ——TB-—> 828eS ——Audio Patch Cable——> 24ai ==AVB==> 828es —TB—> iMac Pro
Attached Thumbnails
Audio Interface - Low Latency Performance Data Base-3217b7ff-bdf0-420c-83e0-c504bc6c9203.jpg   Audio Interface - Low Latency Performance Data Base-7df2fdd8-61cd-451d-aec5-cf4a4b2f78be.jpg   Audio Interface - Low Latency Performance Data Base-8f26099f-16c7-4178-a791-88ac2a3f857d.jpg  
Old 3rd July 2019
  #3308
Gear Maniac
 
Dewdman42's Avatar
 

That is very similar to what happens with my X32-rack over AES50.
Old 3rd July 2019
  #3309
Gear Maniac
 
Dewdman42's Avatar
 

By the way all this RTL testing is not without purpose. When the reported latency in your daw does not match reality, which is probably the case when using AVB or AES50 to connect devices into one virtual DAW device, then it becomes necessary to make adjustments in your daw so that all recording offsets will be correct.

For example if daw reports 6ms but actual latency is 6.4ms, then the record offset needs to be adjusted by .4ms in order to ensure completely correct timing of newly recorded tracks and plugin delay compensation, etc..
Old 3rd July 2019
  #3310
Lives for gear
 
Gomjab's Avatar
 

Quote:
Originally Posted by Dewdman42 View Post
By the way all this RTL testing is not without purpose. When the reported latency in your daw does not match reality, which is probably the case when using AVB or AES50 to connect devices into one virtual DAW device, then it becomes necessary to make adjustments in your daw so that all recording offsets will be correct.

For example if daw reports 6ms but actual latency is 6.4ms, then the record offset needs to be adjusted by .4ms in order to ensure completely correct timing of newly recorded tracks and plugin delay compensation, etc..
Exactly! My DAW has no clue what happens once the bits leave over Thunderbolt. The RTL utility complained on the lower buffers due to the mismatch between reported and measured.

Bottom line is my setup now gives me the flexibility I want and will be under 5ms for the vast majority of my use cases and even lower if I push it.

Thanks again to TAFKAT and all the people contributing to this topic! It really helped when I was looking to update my hardware.

Old 3rd July 2019
  #3311
Gear Guru
 
monkeyxx's Avatar
Does anyone know when you "ping" your External Effects plugin wrapper in Cubase to detect the latency. Is that a pretty accuarate number that comes back?

It seems phase-accurate when running outboard processors in parallel to ITB tracks, so I wonder if this is a quick/easy way for me to measure system latency.
Old 3rd July 2019
  #3312
Lives for gear
Quote:
Originally Posted by monkeyxx View Post
Does anyone know when you "ping" your External Effects plugin wrapper in Cubase to detect the latency. Is that a pretty accuarate number that comes back?

It seems phase-accurate when running outboard processors in parallel to ITB tracks, so I wonder if this is a quick/easy way for me to measure system latency.
I don't know, how accurate is current Cubase plugin, I've never tested that, but for example similar plugin in Reaper is pretty accurate with regards to detection. But it wasn't always like that, because in previous Reaper versions (like 3y back) it was significantly improved.
Before Oblique made its RTL Utility available for Mac, I've recommended Reaper as good alternative method for semi automatic testing.

Just keep in mind, that some insert plugins shows absolute roundtrip latency and others shows relative offset to what's reported by interface driver.
Of course for sake of comparison, absolute latency is necessary. In Reaper there is checkbox for both ways. Not sure, how that's in Cubase.

Best, you can do, is to obtain some ping measurements, and verify it by Oblique tool.
Or always reliable good old manual detection with transient source, record track and manual range selection with length readout in samples, but for that it's necessary to verify and defeat any DAW compensation based on reported latency from interface driver.

Last (likely obvious) thing with measurements in DAW is to align input and output levels and keep some headroom.. sometimes people report weird detected figures, because they do analog loopback at interface with adjustable input sensitivity and forget to set levels right.. so ping signal returns clipped or so weak and buried in noise, that it can't really work.

Michal

Last edited by msmucr; 3rd July 2019 at 11:40 PM..
Old 3rd July 2019
  #3313
Quote:
Originally Posted by Gomjab View Post
And here is my old Quantum on my iMac Pro at 44.1, 48, and 96.

No doubt the Quantum is the winner in the latency department but I knew that going in. The features on the MOTU 828es are fantastic. I paired it with a 24ai over Ethernet using AVB to give me 32 balanced TRS inputs into the internal mixer/routing matrix.

