Thanks :) Yes, the Surface was a pleasant surprise. But this video shows that, on a proper setup, the AudioFuse and its ASIO driver can achieve those ridiculous settings Arturia is promoting, even though, in practice, one would end up settling for 64 samples... 32 if trying to do real-time I/O audio effects to external sources (guitar, voice, etc)
Thanks for this video, it's still informative to me after 5 years as I do own a Surface Pro 3 with an i7-4650U processor that I bought in 2015. This computer has served me faultlessly, but now that I'm looking into an Audio I/F (Audiofuse Studio or Focusrite Scarlett 18i8) and DAW, I might eventually be forced to notch up to something more powerful... maybe a gaming laptop?
Thanks for informing on the Audiofuse. I've been eyeing it for awhile but adamant because of all the reports of heat and people sending them back. What I'm curious about though now is, why not an i7 processor?
I’ve looked into the Audiofuse but 600 euros is a bit too expensive for me. Are there any other less expensive usb audio interfaces with good Asio drivers? I’m using the integrated audio on my mobo at the moment with asio4all but I’ve got a total roundtrip of 15 ms which is a bit too much. I don’t need to connect any mics or guitars. I only use midi keyboards via usb and Ableton.
Peter Walker There is an earlier (very useful) vid of Robin's where he walks through his steps for cutting down OS fluff on Windows 10 creators update.
Well this is pretty frustrating for me because i get RTL of 12,2ms in my M1 Mac Mini 16GB within Logic Pro X. They really need to update their Mac OS drivers and make it compatible with M1 Macs. In my PC desktop the values are significantly better.
Latencies reported via Arturia driver is a total fake ....you need to measure it by analog loopback with centrance utility.... real latency of arturia on 64 samples .44 khz - 7 ms
Round trip yeah? I found that to be at 128 samples, but your mileage may vary. I've found over the years that Cubase gives a pretty accurate report of latency figures and i prefer to base my testing on using audio that i can hear - which you can't with the centrance utility. But it's all good :)
Сubase and other daw - not give a "accurate report" - they give numbers reported by driver ... and arturia driver - lie .... like many other manufacturers ... exclude Rme ,lynx ... "that i can hear " - strange method for professional testing ... some people don`t hear 12 ms latency ,, some people - hear 5 ms .... On your video - you play keyboard ,, - this is a half of roundtrip latency Because you have no input Analog to Digital latency ... if you playing guitar with plugins or tracking vocal - latency will be doubled ... And a second very important thing - how system works with such latencies... For example - on real 5 ms roundtrip with top rme hdsp card i can load 100 plugins and with 7 ms on arturia - only 30 ...before i hear massive dropouts .. check DAWBENCH testing on gearsluts... .. dawbench audio interface testing methodic - best and most professional up to date ... Roundtrip latency ,, overall perfomance on different latency ...
I'm aware of the disparity but i've found that Cubase reports the numbers more accurately than other DAWs or the audio interfaces driver app. It includes any safety buffers or other bits and bobs that are often missed off driver control panels. I'm not talking about hearing the latency, although that's a factor, i'm talking about listening to the audio so that i know the audio under test is a stable data stream - that's all. And yes i understand the difference between output latency and round trip latency. And it's not necessarily a doubling because input and output latencies can be different. And yes plug-in testing at different buffer sizes is very important. And yes RME interfaces can do a heck load more plug-ins than other interfaces at the same buffer size. I do make all of these points in all of my performance test videos and the videos I've done on latency and i use the Dawbench test along with my own test projects. So we're on the same page mate :)
I get 9.30 toal ms latency on Windows 10 Surface Pro 6, at 96 samples on my Roland Quad Capture. How does it compare to the Audiofuse? The Audiofuse seems to have 4 ms less than the Quad Capture at the same sampke rate with no glitching, apparently, and even less if you are using the 2 nucleous Surface Pro 5. I am considering purchasing the audiofuse because of this and its built-in USB hub.
@@MoltenMusicTech I I adjust the buffer, load Eastwest plugg-ins and check the ASIO menu numbers. The only thing is that in my case it generally works well (when loading medium-size projects) at 128 samples, and ideally at playback and recording at 256 samples, which ends up being more than 10 miliseconds, though. I should double-check for cracks and see up to which buffer size I can set it, loading a full project. Or do the same thing up for comparison. I have the Surface Pro 6 i5 model and the Roland Quad Capture. It would be interesting to check, because I would see if upgrading to Audiofuse is a good idea or not necessarily. I believe as you say that latency depends on both the dedicated driver/hardware combination and processor.
