Gearslutz

Gearslutz (https://www.gearslutz.com/board/)
-   Music Computers (https://www.gearslutz.com/board/music-computers/)
-   -   Audio Interface - Low Latency Performance Data Base (https://www.gearslutz.com/board/music-computers/618474-audio-interface-low-latency-performance-data-base.html)

DAW PLUS 30th September 2019 09:33 AM

Quote:

Originally Posted by ponzi (Post 14233285)
I said 'can do' based on the specs of my 828es as an example, I did not say that was the upper limit which would read 'can do no more than'.

With all respect, vague figures and guesses don't contribute to clearness about this already confusing topic, nor does factual I/O numbers of mid sized interfaces explain anything.
A Steinberg UR22 "can do" 2 in 2 out but that does not tell us anything about USB2 nor about USB3 capabilities. It just sounds limiting, which is confusing.

thehightenor 3rd October 2019 08:12 AM

So let me get this straight ....according to this excellent data base my 15 year old RME PCIe host card (to which I connect my Crane Song converters via AES) is still the best option going in terms of low latency and stability!

It looks like anything I buy other than using what I already have is a backward step!! .... thank goodness RME continue to support old hardware!

Dallon426 3rd October 2019 08:18 AM

Quote:

Originally Posted by thehightenor (Post 14243750)
So let me get this straight ....according to this excellent data base my 15 year old RME PCIe host card (to which I connect my Crane Song converters via AES) is still the best option going in terms of low latency and stability!

It looks like anything I buy other than using what I already have is a backward step!! .... thank goodness RME continue to support old hardware!

If it ain't broke.... Don't fix it

thehightenor 3rd October 2019 09:11 AM

Quote:

Originally Posted by Dallon426 (Post 14243757)
If it ain't broke.... Don't fix it

True.

I think you're right!

misa 10th October 2019 10:38 PM

Someone suggested I post my little study on this thread, so here it is in case it helps anyone:

After reformatting an aging Mid-2009 MacBook Pro with OSX 10.11.6 El Capitan running a 2.8 Intel Core 2 Duo and 8GB RAM, I decided to test out RTL using the Mixbus 32C-5 Latency Measurement Tool between a newly arrived Arturia AudioFuse 8pre and my trusty Apogee Duet 2.

Here are the results for the 8pre* in # Samples - Sample Rate (Hz) / Round Trip Latency (ms)
32 Samples - 48k/5.438 96k/3.688
64 Samples - 48k/6.646 96k/4.354
128 Samples - 48k/9.188 96k/5.688
256 Samples - 48k/14.646 96k/8.354
512 Samples - 48k/25.062 96k/13.688
1024 Samples - 48k/46.771 96k/24.354
2048 Samples - 48k/89.312 96k/45.688
4096 Samples - 48k/174.396 96k/88.354

Here are the results for the Duet 2 in # Samples - Sample Rate (Hz) / Round Trip Latency (ms)
32 Samples - 48k/4.104 96k/3.531
64 Samples - 48k/5.438 96k/4.198
128 Samples - 48k/8.104 96k/5.531
256 Samples - 48k/13.438 96k/8.198
512 Samples - 48k/24.104 96k/13.531
1024 Samples - 48k/45.438 96k/24.198
2048 Samples - 48k/88.104 96k/45.531
4096 Samples - 48k/173.438 96k/88.198

*The 8pre exhibited highly variable results per test instance. For example, running the tool multiple times using 32 samples at 48kHz resulted in all of the following results: 5.292, 4.983, 5.167, 5.417, 5.158, 5.438. I ended up running the test multiple times per sample rate and judging what seemed to be the most common latency result. On the other hand, the results on the Duet 2 were very stable and did not deviate. I’m not sure what this points to, but I’d feel a bit more confident with the Apogee. Both were running using Core Audio.

