The No.1 Website for Pro Audio
 Search This Thread  Search This Forum  Search Reviews  Search Gear Database  Search Gear for sale  Search Gearslutz Go Advanced
Audio Interface - Low Latency Performance Data Base :
Old 16th March 2017
  #1441
Gear Guru
 
UnderTow's Avatar
Quote:
Originally Posted by DAW PLUS View Post
Note that if you want low latency at 44.1/48kHz with the AVB line, you need to set the buffers to 64/16.
Hi Leon,

I've seen your advice twice now and it has made me curious: Does the 64/16 number mean that you can set input and output buffers separately? Or is it something completely different?

Alistair
Old 16th March 2017
  #1442
Lives for gear
 
TAFKAT's Avatar
 

Thread Starter
Quote:
Originally Posted by UnderTow View Post
Hi Leon,

I've seen your advice twice now and it has made me curious: Does the 64/16 number mean that you can set input and output buffers separately? Or is it something completely different?

Alistair
From my understanding, the second number is the level of safety buffer assigned, which can be dialed in separately.

There is a discussion further back where the 2 parameters were discussed, it took me a while to get my head around it at the time.

Old 16th March 2017
  #1443
Gear Guru
 
UnderTow's Avatar
Quote:
Originally Posted by TAFKAT View Post
From my understanding, the second number is the level of safety buffer assigned, which can be dialed in separately.

There is a discussion further back where the 2 parameters were discussed, it took me a while to get my head around it at the time.

Ah yes, I remember reading about it earlier in this thread. Thanks for reminding me!

Alistair
Old 16th March 2017
  #1444
Lives for gear
 

Quote:
Originally Posted by TAFKAT View Post
From my understanding, the second number is the level of safety buffer assigned, which can be dialed in separately.

There is a discussion further back where the 2 parameters were discussed, it took me a while to get my head around it at the time.

That is correct.
MOTU has two independent settings:

Host Buffer size (the 64).
Host Safety Offset (the 16).

Both can be changed independent from each other.

What is strange is that several of us have discovered that a Safety Buffer of 16, a low value, works better than say a value of 64 or 128. One would assume the higher value would give better results in terms of loading and clicks/pops but that doesn't seem to be the case.
Not sure why.

64/16 seems to be the sweet spot as a few of us have discovered.

Here is the MOTU readme:

Host Buffer Size


When connected to a Windows host, the
Host Buffer Size
menu (Figure 1) is available. This setting determines the amount of latency (delay) you may hear when live audio is patched through your Windows audio software. Smaller buffer sizes produce lower latency, with sizes of 256 samples or less producing virtually imperceptible delay. Many host applications report audio hardware I/O latency, so you can see what happens to the reported latency when making adjustments to this setting.
Be careful with very small buffer sizes, as they can cause performance issues from your host software or PC.
See your MOTU audio interface User Guide for further information about latency.



At sea level, audio travels approximately one foot (30 cm) per millisecond. A latency of ten milliseconds is about the same as being ten feet (three meters) from an audio source.

Host Safety Offset

When connected to a Windows host, the
Host Safety Offset
menu (Figure 1) also becomes available. This setting allows you to fine tune host latency. Larger offsets allow the driver more time to process audio as it transfers to and from the hardware. Lower settings produce lower latency, but if you go too low, your host software may experience performance issues. Generally speaking, 48 samples should serve as a good baseline setting. You can then experiment with lower settings from there. Be mindful, however, when reducing the safety offset, as this parameter can have a significant impact on your computer system’s performance.
Old 19th March 2017
  #1445
Attached are screen pics of TAFKAT's Benchmarks of the Babyface PRO and thethe Audient id22.

How do these numbers translate to my world under my system ?

I've got them both side by side and have been working with the Babyface for a couple of weeks now. As far as a comparison, here goes :

#1 ) Ease of use - Audient by far...way easy.
#2 ) Sound - it's a pick'em really. After switching back and forth using the same project. I can't really pick a winner...you can't go wrong with either one.
#3 ) Performance - Well, there is a very, very small difference on my system-an i76700HQ laptop under Windows 10 running Cubase 9 PRO. Screen pic attached.