Confirmed: Roland Quad Capture at 96 samples. 44100 hz working well with playback and recording on Surface Pro 6 (i5u quad) shows 5.510 ms IN latency and 4.331 Output Latency. I would send you a screen capture but this does not allow me to do so. How does this compare to Audiofuse? It ios stable at 128 samples 44100 Hz, which shows 5.395 IN and 2.494 output latency. It seems Arturia is a bit faster, but not that much, unless your compouter is a Dual-core i5
It's an i5 dual core Surface Pro in the video and I show you the latency in the video yeah? I ask how you're measuring it because there can be all sorts of hidden "safe mode" latencies that are not shown in the control panel. Anyway, i don't have a SP6 setup anywhere to test it for you but as you can see from the video at 128 samples it's a tiny bit lower than yours.... which is also why i check because 96 samples should be lower than 4.331 which makes me think there are other buffers involved. But honestly for round-trip latency you're not going to get much lower than what you already have unless you go for an RME Audio interface.
@@MoltenMusicTech Now you mention it, there might be some extra buffer, yes. I will check, remove it and re-write the result. RME? I will have a look although anything lower than 10 ms works well for me. I guess if I re-adjust the buffer of my Quad Capture, it might be more than enough for my needs. Both interfaces seem to do the job, despite the mobile processor's limitations.
Didn't have the time to go that deep into it unfortunately. I had to grab an opportunity while it presented itself - there's always something else to test :)
+Molten Music Technology EXCELLENT! Thanks for this update to your original review. 👍👍 Again, as I mentioned in the comments of your initial review, I always want to record or capture audio at 24/96 minimum, and preferably 24/192 in order to match the human auditory system's capability in the Time Domain, to reduce smearing & pre/post ringing artifacts created by the steep brick-wall cutoff filters that are used in the ADC & DAC process, and for an overall lower noise floor/advantage in gain staging. All of your tests still seem to be at a 44.1kHz sample rate, correct? Again, let's see what happens to your RT Latency, Channel Count, Plug-In Count, and VI's at Higher Sample Rates!?!? One reason why all of these USB audio interface manufacturers Still Insist that USB 2.0 is "more than adequate" in terms of bandwidth for good latency performance and 10+ Channels of simultaneous audio I/O is because they are still only referencing a 44.1/48kHz sample rate! 😒😕 :/ The standard in professional recording studios has been a minimum capture and workflow of 24/96 for AGES, with quite a few of them using 24/192 for 10+ years. With the technological advancements of today, why should we be theoretically limited to 44.1-48kHz?
No offense but, looking at your screen it's obvious that you are only changing the desired buffer size and the actual buffer size and output latency do not change after you go lower than 64 samples, so it's kind of a moot point since the cpu has the same amount of time to compute the output in 64, 32, 16 and 8 sample modes.
None taken. True enough the output latency is stuck at the lowest level of 1.5ms but that's not the only thing that changes when reducing the buffer size. It's also about the round-trip latency. But the point i was trying to make is that on the Surface it couldn't handle an output latency of 1.5ms whereas the desktop could - which is what Arturia had claimed. You also have to take into account that the reported numbers are not to be trusted. If you check out my review of the Komplete Audio 2 you can see the latency measuring I did and how far removed it is from the reported latency - and the KA2 appears to use the same drivers as the Audiofuse - ruclips.net/video/mA7_fcdx12I/видео.html
Hi Robin, I was wondering if you could do a review of the UAD Apollo thunderbolt on Windows machines. I've had it for months and have never been able to get it to be stable on my hp spectre x360 i7 8550u 16gb ram.
Laptops are troublesome i'm afraid. I don't think me reviewing it will have any impact on how well it works on your machine as I'd be doing it on a desktop. PC's haven't worked out Thunderbolt quite right yet unfortunately.
Molten Music Technology I fully understand what your saying. The only thing is every other piece of thunderbolt 3 hardware from monitors to hubs to GPU's works flawless in Windows 10. Its just the apollo thats buggy. Maybe its because of the Thunderbolt 2 to 3 conversion but, the guys at apollo say it should work problemless thats why I was wondering if you could do a tour on the apollo see if it works for you. Its great hardware maybe one of the best and there' s a ton of musicians out there either owning it or thinking about buying it. Anyway thanks for the response 👍
Molten Music Technology haha I wish I could but, we're based in the Netherlands and the other guys in the studio use it with their macs on a daily basis.
if you are on windows XPto10 check out the Odeus asio link driver adon..all my interfaces are have now new and more fuctionality and are much more stable now!!!i dont now how they made that,but i never lookt back!??!Now i just look for roundtrip latency and connectivity!!!!!wood be awsome if you Rewiew that Killer app!
That is a testament to your good music PCs also. I'll have to check them out!
Very useful video. Thank you. I’m looking at the 4 channel model now. This helped.