Noctambulant 11th October 2019 05:42 AM

Quote:

Originally Posted by misa (Post 14258358)
Someone suggested I post my little study on this thread, so here it is in case it helps anyone:

After reformatting an aging Mid-2009 MacBook Pro with OSX 10.11.6 El Capitan running a 2.8 Intel Core 2 Duo and 8GB RAM, I decided to test out RTL using the Mixbus 32C-5 Latency Measurement Tool between a newly arrived Arturia AudioFuse 8pre and my trusty Apogee Duet 2.

Here are the results for the 8pre* in # Samples - Sample Rate (Hz) / Round Trip Latency (ms)
32 Samples - 48k/5.438 96k/3.688
64 Samples - 48k/6.646 96k/4.354
128 Samples - 48k/9.188 96k/5.688
256 Samples - 48k/14.646 96k/8.354
512 Samples - 48k/25.062 96k/13.688
1024 Samples - 48k/46.771 96k/24.354
2048 Samples - 48k/89.312 96k/45.688
4096 Samples - 48k/174.396 96k/88.354

Here are the results for the Duet 2 in # Samples - Sample Rate (Hz) / Round Trip Latency (ms)
32 Samples - 48k/4.104 96k/3.531
64 Samples - 48k/5.438 96k/4.198
128 Samples - 48k/8.104 96k/5.531
256 Samples - 48k/13.438 96k/8.198
512 Samples - 48k/24.104 96k/13.531
1024 Samples - 48k/45.438 96k/24.198
2048 Samples - 48k/88.104 96k/45.531
4096 Samples - 48k/173.438 96k/88.198

*The 8pre exhibited highly variable results per test instance. For example, running the tool multiple times using 32 samples at 48kHz resulted in all of the following results: 5.292, 4.983, 5.167, 5.417, 5.158, 5.438. I ended up running the test multiple times per sample rate and judging what seemed to be the most common latency result. On the other hand, the results on the Duet 2 were very stable and did not deviate. I’m not sure what this points to, but I’d feel a bit more confident with the Apogee. Both were running using Core Audio.

Does the 8 pre overheat like the og Audiofuse?

misa 11th October 2019 02:42 PM

Quote:

Originally Posted by Noctambulant (Post 14258860)
Does the 8 pre overheat like the og Audiofuse?

I can’t say I’ve put it into real heavy use yet, but being on for several hours, I didn’t notice any heat.

I’m sure the spread out rack form factor would help with heat compared to the super compact original.

Dooberknob 12th October 2019 03:03 PM

Does thunderbolt offer any advantages in latency of USB?

I've read that as far as latency is concerned, USB3 doesn't really offer any advantages over USB2, and I was wondering if it was the same deal with thunderbolt. I understand there are many other variables that would affect the RTL of an audio interface other than the connector, like processing power, drivers ect., but all things being equal, would thunderbolt have the potential for lower latency?

12Bass 12th October 2019 03:37 PM

Quote:

Originally Posted by Dooberknob (Post 14261596)
Does thunderbolt offer any advantages in latency of USB?

I've read that as far as latency is concerned, USB3 doesn't really offer any advantages over USB2, and I was wondering if it was the same deal with thunderbolt. I understand there are many other variables that would affect the RTL of an audio interface other than the connector, like processing power, drivers ect., but all things being equal, would thunderbolt have the potential for lower latency?

In principle, yes, as Thunderbolt is essentially PCIe over a cable. However, in practice, it depends. Take at look at the LLP database and you'll see some excellent TB performers and a few that are beaten by the better USB interfaces. It's best to look at them on a case-by-case basis.

GMSweet 14th October 2019 01:32 PM

Here are my results from testing my new Audient iD44.

