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)

Xichael 28th April 2017 06:51 AM

Any chance a Zoom UAC device might make it into the database? I've been reading that they have extremely low latency for USB.

SEA 27th May 2017 02:01 PM

I'm looking at picking up the RME HDSPe AIO PCIe to get my latency down. However, would a FireWire unit like the RME's Fireface line be as fast?

My plan was to go with PCIe then ADAT or AES, S/PDIF with a Hilo and a Lyra 2 sometime this year.

However, my current situation is I have the UR28M and the latency is an issue. Until I pick up the Hilo I was going to go S/PDIF out of the UR28M and into the RME HDSPe AIO.

I also have a Liquid Mix on Fire Wire so a separate FW card should be fine.

Any tips?

Thanks!

Skap 27th May 2017 06:14 PM

Quote:

Originally Posted by SEA (Post 12646596)
I'm looking at picking up the RME HDSPe AIO PCIe to get my latency down. However, would a FireWire unit like the RME's Fireface line be as fast?

My plan was to go with PCIe then ADAT or AES, S/PDIF with a Hilo and a Lyra 2 sometime this year.

However, my current situation is I have the UR28M and the latency is an issue. Until I pick up the Hilo I was going to go S/PDIF out of the UR28M and into the RME HDSPe AIO.

I also have a Liquid Mix on Fire Wire so a separate FW card should be fine.

Any tips?

Thanks!

I'm not sure how you're planning on using the Hilo, but be aware of that the AES and S/PDIF connections on the Hilo can not be used simultaneously, so if you want to use both the Main out and the separate Monitor out you might want to consider using the ADAT connection on the Hilo. (By the way Lynx's USB card is not that bad, and their Thunderbolt card is the same as in Aurora, i.e. excellent, but it's very expensive.)

Regarding latency, I think the RME PCIe solution is both a bit faster and a bit more efficient (able to run more plugins @ same buffer setting) than RME's own FW solutions or any other solution out there. If you're using a desktop anyway, a RME PCIe card is probably as good a solution as you can get. I recently bought a used RME RayDat PCIe card that I'm going to use with my Hilo when I make the transition from laptop to desktop later this year (probably when Skylake-X and Threadripper are released), and I expect it to be better than my old RME Fireface 400, which still is pretty impressive in terms of latency and efficiency.

Alndln 27th May 2017 07:03 PM

Quote:

Originally Posted by Xichael (Post 12593151)
Any chance a Zoom UAC device might make it into the database? I've been reading that they have extremely low latency for USB.

There was a discussion about it some pages back (try searching these pages). I think the problem was that Zoom was unwilling to provide a unit for testing so that left waiting for someone to provide a unit for testing. I'd like to see those numbers myself.

SEA 27th May 2017 07:13 PM

Quote:

Originally Posted by Skap (Post 12646922)
I'm not sure how you're planning on using the Hilo, but be aware of that the AES and S/PDIF connections on the Hilo can not be used simultaneously, so if you want to use both the Main out and the separate Monitor out you might want to consider using the ADAT connection on the Hilo. (By the way Lynx's USB card is not that bad, and their Thunderbolt card is the same as in Aurora, i.e. excellent, but it's very expensive.)

Regarding latency, I think the RME PCIe solution is both a bit faster and a bit more efficient (able to run more plugins @ same buffer setting) than RME's own FW solutions or any other solution out there. If you're using a desktop anyway, a RME PCIe card is probably as good a solution as you can get. I recently bought a used RME RayDat PCIe card that I'm going to use with my Hilo when I make the transition from laptop to desktop later this year (probably when Skylake-X and Threadripper are released), and I expect it to be better than my old RME Fireface 400, which still is pretty impressive in terms of latency and efficiency.

Thanks Skap! I would like the Prism to connect to the Hilo and use the Hilo for routing (if that's the best way to go). I was also looking into the Lynx AES16e as an option vs. the RME HDSPe AIO but it's only AES (16 channels) and perhaps that would be fine.

My goal is to be able to record through whichever AD I want and listen to whichever DA I want AS WELL as loop OUT of the DA of choice through some hardware and print through which ever AD of choice.

In most case scenarios I would record through the Prism AD, capture the Prism DA into the Hilo AD (with maybe some hardware in-between).