There may be desktops where the performance of the RME really shines. I don't know. Maybe there needs to be something "RESET" on my laptop...a USB that is not functioning as it should ? maybe? Giving the babyface PRO the benefit of the doubt. All things being equal, if you go by the benchmarks, the Babyface PRO should be showing a much greater bump in performance imho.

TAFKAT, let me know if you'd prefer this in it's own thread. I delete it and post it separately.
Attached Thumbnails
Audio Interface - Low Latency Performance Data Base :-daw-bench-babyface.png   Audio Interface - Low Latency Performance Data Base :-daw-bench-audient-id22.png   Audio Interface - Low Latency Performance Data Base :-audient-asio-load.png   Audio Interface - Low Latency Performance Data Base :-babyface-asio-load.png  
Old 19th March 2017
  #1446
P99
Gear Maniac
 

Quote:
Originally Posted by eightyeightkeys View Post
Attached are screen pics of TAFKAT's Benchmarks of the Babyface PRO and thethe Audient id22.

How do these numbers translate to my world under my system ?

I've got them both side by side and have been working with the Babyface for a couple of weeks now. As far as a comparison, here goes :

#1 ) Ease of use - Audient by far...way easy.
#2 ) Sound - it's a pick'em really. After switching back and forth using the same project. I can't really pick a winner...you can't go wrong with either one.
#3 ) Performance - Well, there is a very, very small difference on my system-an i76700HQ laptop under Windows 10 running Cubase 9 PRO. Screen pic attached.

There may be desktops where the performance of the RME really shines. I don't know. Maybe there needs to be something "RESET" on my laptop...a USB that is not functioning as it should ? maybe? Giving the babyface PRO the benefit of the doubt. All things being equal, if you go by the benchmarks, the Babyface PRO should be showing a much greater bump in performance imho.

TAFKAT, let me know if you'd prefer this in it's own thread. I delete it and post it separately.
That's the original Babyface, not the Babyface Pro. The Babyface Pro is, from my experience 20%-30% less RTL than the BF. I had bought a Babyface on sale and after a week returned it and bought the Pro. The Pro was, as I said about 20%-30% faster than the BF.
Old 19th March 2017
  #1447
Quote:
Originally Posted by P99 View Post
That's the original Babyface, not the Babyface Pro. The Babyface Pro is, from my experience 20%-30% less RTL than the BF. I had bought a Babyface on sale and after a week returned it and bought the Pro. The Pro was, as I said about 20%-30% faster than the BF.
Your talking about the benchmark ? Yes thanks for the correction.
Old 20th March 2017
  #1448
Lives for gear
 
TAFKAT's Avatar
 

Thread Starter
Quote:
Originally Posted by eightyeightkeys View Post
Attached are screen pics of TAFKAT's Benchmarks of the Babyface PRO and thethe Audient id22.
If you have ASIO Guard On, all you are showing is an ASIO meter reading at 20+ms playback , not really relevant to the charts I am presenting, and/or anyone running a DAW without a Hybrid playback engine. It also has no relevance to the efficiency of the driver once forced to the hardware input/output when track armed/input monitoring. Nor does it show where the actual break point is for the respective drivers.

Comparing ASIO/CPU meters is something I specifically moved away from all those years back, as they don't really represent anything that can be measured empirically, they are more a visual rough estimation.


Last edited by TAFKAT; 20th March 2017 at 04:29 AM..
Old 20th March 2017
  #1449
Lives for gear
 
TS-12's Avatar
Anymore interface tests coming ?

Like in the range between steinberg and better ?

Recently upgraded from motu ultrlite to steinberg ur22. Much much better latency, can enjoy playing software instruments finally.
But want to upgrade to a batter interface anyways with lower latency, maybe some kind of good PCIe interface
Old 20th March 2017
  #1450
P99
Gear Maniac
 

Check out the RME Babyface Pro for some of the lowest latency available.
Old 20th March 2017
  #1451
Quote:
Originally Posted by TAFKAT View Post
If you have ASIO Guard On, all you are showing is an ASIO meter reading at 20+ms playback , not really relevant to the charts I am presenting, and/or anyone running a DAW without a Hybrid playback engine. It also has no relevance to the efficiency of the driver once forced to the hardware input/output when track armed/input monitoring. Nor does it show where the actual break point is for the respective drivers.