System: Intel i7-8700K @ 3.7GHz on an Asus PrimeZ370A motherboard w/ 32GB RAM running Windows 10
Connection: USB3 controller on motherboard (even though the iD44 is a USB2 controller)

Personal commentary: I typically record at 24/96K. I'm not sure how the math works out, but I was expecting the iD44 to perform worse than the Saffire Pro 24DSP it replaced based on the charts here. I was surprised to see that if I was still running at 24/48K like I used to, my latency scores would be nearly double that at 96K with buffers at 128 and 256 samples. On top of that. I was able to re-test my old Saffire at 96K and found that that at a buffer of 128, the USB driven iD44 has a lower RTL than the Firewire Saffire and is just a bit slower at 256 samples. I added the Saffire data I was able to gather that was pertinent to my usage, in parenthesis in the table below. My Saffire had the Firewire latency set to medium as that was the most stable in usage for me. For benchmarking, I was able to get slightly faster results, but I never had much success there with actually DAW usage.

Code:

Sample Rate:  44.1K
Buffer                Measured RTL        Measured RTL (ms)
8                233.3                5.291
16                231.4                5.248
32                231.4                5.248
64                289.6 (306)        6.566 (6.940)
128                464.4 (434)        10.531 (9.843)
256                885.4 (690)        20.077 (15.648)
512                1272.4                28.853
1024                2319.4                52.595
2048                4407.6                99.945
Parenthetical numbers are from a firewire Saffire Pro 24DSP

Sample Rate: 48K
Buffer                Measured RTL        Measured RTL (ms)
8                230.5                4.802
16                238.6                4.97
32                248.5                5.177
64                262.5 (306)        5.469 (6.376)
128                494.5 (434)        10.302 (9.044)
256                886.4 (690)        18.466 (14.376)
512                1382.4                28.799
1024                2422.2                50.462
2048                4616.5                96.177
Parenthetical numbers are from a firewire Saffire Pro 24DSP

Sample Rate: 88.2K
Buffer                Measured RTL        Measured RTL (ms)
16                374.1                4.241
32                401.1                4.548
64                411.8                4.669
128                498.1                5.647
256                879                9.966
512                1675.1                18.992
1024                2517.9                28.548
2048                4623                52.415
4096                8776.9                99.512

Sample Rate: 96K
Buffer                Measured RTL        Measured RTL (ms)
16                372.4                3.879
32                400.4                4.171
64                432.3 (399)        4.503 (4.160)
128                484.3 (527.6)        5.045 (5.496)
256                900.5 (783.7)        9.38 (8.163)
512                1732.4 (1296)        18.046 (13.495)
1024                2628.4                27.38
2048                4804.3                50.045
4096                9180.4                95.629
Parenthetical numbers are from a firewire Saffire Pro 24DSP


12Bass 14th October 2019 11:20 PM

Quote:

Originally Posted by GMSweet (Post 14264520)
Here are my results from testing my new Audient iD44.

System: Intel i7-8700K @ 3.7GHz on an Asus PrimeZ370A motherboard w/ 32GB RAM running Windows 10
Connection: USB3 controller on motherboard (even though the iD44 is a USB2 controller)

Personal commentary: I typically record at 24/96K. I'm not sure how the math works out, but I was expecting the iD44 to perform worse than the Saffire Pro 24DSP it replaced based on the charts here. I was surprised to see that if I was still running at 24/48K like I used to, my latency scores would be nearly double that at 96K with buffers at 128 and 256 samples. On top of that. I was able to re-test my old Saffire at 96K and found that that at a buffer of 128, the USB driven iD44 has a lower RTL than the Firewire Saffire and is just a bit slower at 256 samples. I added the Saffire data I was able to gather that was pertinent to my usage, in parenthesis in the table below. My Saffire had the Firewire latency set to medium as that was the most stable in usage for me. For benchmarking, I was able to get slightly faster results, but I never had much success there with actually DAW usage.

Did you test the both iD44 and Saffire Pro 24 DSP on the new machine, or are the Saffire numbers from the old system? I ask because it should be providing better LLP than the Audient if both are on the same system.