It is interesting that the MOTU is not accurately reporting the RTL. The Quantum’s reported RTL was always pretty close to measured.
These results look really good. I would be happy with that setup. Alas Motu delivered! Glad it is working for you
Old 4th July 2019
  #3314
Here for the gear
 

The trivial method for measuring RTL through the DAW is to use a 2 in/2 out interface, connect some source to the left input, connect the left output to the right input and leave the right output unconnected.
Now record as a stereo track.
Use the range selection to measure the difference between left and right recorded channels.
The difference is the true RTL through the DAW with any latency compensation defeated.
Old 4th July 2019
  #3315
Gear Maniac
 
Dewdman42's Avatar
 

https://ask.video/video/logic9207/19...-loopback-test

Last edited by Dewdman42; 4th July 2019 at 06:41 AM..
Old 4th July 2019
  #3316
Gear Maniac
 
Dewdman42's Avatar
 

So, there are a few things to think about. First there is the RTL test without using any DAW. That will tell you the actual latency, regardless of what all the DAW's are reporting. if they are the same, then you're good. If they are different, then you know your audio device driver is inaccurate by some number of samples.

Write this down.

Some audio drivers are simply not accurate. But usually it has more to do with the fact that you're mixing and matching multiple devices, perhaps some of them are expansions via ADAT, AES, AES50, AVB, etc. The audio driver doesn't know about that extra latency, so it won't be included it in the added up number reported to the DAW, and to you. It can definitely matter.

Alright once you know that we turn to your DAW. when you record tracks into your DAW, your DAW knows the latencies reported by your soundcard. Assuming they are correct. So when incoming audio comes in, your DAW will conclude that there was so many ms of input latency and that the sound hit the input on your sound card that many milliseconds before when it actually received it as digital data, due to the latency, and will then register that newly recorded audio onto a region, exactly that many milliseconds earlier then it actually received it. In this way, it ends up on the track exactly where you intended it to be when you used your ears while recording.

That is all good.

But if the audio driver is reporting the incorrect latency, then the stuff you're recording could be getting registered to the track regions early or late because of that incorrect latency report. That is why most DAW's will have an offset parameter that you can use to adjust for this. if you know that your actual RTL is .5ms longer then reported by your DAW, then you need to nudge the offset value by that many samples so that events will be registered exactly where they are supposed to be when you record. After making that adjustment the above test should record the loopback test exactly to the sample on the same point of the timeline as the source.

This will not get rid of the delay you hear while monitoring through the DAW. It will just make your DAW smarter about how to record audio events to the timeline.

You can determine some of that by comparing the RTL reported latency vs actual. But perhaps a more sure fire method is to use your DAW just in case the DAW itself adds any more latency through some extra buffer, which is always a possibility.

The above method as described, and in the video for logic, would work with any DAW, but its also a bit the hard way. in LogicPro you can use the I/O plugin to measure it, for example. this can help you set the record offset parameter correctly so that your recordings will line up properly on the timeline.



The general RTL utility is interesting, IMHO, more to have an idea of how much actual latency you will be experiencing. If you know you can't live with more than say 8ms of latency, you can find out for sure exactly how much latency is being added going through your sound card, regardless of what is actually reported, and set the buffer accordingly. If you have ASIO Guard turned on in Cubase for example, your latency will be higher then reported by the sound card, though in that case Cubase is probably taking it into account, so an offset adjustment may not be needed, but you might be dealing with 30ms, for example, of latency instead of the 8ms that your sound card is reporting.

As I said, you can also compare actual vs reported to find out if you need a record offset, but I prefer to test that in the DAW to be absolutely sure I'm making up for anything the DAW might be doing to the audio signal to get different latency then what is reported.

Last edited by Dewdman42; 4th July 2019 at 06:45 AM..
Old 4th July 2019
  #3317
Quote:
Originally Posted by Gomjab View Post
It is interesting that the MOTU is not accurately reporting the RTL. The Quantum’s reported RTL was always pretty close to measured.
Could it be that it reports digital loopback and you are measuring analog loopback, i.e. the difference is the conversion latency? For each sample rate, the value stays identical independent of the buffer size, and changes with the sample rate.


Quote:
Originally Posted by Dewdman42 View Post
By the way all this RTL testing is not without purpose. When the reported latency in your daw does not match reality, which is probably the case when using AVB or AES50 to connect devices into one virtual DAW device, then it becomes necessary to make adjustments in your daw so that all recording offsets will be correct.

For example if daw reports 6ms but actual latency is 6.4ms, then the record offset needs to be adjusted by .4ms in order to ensure completely correct timing of newly recorded tracks and plugin delay compensation, etc..
If you record all tracks with the same buffer, the offset will always be the same. You may want to correct only for very tight timing on the grid.


Quote:
Originally Posted by Drumfix View Post
The trivial method for measuring RTL through the DAW is to use a 2 in/2 out interface, connect some source to the left input, connect the left output to the right input and leave the right output unconnected.
Now record as a stereo track.
Use the range selection to measure the difference between left and right recorded channels.
The difference is the true RTL through the DAW with any latency compensation defeated.
This is not reliable as most hosts have an automatic latency correction pulling the recording to the correct spot, meaning the latency is corrected. Then again, in my measurements this has not been perfect so using RTL Utility is the solid way.
Reminds me I should look at the stability of that feature again, but it is not a priority for me.