Thanks :) Yes, the Surface was a pleasant surprise. But this video shows that, on a proper setup, the AudioFuse and its ASIO driver can achieve those ridiculous settings Arturia is promoting, even though, in practice, one would end up settling for 64 samples... 32 if trying to do real-time I/O audio effects to external sources (guitar, voice, etc)
Thanks for this video, it's still informative to me after 5 years as I do own a Surface Pro 3 with an i7-4650U processor that I bought in 2015. This computer has served me faultlessly, but now that I'm looking into an Audio I/F (Audiofuse Studio or Focusrite Scarlett 18i8) and DAW, I might eventually be forced to notch up to something more powerful... maybe a gaming laptop?
Thanks for informing on the Audiofuse. I've been eyeing it for awhile but adamant because of all the reports of heat and people sending them back. What I'm curious about though now is, why not an i7 processor?
The i5 is the sweet spot I think. I7 is too expensive and has a fan, the i5 is fanless and silent. Yes it does get hot.
Ah I forgot about the fan. Cool, thanks. Cheers!
@@Nomo_Popo And Audiofuse hs the "Audiofuse Studio" which I think fixes the heat issue. I don't have one so I have not checked, though.
@@MoltenMusicTech Is this the i5 Dual or Quad processor?
I’ve looked into the Audiofuse but 600 euros is a bit too expensive for me. Are there any other less expensive usb audio interfaces with good Asio drivers? I’m using the integrated audio on my mobo at the moment with asio4all but I’ve got a total roundtrip of 15 ms which is a bit too much. I don’t need to connect any mics or guitars. I only use midi keyboards via usb and Ableton.
Yes, loads - check out the UR range from Steinberg, the NI Komplete Audio 6, anything by RME or MOTU.
Molten Music Technology Thanks for the reply! I looked into RME but they are expensive too. I’ll check out the others!
Excellent comparison.
Doesn't the desktop have 4 physical cores though? And much higher clock?
Yes it does, that's precisely the point.
How much effort do you put into shutting down services and fluff in the OS?
Peter Walker There is an earlier (very useful) vid of Robin's where he walks through his steps for cutting down OS fluff on Windows 10 creators update.
Thanks for that!
Well this is pretty frustrating for me because i get RTL of 12,2ms in my M1 Mac Mini 16GB within Logic Pro X. They really need to update their Mac OS drivers and make it compatible with M1 Macs. In my PC desktop the values are significantly better.
Latencies reported via Arturia driver is a total fake ....you need to measure it by analog loopback with centrance utility.... real latency of arturia on 64 samples .44 khz - 7 ms
Round trip yeah? I found that to be at 128 samples, but your mileage may vary. I've found over the years that Cubase gives a pretty accurate report of latency figures and i prefer to base my testing on using audio that i can hear - which you can't with the centrance utility. But it's all good :)
Сubase and other daw - not give a "accurate report" - they give numbers reported by driver ... and arturia driver - lie .... like many other manufacturers ... exclude Rme ,lynx ...
"that i can hear " - strange method for professional testing ... some people don`t hear 12 ms latency ,, some people - hear 5 ms ....
On your video - you play keyboard ,, - this is a half of roundtrip latency
Because you have no input Analog to Digital latency ...
if you playing guitar with plugins or tracking vocal - latency will be doubled ...
And a second very important thing - how system works with such latencies...
For example - on real 5 ms roundtrip with top rme hdsp card i can load 100 plugins
and with 7 ms on arturia - only 30 ...before i hear massive dropouts ..
check DAWBENCH testing on gearsluts... .. dawbench audio interface testing methodic - best and most professional up to date ...
Roundtrip latency ,, overall perfomance on different latency ...
I'm aware of the disparity but i've found that Cubase reports the numbers more accurately than other DAWs or the audio interfaces driver app. It includes any safety buffers or other bits and bobs that are often missed off driver control panels. I'm not talking about hearing the latency, although that's a factor, i'm talking about listening to the audio so that i know the audio under test is a stable data stream - that's all. And yes i understand the difference between output latency and round trip latency. And it's not necessarily a doubling because input and output latencies can be different. And yes plug-in testing at different buffer sizes is very important. And yes RME interfaces can do a heck load more plug-ins than other interfaces at the same buffer size.
I do make all of these points in all of my performance test videos and the videos I've done on latency and i use the Dawbench test along with my own test projects. So we're on the same page mate :)
I get 9.30 toal ms latency on Windows 10 Surface Pro 6, at 96 samples on my Roland Quad Capture. How does it compare to the Audiofuse? The Audiofuse seems to have 4 ms less than the Quad Capture at the same sampke rate with no glitching, apparently, and even less if you are using the 2 nucleous Surface Pro 5. I am considering purchasing the audiofuse because of this and its built-in USB hub.