GMSweet 14th October 2019 11:50 PM

Quote:

Originally Posted by 12Bass (Post 14265446)
Did you test the both iD44 and Saffire Pro 24 DSP on the new machine, or are the Saffire numbers from the old system? I ask because it should be providing better LLP than the Audient if both are on the same system.

Same system. I ran both through the RTL tool this morning. In most cases, the Saffire did beat out the Audient, but not by as much as I expected based on the existing published charts. I'm pleasantly surprised, but I was more shocked at the difference in performance of the iD44 at 44.1K vs 96K. At 256 samples, the RTL was less than half at the faster rate. I just expected higher sample rates to take more time.

DAW PLUS 15th October 2019 09:00 AM

The latency should be the same on any system with the same OS.
Performance benchmarks will differ of course.

jcsm 16th October 2019 05:27 PM

This is a great thread. I read thirty pages last night. Thank you so much. I have a couple of questions;

1. There was mention of the M-Audio M-Track 8X4M which has a bold latency claim, but are there any official results?
2. Anyone seen scores for the Native Instruments Komplete Audio 6 Mk2?

Thanks,
John

Gomjab 16th October 2019 06:35 PM

Quote:

Originally Posted by GMSweet (Post 14265501)
Same system. I ran both through the RTL tool this morning. In most cases, the Saffire did beat out the Audient, but not by as much as I expected based on the existing published charts. I'm pleasantly surprised, but I was more shocked at the difference in performance of the iD44 at 44.1K vs 96K. At 256 samples, the RTL was less than half at the faster rate. I just expected higher sample rates to take more time.

It’s the opposite. At higher rates the buffers are flushed faster. The CPU load will be higher to process the higher sample rate audio but the latency will be lower.

woodslanding 22nd October 2019 06:11 AM

Quote:

Originally Posted by Gomjab (Post 14268701)
It’s the opposite. At higher rates the buffers are flushed faster. The CPU load will be higher to process the higher sample rate audio but the latency will be lower.

Yeah, 128 samples at 48K === 256 samples at 96k

You're spitting them out twice as fast....

Dallon426 22nd October 2019 09:25 AM

Quote:

Originally Posted by jcsm (Post 14268555)
This is a great thread. I read thirty pages last night. Thank you so much. I have a couple of questions;

1. There was mention of the M-Audio M-Track 8X4M which has a bold latency claim, but are there any official results?
2. Anyone seen scores for the Native Instruments Komplete Audio 6 Mk2?

Thanks,
John


Native instruments uses generic drivers. I wouldn't put my money in the pocket of a company like that. Better off buying a UAD or RME product IMHO.

You just read the results. If it's USB you need go rme. There really is not any other competing USB interfaces.

analogcabin 22nd October 2019 01:34 PM

Zoom UAC has solid drivers with very low latency.

Dallon426 22nd October 2019 01:35 PM

Quote:

Originally Posted by analogcabin (Post 14278935)
Zoom UAC has solid drivers with very low latency.


I experienced a lot of problems with the zoom UAC. For me it was very unstable.

analogcabin 22nd October 2019 01:39 PM

That’s a shame.
I’m about to move from Mac to windows. Is that what you’re on?

daskeladden 22nd October 2019 02:20 PM

Quote:

Originally Posted by Dallon426 (Post 14278938)
I experienced a lot of problems with the zoom UAC. For me it was very unstable.

Me too, Zoom UAC-8 had terrible drivers for Windows. From what I have read Zoom UAC-2 is okay, but Zoom UAC-8 is not. My advice stay away from Zoom on Windows also due to terrible support.

bullo 22nd October 2019 04:16 PM

Quote:

Me too, ZOOM UAC-8 had terrible drivers for Windows. From what I have read ZOOM UAC-2 is okay, but ZOOM UAC-8 is not. My advice stay away from Zoom on Windows also due to terrible support.
I think you may have something wrong in your machine/configuration. I own the UAC-2 and in Windows, it is the MOST stable interface I've ever used, and it has rock-solid low-latency....I can run it at 96khz/32 sample-buffer for hours, when rehearsing at home or playing live...not a single drop-out. Just make sure that:

1) You have low DPC latency (that applies to EVERY interface - it is a MUST for low latency performance)
2) In the case of the UAC-2/UAC-8, Zoom says it ONLY works with Intel chipset. They do not guarantee it works with other. I cannot talk about other system as I only have Intel based machines.