Comparing ASIO/CPU meters is something I specifically moved away from all those years back, as they don't really represent anything that can be measured empirically, they are more a visual rough estimation.

O.K. got it.

But, for a composer/producer running audio tracks, plugins and VSTi's (extremely heavy on the VSTi's), the ASIO meter is all that matters because when you've max'ed out, that's it. In practical terms, you're done unless you start freezing instruments and channels which is a bad option when clients are almost always asking for revisions until the end.

So, to carry it further, in part, I used TAFKAT's benchmarks as a basis for the purchase of the RME Babyface Pro, because I believed that I'd get a significant bump in CPU efficiency, Was I wrong to do so and if so, where did I go wrong ?
Old 20th March 2017
  #1452
Here for the gear
 

Quote:
Originally Posted by eightyeightkeys View Post
O.K. got it.

But, for a composer/producer running audio tracks, plugins and VSTi's (extremely heavy on the VSTi's), the ASIO meter is all that matters because when you've max'ed out, that's it. In practical terms, you're done unless you start freezing instruments and channels which is a bad option when clients are almost always asking for revisions until the end.

So, to carry it further, in part, I used TAFKAT's benchmarks as a basis for the purchase of the RME Babyface Pro, because I believed that I'd get a significant bump in CPU efficiency, Was I wrong to do so and if so, where did I go wrong ?

I agree. This is probably the most important thing to know for a producer/mixer that works mainly inside the box and that works on songs from the beginning to the end (from production to mixing/mastering). Would be great to know how these interface actually perform in this regard.
Old 20th March 2017
  #1453
Lives for gear
 
Bstapper's Avatar
 

Quote:
Originally Posted by eightyeightkeys View Post
O.K. got it.

But, for a composer/producer running audio tracks, plugins and VSTi's (extremely heavy on the VSTi's), the ASIO meter is all that matters because when you've max'ed out, that's it. In practical terms, you're done unless you start freezing instruments and channels which is a bad option when clients are almost always asking for revisions until the end.

So, to carry it further, in part, I used TAFKAT's benchmarks as a basis for the purchase of the RME Babyface Pro, because I believed that I'd get a significant bump in CPU efficiency, Was I wrong to do so and if so, where did I go wrong ?
You didn't go wrong. You used the chart and chose the device that has better latency and will perform more efficiently; especially as you reach the upper end of your system's capabilities.

I am not sure what you expected beyond that - but good work you figured it out...

Cheers,
Brock
Old 20th March 2017
  #1454
Lives for gear
 
TAFKAT's Avatar
 

Thread Starter
Quote:
Originally Posted by eightyeightkeys View Post
O.K. got it.

But, for a composer/producer running audio tracks, plugins and VSTi's (extremely heavy on the VSTi's), the ASIO meter is all that matters because when you've max'ed out, that's it. In practical terms, you're done unless you start freezing instruments and channels which is a bad option when clients are almost always asking for revisions until the end.
The ASIO Meter is a reference of the overall audio engine efficiency , but its not always an accurate reference of when the driver will break up. Some drivers will break up at 50-60% ASIO load for example, some will push past the red and continue without breakup.

ASIO Guard mask some of the performance issues of the respective drivers because of the high playback latency. There are instances where end users cannot use ASIO Guard - VSL, VEP for example, so any benefit of the looser playback latency is negated. It then comes back to the actual efficiency of the driver at the respective hardware latency. In those instances the efficiency of the driver is paramount , which is what is represented in the charts. As I noted earlier, that is also relevant for all users who use a DAW without a hybrid playback engine- Ableton, Sonar, Reaper, StudioOne, etc.

Quote:
So, to carry it further, in part, I used TAFKAT's benchmarks as a basis for the purchase of the RME Babyface Pro, because I believed that I'd get a significant bump in CPU efficiency, Was I wrong to do so and if so, where did I go wrong ?
I think you are confusing the ASIO Meter as a CPU Meter, it doesn't represent CPU efficiency at all, maybe we are getting lost in semantics.