So any suggestions would be greatly appreciated. :-)

SEA 27th May 2017 07:40 PM

Quote:

Originally Posted by Skap (Post 12646922)
I'm going to use with my Hilo when I make the transition from laptop to desktop later this year (probably when Skylake-X and Threadripper are released)

That's my plan as well! I have a dual Xeon which is great for video editing but I need more power for VSTis like Falcon and Omnisphere.

If I was to build new now I would go with the X99 MB and the 6850K but waiting on the new X299 CPU's. 8 or 10 cores would be great at 4.5 without having to OC (if possible).

Skap 30th May 2017 10:58 PM

SEA, if you're certain that you'll be using the Intel platform (which probably makes sense for use with demanding VSTis), you could always go the Thunderbolt route with a LT-TB card in the Hilo and connect the Lyra 2 to the Hilo with the ADAT connections which will give you access to all of Lyra 2's analog I/Os (including headphones out, which could come in handy).

The LT-TB card is expensive, but so is the RME HDSPe cards.

Intel has recently revealed that it has decided to make Thunderbolt more accessible for developers in the future, and to include it in the chipsets themselves, so it could be that it is a protocol worth investing in.

It also gives you the advantage of being compatible with portable systems if that kind of flexibility is important to you.

If you buy the Hilo with the LT-TB card already installed you'll probably get a better deal than you would if you were buying them separately. I bought the Hilo before the LT-TB card was released, so mine came with the LT-USB card.

Because the Lynx LT-TB card is so expensive I decided to go with the RME PCIe card solution instead and be free to choose any chipset and motherboard (Intel or AMD based, with or without Thunderbolt).

Both Lynx and RME have great drivers with low latencies for all protocols and both MacOS and Windows. Hilo works great as a hub for controlling everything, very intuitive, but RME's software is prossibly even more flexible + the metering suite is great. By going RME PCIe + Hilo, you'd have access to both, but I guess you'd mostly be using the Hilo's touch screen and routing features anyway. You probably can't go wrong either way.

By the way, what samplerates do you use?

SEA 31st May 2017 01:45 AM

Quote:

Originally Posted by Skap (Post 12652409)
SEA, if you're certain that you'll be using the Intel platform (which probably makes sense for use with demanding VSTis), you could always go the Thunderbolt route with a LT-TB card in the Hilo and connect the Lyra 2 to the Hilo with the ADAT connections which will give you access to all of Lyra 2's analog I/Os (including headphones out, which could come in handy).

The LT-TB card is expensive, but so is the RME HDSPe cards.

Intel has recently revealed that it has decided to make Thunderbolt more accessible for developers in the future, and to include it in the chipsets themselves, so it could be that it is a protocol worth investing in.

It also gives you the advantage of being compatible with portable systems if that kind of flexibility is important to you.

If you buy the Hilo with the LT-TB card already installed you'll probably get a better deal than you would if you were buying them separately. I bought the Hilo before the LT-TB card was released, so mine came with the LT-USB card.

Because the Lynx LT-TB card is so expensive I decided to go with the RME PCIe card solution instead and be free to choose any chipset and motherboard (Intel or AMD based, with or without Thunderbolt).

Both Lynx and RME have great drivers with low latencies for all protocols and both MacOS and Windows. Hilo works great as a hub for controlling everything, very intuitive, but RME's software is prossibly even more flexible + the metering suite is great. By going RME PCIe + Hilo, you'd have access to both, but I guess you'd mostly be using the Hilo's touch screen and routing features anyway. You probably can't go wrong either way.

By the way, what samplerates do you use?

Thanks Skap! I believe PCIe is the way to go. I'll also try to find an ASUS motherboard that has TB as well when the time comes. Going to wait and see how the new X299 pan out first.

Skap 31st May 2017 07:55 AM

Quote:

Originally Posted by SEA (Post 12652616)
Thanks Skap! I believe PCIe is the way to go. I'll also try to find an ASUS motherboard that has TB as well when the time comes. Going to wait and see how the new X299 pan out first.

You're welcome.

Be aware, though, that if you choose PCIe you'd probably want to connect the Hilo to the PCIe via ADAT. This leaves the AES and S/PDIF connections on the Hilo available for connecting to the Lyra 2. AES and S/PDIF can't be used simultaneously on the Hilo, so you'll only be able to fully utilize one of them, f.i. AES (i.e. stereo in and out). This gives you stereo in and stereo out from the Lyra 2 into the Hilo via AES.