The Zoom UAC series is a winner. The sound? It sounds amazing to me. Granted, I'm not claiming its sound is on pair with a +2k interface....But it is more than decent.

My 2 cents.
Bullo

daskeladden 22nd October 2019 04:35 PM

Quote:

Originally Posted by bullo (Post 14279160)
I think you may have something wrong in your machine/configuration. I own the UAC-2 and in Windows, it is the MOST stable interface I've ever used, and it has rock-solid low-latency....I can run it at 96khz/32 sample-buffer for hours, when rehearsing at home or playing live...not a single drop-out. Just make sure that:

1) You have low DPC latency (that applies to EVERY interface - it is a MUST for low latency performance)
2) In the case of the UAC-2/UAC-8, Zoom says it ONLY works with Intel chipset. They do not guarantee it works with other. I cannot talk about other system as I only have Intel based machines.

The Zoom UAC series is a winner. The sound? It sounds amazing to me. Granted, I'm not claiming its sound is on pair with a +2k interface....But it is more than decent.

My 2 cents.
Bullo

This has been discussed before. UAC-2 and UAC-8 are not the same. On the UAC-8 you have to have external power which seems to cause trouble. Like I have said before many people have trouble with the UAC-8 driver on Windows (and of course I had Intel chipset then and now).
As for latency Zoom Asio driver is pretty good. I have Presonus Quantum now which blows the Zoom out of the water on every level (except that the Quantum don't have a standalone function)

bullo 22nd October 2019 10:32 PM

Quote:

This has been discussed before. UAC-2 and UAC-8 are not the same. On the UAC-8 you have to have external power which seems to cause trouble. Like I have said before many people have trouble with the UAC-8 driver on Windows (and of course I had Intel chipset then and now).
Dude, I have a HARD time believing that the external power supply alone is the source of problems. I don't own a UAC-8, just the UAC-2, but I worked on one with my friend and was EXACTLY like the UAC-2...just more inputs, and the possibility to use it as stand-alone (which I haven't). Maybe a faulty unit?

Quote:

I have PreSonus Quantum now which blows the Zoom out of the water on every level (except that the Quantum don't have a standalone function)
I'm sure you are right about that. However, the Presonus is much more expensive, and it is thunderbolt. The UAC-2 and UAC-8 are USB and you can make it work with (relatively) old laptops more than decently well. IMO For the current lowest round-trip latency in the world, the Quantum rules....but you have to pay for it and a decent computer....The Zoom UAC series comes after.

Bullo

Dallon426 23rd October 2019 12:20 AM

Quote:

Originally Posted by bullo (Post 14279916)
Dude, I have a HARD time believing that the external power supply alone is the source of problems. I don't own a UAC-8, just the UAC-2, but I worked on one with my friend and was EXACTLY like the UAC-2...just more inputs, and the possibility to use it as stand-alone (which I haven't). Maybe a faulty unit?



I'm sure you are right about that. However, the Presonus is much more expensive, and it is thunderbolt. The UAC-2 and UAC-8 are USB and you can make it work with (relatively) old laptops more than decently well. IMO For the current lowest round-trip latency in the world, the Quantum rules....but you have to pay for it and a decent computer....The Zoom UAC series comes after.

Bullo

Actually I would aim for a second hand Apogee if I could not afford a better interface. I would stay clear of Zoom. I had problems with two different units.
The only interfaces I would stay clear of are Keith Mcmillen, Zoom, and Iconnect audio. They were the worst of the lot that I tested. You are better off with a focusrite or something to that affect.

nobs 25th October 2019 03:14 PM

Hello, i need a windows usb interface to record guitar only, through amp simulations like helix native.