Last edited by DAW PLUS; 4th July 2019 at 03:54 PM..
Old 4th July 2019
  #3318
Lives for gear
 

DAW Recording Latency Compensation compensates for Output Latency+Input Latency (i.e. a Roundtrip), not just Input Latency.
Old 4th July 2019
  #3319
Gear Maniac
 
Dewdman42's Avatar
 

The latency reported by daws is not measured, the audio driver will have some hard coded values and then it’s calculated based on buffer size.

The daw can only place the Audio at the correct spot if the calculated latency of input and output is accurate. When you use the method shown in the first video I posted you will find out exactly how many samples off it is and after adjusting the record delay parameter of the daw another test will be spot on.
Old 4th July 2019
  #3320
Quote:
Originally Posted by DanRand View Post
DAW Recording Latency Compensation compensates for Output Latency+Input Latency (i.e. a Roundtrip), not just Input Latency.
Yes, correct, my bad. A glitch in my matrix...corrected in my post.
Old 4th July 2019
  #3321
Here for the gear
 

Quote:
Originally Posted by DAW PLUS View Post
This is not reliable as most hosts have an automatic latency correction pulling the recording to the correct spot, meaning the latency is corrected.
Wrong, this is absolutely reliable and works with any DAW, no matter what latency compensation it uses.
Old 5th July 2019
  #3322
Quote:
Originally Posted by Drumfix View Post
Wrong, this is absolutely reliable and works with any DAW, no matter what latency compensation it uses.
Interesting, as you need to switch off the ASIO latency compensation (not the plugin latency compensation) to get the real offset. If it is on, if the drivers report well and the compensation works ideal, there is no offset.
However, I have not been able to measure remotely correct offsets which comply with the actual driver latency (with and without AD/DA). It could have been an error in the host but that is what I mean with unreliable. I have not tried this since then as RTL Utility already worked great and fast.
Old 5th July 2019
  #3323
Lives for gear
 

Drumfix's method doesn't rely on accurate reporting of the actual latency by the driver.

He's looking at the relative distance between two recordings* that have gone through the same amount of DAW Recording Latency Compensation, which will be the same amount of compensation for both recordings.

The second recording (edit*: right side of the stereo track) will have gone through a Roundtrip more latency than the first recording. By comparing the difference between the two recordings (edit: left vs right) in a sample editor, you get the exact roundtrip latency in samples.



* Drumfix is recording the test to a stereo track, but the same principle applies

Last edited by DanRand; 5th July 2019 at 09:08 PM..
Old 5th July 2019
  #3324
Quote:
Originally Posted by DanRand View Post
Drumfix's method doesn't rely on accurate reporting of the actual latency by the driver.

He's looking at the relative distance between two recordings that have gone through the same amount of DAW Recording Latency Compensation, which will be the same amount of compensation for both recordings.

The second recording will have gone through a Roundtrip more latency than the first recording. By comparing the difference between the two recordings in a sample editor, you get the exact roundtrip latency in samples.
Duh, thanks for explaining. I missed the "connect some source to the left input," part.
Yes, I absolutely agree, this will work.

I think my head has some latency this week...I will go to the bar now and fix it.
Old 5th July 2019
  #3325
Lives for gear
 
TAFKAT's Avatar
 

Quote:
Originally Posted by DAW PLUS View Post

I think my head has some latency this week...I will go to the bar now and fix it.
10, 9,..9, 8,..8,...8, ..... , in joke... :-)

Old 7th July 2019
  #3326
Here for the gear
Signed up to show my gratitude for this thread, an excellent wealth of knowledge here. I ended up with the Quantum 2, a lovely piece of hardware.
Old 8th July 2019
  #3327
Gear Addict
 
MontyMakesMusic's Avatar
 

Quote:
Originally Posted by dkg_uk View Post
Signed up to show my gratitude for this thread, an excellent wealth of knowledge here. I ended up with the Quantum 2, a lovely piece of hardware.
+1.
I ended up with the quantum. Absolutely blistering latency.
Old 8th July 2019
  #3328
Lives for gear
 
daskeladden's Avatar
 

I saw that the Steinberg AXR4 Thunderbolt 2 audio interface now have Windows drivers. Anybody planning to test it on Windows?

dkg_uk and MontyMakesMusic Quantum is a great choice. You on Mac or Pc?
Old 8th July 2019
  #3329
Here for the gear
Quote:
Originally Posted by daskeladden View Post
Quantum is a great choice. You on Mac or Pc?
PC
More Info
Old 8th July 2019
  #3330
Lives for gear
 

Quote:
Originally Posted by daskeladden View Post
I saw that the Steinberg AXR4 Thunderbolt 2 audio interface now have Windows drivers. Anybody planning to test it on Windows?
I had one up and running last week, so it definitely works fine. Didn't get a chance to pull benches at the time through.
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump
Forum Jump