Example: Hilo will be the router and will output its own analog stereo inputs to its ADAT out 1 and 2 and pass through the AES signal it received from the Lyra 2 to f.i. ADAT 3 and 4. These four audio signals are transmitted to the PCIe card's ADAT inputs. The outputs of the PCIe card will be sent to the Hilo's ADAT input with ADAT cannel 1 and 2 sent to Hilo's main out, ADAT 3 and 4 sent to Hilo's monitor out, ADAT 5 and 6 sent to Hilo's headphones out, and ADAT 7 and 8 sent to Hilo's stereo AES out and passed through to Lyra 2's AES input and thereafter one(!) of its analog out.

As you can see, if you're using a PCIe card and routing everything through the Hilo, you'll only be able to use one of the analog outputs of the Lyra 2.

A way to utilize all of its outputs is to connect all or some of the digital I/Os directly from the Lyra 2 to the PCIe card. The Hilo will not be able to control or store presets for routing through those outputs, i.e. Lyra 2's monitor out and headphones out, but you'll be able to use them and control them with the software of the PCIe card.

Another way to utilize all of the analog I/Os on both devices is to use the LT-TB card option.

The Thunderbolt option will also let you use higher samplerates (that's the reason I asked what SR you expect to be using) because when using higher samplerates the ADAT will have to be used in S/MUX mode this will cut the number of ADAT channels in half, but if the ADAT connections can be reserved for use between the Hilo and the Lyra 2 instead of between the Hilo and the PCIe card, you'll be able to use all or most channels on both devices also in 96 KHz. I just wanted to make sure that you are aware of this.

msmucr 31st May 2017 08:09 AM

@ Skap

It's bit OT here... and we quite extensively talked about that in other SEA's threads..

https://www.gearslutz.com/board/musi...iple-adda.html
https://www.gearslutz.com/board/musi...l#post12648105

Michal

Skap 31st May 2017 09:06 AM

Quote:

Originally Posted by msmucr (Post 12653011)
@ Skap

It's bit OT here... and we quite extensively talked about that in other SEA's threads..

https://www.gearslutz.com/board/musi...iple-adda.html
https://www.gearslutz.com/board/musi...l#post12648105

Michal

You're right, I guess it's quite a bit OT, so we'll leave it there.

Loving this thread by the way. Learning a lot from it. It's what made me buy the good old RayDat. Thanks to everyone who has contributed.

SEA 31st May 2017 02:27 PM

Quote:

Originally Posted by Skap (Post 12652999)
A way to utilize all of its outputs is to connect all or some of the digital I/Os directly from the Lyra 2 to the PCIe card. The Hilo will not be able to control or store presets for routing through those outputs, i.e. Lyra 2's monitor out and headphones out, but you'll be able to use them and control them with the software of the PCIe card.

Another way to utilize all of the analog I/Os on both devices is to use the LT-TB card option.

The Thunderbolt option will also let you use higher sample rates (that's the reason I asked what SR you expect to be using) because when using higher sample rates the ADAT will have to be used in S/MUX mode this will cut the number of ADAT channels in half, but if the ADAT connections can be reserved for use between the Hilo and the Lyra 2 instead of between the Hilo and the PCIe card, you'll be able to use all or most channels on both devices also in 96 KHz. I just wanted to make sure that you are aware of this.

So connecting the Hilo and Prism separately to the PCI card lets me use and monitor both converts at the same time! For instance, If I want to mix out of the Prism DA and back into the Prism AD, then I can switch between listening to the mix through either the Hilo DA or the Prism DA correct?

I could also go out the Prism DA, through some hardware, then back into the Hilo AD or the Prism AD to render the mix (or vice versa).

With TB I would have to use route the Prism through the Hilo and then I wouldn't be able to use all the converters simultaneously.

What I like about the RayDAT is you have more than one ADAT. More connectivity if needed.

Skap 31st May 2017 02:43 PM

Sorry for staying off topic, guys.