Looking at the charts the obvious choice is rme babyface pro for 4.3 RTL at 64 buffers.
But in that same chart the audient id4 has 5.7 RTL at 64 buffers, which is pretty good yet is ranked a lot lower, why is that ?

Considering the huge price difference, 100€ vs 600€ & extra IOs i don't need on the babyface, i need your advice thank you.

DAW PLUS 25th October 2019 03:18 PM

Quote:

Originally Posted by nobs (Post 14284898)
Hello, i need a windows usb interface to record guitar only, through amp simulations like helix native.

Looking at the charts the obvious choice is rme babyface pro for 4.3 RTL at 64 buffers.
But in that same chart the audient id4 has 5.7 RTL at 64 buffers, which is pretty good yet is ranked a lot lower, why is that ?

Considering the huge price difference, 100€ vs 600€ & extra IOs i don't need on the babyface, i need your advice thank you.

The driver of the Audient is by far not as efficient, i.e. you can load a lot more/heavier plugins with small buffers with the Babyface.

96k 26th October 2019 02:44 PM

I have a zoom UAC and a zoom TAC and on a Mac they are by far the most stable interfaces I have owned. The latency is amazing on the TAC I run logic at 32 buffer with 1.9 ms on day long on large projects. For the money they are amazing!

T11 26th October 2019 04:42 PM

Yeah, I'm really curious which placements Zoom (UAC-2, UAC-8 & U44) and Tascam (208i & 102i) would get in TAFKAT's list.

If I'm correct, both Zoom & Tascam nowadays writes their own ASIO drivers?

Is TAFKAT's list available in text form or Excel sheet somewhere?

These are few of those I would like to see on that list:

Tascam Series 208i - 369 USD
https://tascam.com/us/product/series_208i/top
https://www.thomann.de/tascam_series_208i.htm

Zoom UAC-2 - 183 USD
https://www.zoom.co.jp/products/audi...udio-converter
https://www.thomannmusic.com/zoom_uac_2.htm

Zoom UAC-8 - 549 USD
https://www.zoom.co.jp/products/audi...udio-converter
https://www.thomannmusic.com/zoom_uac_8.htm

Zoom U44 - 139 USD
https://www.zoom.co.jp/products/prod...udio-interface
https://www.sweetwater.com/store/det...udio-interface

ESI MAYA44 EX (not the same as MAYA44 E) - 157 USD
https://www.esi-audio.com/products/maya44ex/
https://www.thomannmusic.com/esi_maya44_ex.htm

Behringer UMC1820 - 175 USD
https://www.behringer.com/Categories...MC1820/p/P0B2J
https://www.thomannmusic.com/behringer_umc1820.htm

Arturia Audiofuse - 433 USD
https://www.arturia.com/products/aud...ofuse/overview
https://www.thomannmusic.com/arturia...dark_black.htm

Steinberg UR22mkII - 107 USD
https://www.steinberg.net/en/product.../ur22mkii.html
https://www.thomannmusic.com/steinberg_ur22_mk2.htm

Audient iD22 - 333 USD
https://audient.com/products/audio-i...id22/overview/
https://www.thomannmusic.com/audient_id22.htm

MiniDSP USBstreamer B - 105 USD
https://www.minidsp.com/products/usb...sbstreamer-box

Quote:

Originally Posted by 96k (Post 14286659)
I have a zoom UAC and a zoom TAC and on a Mac they are by far the most stable interfaces I have owned. The latency is amazing on the TAC I run logic at 32 buffer with 1.9 ms on day long on large projects. For the money they are amazing!


nobs 28th October 2019 04:15 PM

Thanks, i'm gonna get rme babyface for drivers & durability peace of mind after looking mixed reviews on zoom uac, although i had to stop my purchase because i got infos on a new babyface pro "FS" model next month, any more infos on that ?