As I noted earlier, ASIO guard does mask some of the driver issues, the devil will be in the details.What you have gained is superior low latency drivers at any given working latency for playing in realtime and RTL, as well in your instance with C9 and ASIO guard, when dropping in, overdubbing later in the session.

Another thing to note is - what actual hardware latency have you set on the respective interfaces when playing live - not the panel label listing, the actual delivered latencies of the respective interfaces ?


Last edited by TAFKAT; 21st March 2017 at 02:52 AM..
Old 21st March 2017
  #1455
Quote:
Originally Posted by TAFKAT View Post
Another thing to note is - what actual hardware latency have you set on the respective interfaces when playing live - not the panel label listing, the actual delivered latencies of the respective interfaces ?

You've got me there....not sure I understand the difference. The latency can be set from within the RME control panel or within Cubase, but, they are redundant.

Where would I find the "actual hardware latency" ? versus the one set within Cubase itself ?
Old 21st March 2017
  #1456
Lives for gear
 
TAFKAT's Avatar
 

Thread Starter
Quote:
Originally Posted by eightyeightkeys View Post
You've got me there....not sure I understand the difference. The latency can be set from within the RME control panel or within Cubase, but, they are redundant.

Where would I find the "actual hardware latency" ? versus the one set within Cubase itself ?
Panel settings - i.e, 064/128/256, etc, are just nominal numbers , the actual reported/delivered latency varies significantly from interface to interface. That is what I have listed in the charts - Reported I/O is listed in Cubase, RTL is the actual measured Round Trip Latency. Some interfaces do not report AD/DA for example.

Example:
RME Babyface @ 128 : Reported In 4.150/ Out 4.331 : RTL 8.481
Audient iD22 @128 : Reported In 3.900/ Out 3.900 : RTL 10.380 .

On the Audient there is a 2.58 ms discrepancy due to them not reporting AD/DA and safety buffers, so lets say its approx 1.3ms respectively, so the playback latency is actually 5.2 ms.

Sounding like a broken record, *Latency is only 1/2 the equation, the other 1/2 is the actual efficiency of the driver at the respective playback latency*, which is where the RME has the greater advantage.

*I need to make that my signature.

Old 22nd March 2017
  #1457
Lives for gear
 

So what's the beef with TB vs USB on the UFX+? Percentages wouldn't enable viewers who haven't swallowed the red pill to get confused!
Old 23rd March 2017
  #1458
Lives for gear
 
TS-12's Avatar
Dear thread starter.
It would be great pleasure if you added popular interfaces like babyface pro, and UAD apollo series, it would be very useful. Especially for me since I'm on the search for low latency and good conversion interface right now.
Old 23rd March 2017
  #1459
Quote:
Originally Posted by TS-12 View Post
Dear thread starter.
It would be great pleasure if you added popular interfaces like babyface pro, and UAD apollo series, it would be very useful. Especially for me since I'm on the search for low latency and good conversion interface right now.
The benchmark system is not Thunderbolt compatible.
The Babyface Pro drivers are the same as the UCX USB drivers.
Old 23rd March 2017
  #1460
Lives for gear
 
TS-12's Avatar
Quote:
Originally Posted by DAW PLUS View Post
The benchmark system is not Thunderbolt compatible.
The Babyface Pro drivers are the same as the UCX USB drivers.
there is USB version Apollo.
thats the one im considering since im on PC.
(debating between BabyFace Pro or UAD Apollo USB)
Old 23rd March 2017
  #1461
Lives for gear
 
TAFKAT's Avatar
 

Thread Starter
Quote:
Originally Posted by TS-12 View Post
there is USB version Apollo.
thats the one im considering since im on PC.
(debating between BabyFace Pro or UAD Apollo USB)
UAD Apollo Twin results were posted in January Here

Old 23rd March 2017
  #1462
Lives for gear
 
TS-12's Avatar
Quote:
Originally Posted by TAFKAT View Post
UAD Apollo Twin results were posted in January Here

Very much thank you.