Quote:

Originally Posted by SEA (Post 12653356)
So connecting the Hilo and Prism separately to the PCI card lets me use and monitor both converts at the same time! For instance, If I want to mix out of the Prism DA and back into the Prism AD, then I can switch between listening to the mix through either the Hilo DA or the Prism DA correct?

I could also go out the Prism DA, through some hardware, then back into the Hilo AD or the Prism AD to render the mix (or vice versa).

With TB I would have to use route the Prism through the Hilo and then I wouldn't be able to use all the converters simultaneously.

What I like about the RayDAT is you have more than one ADAT. More connectivity if needed.

SAE, I think you're misunderstanding. Using even the LT-USB (or the LT-TB) card with the Hilo will give you 16 channels of audio from your computer that you can route freely to whichever digital or analog output you may choose.

Let's not clutter this thread any more, we're wasting people's time. If you have more questions, feel free to PM me.

SEA 31st May 2017 03:24 PM

Quote:

Originally Posted by Skap (Post 12653378)
Let's not clutter this thread any more, we're wasting people's time. If you have more questions, feel free to PM me.

Of course, there will be "newbies" who don't get it either so this may come in handy for future use. :lol:

However, my apologies for not "getting it" the first time I read it. I just woke up a bit a go and now having my morning joe! :chug:

Now... After re-reading your last post, I see where you said
Quote:

Another way to utilize all of the analog I/Os on both devices is to use the LT-TB card option,
but if the ADAT connections can be reserved for use between the Hilo and the Lyra 2 instead of between the Hilo and the PCIe card, you'll be able to use all or most channels on both devices also in 96 KHz.
So yeah! TB lets you use them all at a higher sample rate using all channels vs connecting to the PCIe card! Sounds good! kfhkh

themiracle 11th June 2017 05:41 AM

Rme ufx+
 
Just another plug for the RME UFX+, as I'd love to see those numbers.

thanks for all this!

TNM 12th June 2017 01:35 PM

any chance of getting apollos? I believe there are thunderbolt drivers for windows now, and a usb3 windows only apollo duo. would love to see those.

UnderTow 13th June 2017 02:08 AM

Quote:

Originally Posted by TNM (Post 12675528)
any chance of getting apollos? I believe there are thunderbolt drivers for windows now, and a usb3 windows only apollo duo. would love to see those.

TAFKAT has already posted results for the Apollo Twin:

https://www.gearslutz.com/board/12415420-post1207.html

Alistair

TNM 13th June 2017 02:37 AM

Quote:

Originally Posted by UnderTow (Post 12676912)
TAFKAT has already posted results for the Apollo Twin:

https://www.gearslutz.com/board/12415420-post1207.html

Alistair

thanks.. doesn't look great.. i couldn't find it on the front page post, which is the one i checked before i asked.. sorry about that

TAFKAT 21st June 2017 10:33 PM

Apologies for the lack of updates recently, a lot of other things taking precedent over interface testing lately.

I need to formulate a procedure to include Thunderbolt interfaces into the Database , which is proving difficult as I do the TB testing on a completely different system to the default system that has been locked down all these years to maintain the consistency. I have a few ideas involving using an RME HDSPe reference on the new system to then correlate the results against , and then transpose the results against the earlier system. It isn't that straight forward tho as the current architecture is more efficient with USB3 for example, but I will connect the dots eventually.

Hang in there, I should have some new interface to run through their paces shortly.

Quick response re the Zoom, the local distro was not interested in even selling me a unit, actually they never even replied to multiple requests to contact me back about the product, despite having the product spread across general MI retailers all over the country. In short, they have no idea what they have or what to do with the audio interfaces as their core focus has been guitar orientated. As good as the units may be, if the parent company cannot connect the dots with their local reps to even know what the product focus is, I don't have the time or energy to waste on them.

peachh

nonstatic 2nd July 2017 12:05 PM

Any possibility to test the new Arturia AudioFuse? Love the audio specs but wouldn't consider it unless the drivers perform near RME. It's their first interface so we have no idea if they did the drivers themselves or outsourced.

evo6 2nd July 2017 07:40 PM

Quote:

Originally Posted by nonstatic (Post 12713706)
Any possibility to test the new Arturia AudioFuse? Love the audio specs but wouldn't consider it unless the drivers perform near RME. It's their first interface so we have no idea if they did the drivers themselves or outsourced.