So apollo better monitor mixer, ability to run UAD plugins.
But Babyface pro better latency
Old 23rd March 2017
  #1463
Lives for gear
Quote:
Originally Posted by TS-12 View Post
But Babyface pro better latency
Yes. Check any RME USB 2 interface already tested since they use the same unified driver. That info was also posted in this thread somewhere.
Old 24th March 2017
  #1464
P99
Gear Maniac
 

Quote:
Originally Posted by Alndln View Post
Yes. Check any RME USB 2 interface already tested since they use the same unified driver. That info was also posted in this thread somewhere.
Babyface PRO even better by 20%-30% than Babyface... Just my experience after owning both.
Old 24th March 2017
  #1465
Lives for gear
Quote:
Originally Posted by P99 View Post
Babyface PRO even better by 20%-30% than Babyface... Just my experience after owning both.
You tested them both properly? Same OS? I would be surprised if there was any difference because I just checked and it's the same driver, same download.
Old 24th March 2017
  #1466
P99
Gear Maniac
 

Quote:
Originally Posted by Alndln View Post
You tested them both properly? Same OS? I would be surprised if there was any difference because I just checked and it's the same driver, same download.
I am not an expert, but yes I did try them both on the exact same system and software... everything. The only reading I get is from Studio One Pro. I cannot remember the exact numbers, but for sure it was definitely milliseconds faster.

I had mentioned this to Tafkat and he said it was because the BF Pro used newer stuff and the better driver.

I will post his response shortly when I did it up out of this thread.

EDIT: Here is his response to me about the Babyface Pro:

I haven't tested the Babyface Pro as yet , benchmark results will be on par to the any of the RME UCX USB2 results posted as its a universal driver, so not exact but close enough for a reference , I'll confirm I/O RTL's when next in front of one.
https://www.gearslutz.com/board/12483771-post1398.html


Here is where Tafkat quickly responds to the differences between the BF and BF Pro:

Yep, the Babyface Pro has later gen/lower latency converters, so I/O and RTL will be lower than the first Gen Babyface.
https://www.gearslutz.com/board/12484019-post1400.html
Old 27th March 2017
  #1467
Gear Addict
 
firubbi's Avatar
Quote:
Originally Posted by captain caveman View Post
So what's the beef with TB vs USB on the UFX+? Percentages wouldn't enable viewers who haven't swallowed the red pill to get confused!
TB would be same or close to Pcie I guess. How good is The UFX+ compare to Aes32?
Old 28th March 2017
  #1468
Lives for gear
 
TS-12's Avatar
Old 28th March 2017
  #1469
Lives for gear
 

Quote:
Originally Posted by firubbi View Post
TB would be same or close to Pcie I guess. How good is The UFX+ compare to Aes32?
That's the assumption everyone is going by, but TAFKAT has experienced different results...

"With the UFX + characteristics of the TB driver is more akin to the USB2/3 than PCIe."
Old 28th March 2017
  #1470
Lives for gear
 
TAFKAT's Avatar
 

Thread Starter
Quote:
Originally Posted by captain caveman View Post
That's the assumption everyone is going by, but TAFKAT has experienced different results...

"With the UFX + characteristics of the TB driver is more akin to the USB2/3 than PCIe."
The UFX+ under USB3/TB comparatively is much the same as what we have seen with the UCX/UFX, etc units in comparison to the HDSPe PCIe units.

Its not that the performance is bad in any way, its simply that TB is not behaving in the same manner as PCIe under extreme loadings.

TB performance equaling PCIe has been accepted, so to speak, because its technically external PCIex4, but there are some additional curves being navigated because of the actual TB controller IMO.

Once that is navigated, there is still the dedicated drivers and controllers of the respective interfaces that will be a contributing factor.

What I am saying is simply is don't take it for granted that TB is delivering PCIe level performance, most manufacturers do not even have the capability to be able to qualify that statement as they have no PCIe experience to even compare to.

I'll do my best to test more TB interfaces as they become available to me.


Last edited by TAFKAT; 28th March 2017 at 10:31 PM..
Loading mentioned products ...
New Reply Submit Thread to Facebook Facebook  Submit Thread to Twitter Twitter  Submit Thread to LinkedIn LinkedIn  Submit Thread to Google+ Google+  Submit Thread to Reddit Reddit 
 
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
Similar Threads
Thread
Thread Starter / Forum
Replies
Chieftain Jake / Low End Theory
14
digirira / Low End Theory
7

Forum Jump