Agreed. In the market for a good portable interface. My brain and past experience say RME, but my curiosity says Audiofuse. Would love to have more info.

E6

Alndln 3rd July 2017 06:11 AM

Quote:

Originally Posted by nonstatic (Post 12713706)
Any possibility to test the new Arturia AudioFuse? Love the audio specs but wouldn't consider it unless the drivers perform near RME. It's their first interface so we have no idea if they did the drivers themselves or outsourced.

My brother told me he found a review that basically said it sounded great but the latency was terrible. The next time I see him I'll ask him for a link so I can post it here. I know a review is not the same as a proper test but it at least should give us an idea of the drivers until Tafkat can possibly get his hands on a unit.

skyform 13th July 2017 10:56 AM

Quote:

Originally Posted by TAFKAT (Post 12484034)
You will find the I/O and RTL's will be actually lower on the Babyface Pro than the AIO at the available latencies , due to the older AD/DA's used on the AIO. Where the difference is with the AIO will be in the higher efficiency at extreme loadings.

If the feature set with the BF Pro is more suitable, I don't think you will be disappointed.

My clients who opt for the AIO are mostly TV/Film composers who do not have a requirement for inbuilt Mic pres , and need to squeeze every last ounce out of their available resources at their preferred latencies when dealing with huge VI templates..

Horses for Courses.

peachh

Hello, I'm also working with huge VI templates, big orchestral libraries with projects over 128 midi tracks and I'm deciding between AIO and Babyface Pro.

I worried about the AIO digital interference issues which many people report, I find this very dissapointing for such a expensive card, Babyface wouldn't have this problem because it's an external card.

I would also need a mic input in the future, plus I like the idea to have the the thing on my desk for changing the volume and headphones connection.

My current audio card is M-Audio Audiophile Delta 192 and I'm wondering how would Babyface Pro compare to my Delta 192 and AIO for big projects.

TAFKAT 13th July 2017 11:28 PM

Quote:

Originally Posted by skyform (Post 12734627)
I worried about the AIO digital interference issues which many people report, I find this very dissapointing for such a expensive card, Babyface wouldn't have this problem because it's an external card.


FWIW - I have sold , installed and configured dozens of these cards without issue !

I always use the balanced breakout ( Optional) , never use the standard rca. Is this what people are reporting having issues with ?

If so, its common knowledge to use a balanced connection between any interface and your speakers to inoculate against electrical interference from computers. Using an unbalanced connection on an external interface can also result in interference.

No idea on the comparative performance of the Delta cards, they haven't been available for over a decade.


peachh

zephonic 14th July 2017 12:34 AM

Quote:

Originally Posted by TAFKAT (Post 12415420)
Hey All,

O.K, Finally got a chance to put the UAD Apollo twin USB 3 through its paces.

...............................

Low latency Performance is pretty much as I expected, below average , ( I had already been warned by some colleagues who will remain nameless ) so for anything above basic tracking it falls over pretty quickly when adding Plugins, Virtual Instruments. I didn't go into testing the DSP side of the unit, as that wasn't my focus.

I know those requiring LLP are not the target market for the unit, and the results will pretty much be dismissed by most who will gravitate to the unit.


I'm definitely not in the market for an Apollo, but the word out here is that it works great with Logic. Lot of people use that combination and it seems to be a real winner.

TAFKAT 14th July 2017 01:54 AM

Quote:

Originally Posted by zephonic (Post 12735927)
I'm definitely not in the market for an Apollo, but the word out here is that it works great with Logic. Lot of people use that combination and it seems to be a real winner.

I'm sure there are a lot of people using the Apollo's and are happy with them , horses for courses.

I simply test and report the numbers comparatively to the rest of the database , and within context of LLP on Windows.

The rest is up to the individual to gauge the value and relevance from that information.

peachh

heder 18th July 2017 05:57 PM

Hi all, first poster.

I wish I had done my research and found this thread earlier...

I just bought the ESI Maya 44eX (2017 version) for what I though was a quite cheap, really low latency card, being PCI Express and everything.
For example, the driver advertises 64 samples (1.45 ms latency @ 44.1), but the CEntrance utility gives 9.93 ms RTL. I know this numbers has been posted in this thread before, but right now I am a bit pissed off (both at myself for not doing research, and at ESI for the issue...) and I want to make sure noone else makes the same mistake.
So, I guess this card is going back, and I would have to look at maybe a Zoom UAC-2 or perhaps RME instead...

BR
//Thomas

...and thanks for all the hard work that has been put into this thread.

heder 20th July 2017 07:23 PM

Quote:

Originally Posted by heder (Post 12744211)
Hi all, first poster.

I wish I had done my research and found this thread earlier...

I just bought the ESI Maya 44eX (2017 version) for what I though was a quite cheap, really low latency card, being PCI Express and everything.
For example, the driver advertises 64 samples (1.45 ms latency @ 44.1), but the CEntrance utility gives 9.93 ms RTL.

...

Now, I don't want to hog this thread all for my own little issue here, but - I was in contact with ESI and asked if my results was to be expected, and the answer a got was that the CEntrance Latency checker not really could be 100% trusted.
It's result could according to ESI vary quite a bit depending on the hardware. I can understand this, but I got a screendump from them where they had ran the same test as I (44,1/64 samples) with a measured RTL of 3.11 ms with the same driver where I got 9.93 ms.
My computer is a quite optimized i7 with SSD:s and 16Gb RAM, etc. I ran the tests multiple times? Should differences this big be expected between systems?

BR
//Thomas

msmucr 20th July 2017 08:49 PM

Quote:

Originally Posted by heder (Post 12748284)
Now, I don't want to hog this thread all for my own little issue here, but - I was in contact with ESI and asked if my results was to be expected, and the answer a got was that the CEntrance Latency checker not really could be 100% trusted.
It's result could according to ESI vary quite a bit depending on the hardware. I can understand this, but I got a screendump from them where they had ran the same test as I (44,1/64 samples) with a measured RTL of 3.11 ms with the same driver where I got 9.93 ms.
My computer is a quite optimized i7 with SSD:s and 16Gb RAM, etc. I ran the tests multiple times? Should differences this big be expected between systems?

BR
//Thomas

Hi Thomas.. computer optimization or CPU doesn't play any role in such simple round-trip latency measurement. Obtained figure depends only on audio interface hardware, its driver.. and how precise is used measurement software.

CPU or computer optimizations became important, when you want to use the interface for real-world projects.. or in another way.. how much load you can do at particular buffer size.. In that aspect.. all of mentioned elements (CPU, interface, system..) plays its role. But latency figure at give buffer size doesn't change.

I haven't really used CEntrance tester for long time.. However Oblique RTL Utilily or even Reaper ReaInsert plugin is fine for such measurements.
RTL Utility | Oblique Audio

Anyway.. it's important, that for proper detection, you have to keep audio levels right (eg. pay attention on input overload or too weak signal).
So before such RTL measurement I usually open DAW, insert some internal signal generator (say with -12dBFS sine) and monitor external loopback and possibly adjust output sensitivity or input gain, where it's applicable.. util I have about the same level also on interface input.
Then I measure RTL.

With regard to your particular ESI card.. I've never seen it, but according to your description, one of your measurement is completely off.. it's not possible to have such big discrepancy among different measurement passes.
Other thing is, if it reports 1.45ms latency to DAW, it can't be really right. Because it's just sheer time of 64 samples buffer.. you can easily calculate it 64/44.1 (that will give you ms figure).
There is certainly also another buffer and latency of converter chip itself.
You can imagine complete roundtrip latency as: DAW output buffer + additional data transfer buffer + DAC + ADC + input data transfer buffer (if it's used) + DAW input buffer.. all of that makes that figure, so if interface reports just sole DAW buffer lengths, it's almost certainly incorrect.
However in most DAWs, it's possible to further compensate that by manual offset.. just don't rely on that.

Michal

12Bass 20th July 2017 08:54 PM

Quote:

Originally Posted by heder (Post 12748284)
My computer is a quite optimized i7 with SSD:s and 16Gb RAM, etc. I ran the tests multiple times? Should differences this big be expected between systems?

No, 3.11ms is extremely low analog loopback latency for 64 buffer. Only a few interfaces are capable of going that low at 32 let alone 64. Might be more realistic for digital loopback. System differences should not affect absolute RTL but will affect stability at specific buffer settings.