How are you measuring the latency?
@@MoltenMusicTech I I adjust the buffer, load Eastwest plugg-ins and check the ASIO menu numbers. The only thing is that in my case it generally works well (when loading medium-size projects) at 128 samples, and ideally at playback and recording at 256 samples, which ends up being more than 10 miliseconds, though. I should double-check for cracks and see up to which buffer size I can set it, loading a full project. Or do the same thing up for comparison. I have the Surface Pro 6 i5 model and the Roland Quad Capture. It would be interesting to check, because I would see if upgrading to Audiofuse is a good idea or not necessarily. I believe as you say that latency depends on both the dedicated driver/hardware combination and processor.
Confirmed: Roland Quad Capture at 96 samples. 44100 hz working well with playback and recording on Surface Pro 6 (i5u quad) shows 5.510 ms IN latency and 4.331 Output Latency. I would send you a screen capture but this does not allow me to do so. How does this compare to Audiofuse? It ios stable at 128 samples 44100 Hz, which shows 5.395 IN and 2.494 output latency. It seems Arturia is a bit faster, but not that much, unless your compouter is a Dual-core i5
It's an i5 dual core Surface Pro in the video and I show you the latency in the video yeah? I ask how you're measuring it because there can be all sorts of hidden "safe mode" latencies that are not shown in the control panel. Anyway, i don't have a SP6 setup anywhere to test it for you but as you can see from the video at 128 samples it's a tiny bit lower than yours.... which is also why i check because 96 samples should be lower than 4.331 which makes me think there are other buffers involved. But honestly for round-trip latency you're not going to get much lower than what you already have unless you go for an RME Audio interface.
@@MoltenMusicTech Now you mention it, there might be some extra buffer, yes. I will check, remove it and re-write the result. RME? I will have a look although anything lower than 10 ms works well for me. I guess if I re-adjust the buffer of my Quad Capture, it might be more than enough for my needs. Both interfaces seem to do the job, despite the mobile processor's limitations.
Why is the audio in this video so low? Had to turn my volume all the way up to listen... not complaining just making an observation.
It's those volume pixels chewing on the audio cables
Results seem to look really really impressive, but what happens when you start to pile on the plug-ins?
Didn't have the time to go that deep into it unfortunately. I had to grab an opportunity while it presented itself - there's always something else to test :)
+Molten Music Technology
EXCELLENT! Thanks for this update to your original review. 👍👍
Again, as I mentioned in the comments of your initial review, I always want to record or capture audio at 24/96 minimum, and preferably 24/192 in order to match the human auditory system's capability in the Time Domain, to reduce smearing & pre/post ringing artifacts created by the steep brick-wall cutoff filters that are used in the ADC & DAC process, and for an overall lower noise floor/advantage in gain staging.
All of your tests still seem to be at a 44.1kHz sample rate, correct? Again, let's see what happens to your RT Latency, Channel Count, Plug-In Count, and VI's at Higher Sample Rates!?!?
One reason why all of these USB audio interface manufacturers Still Insist that USB 2.0 is "more than adequate" in terms of bandwidth for good latency performance and 10+ Channels of simultaneous audio I/O is because they are still only referencing a 44.1/48kHz sample rate! 😒😕 :/
The standard in professional recording studios has been a minimum capture and workflow of 24/96 for AGES, with quite a few of them using 24/192 for 10+ years. With the technological advancements of today, why should we be theoretically limited to 44.1-48kHz?
Thank you!
No offense but, looking at your screen it's obvious that you are only changing the desired buffer size and the actual buffer size and output latency do not change after you go lower than 64 samples, so it's kind of a moot point since the cpu has the same amount of time to compute the output in 64, 32, 16 and 8 sample modes.
None taken. True enough the output latency is stuck at the lowest level of 1.5ms but that's not the only thing that changes when reducing the buffer size. It's also about the round-trip latency. But the point i was trying to make is that on the Surface it couldn't handle an output latency of 1.5ms whereas the desktop could - which is what Arturia had claimed. You also have to take into account that the reported numbers are not to be trusted. If you check out my review of the Komplete Audio 2 you can see the latency measuring I did and how far removed it is from the reported latency - and the KA2 appears to use the same drivers as the Audiofuse - ruclips.net/video/mA7_fcdx12I/видео.html
Hi Robin, I was wondering if you could do a review of the UAD Apollo thunderbolt on Windows machines. I've had it for months and have never been able to get it to be stable on my hp spectre x360 i7 8550u 16gb ram.
Laptops are troublesome i'm afraid. I don't think me reviewing it will have any impact on how well it works on your machine as I'd be doing it on a desktop. PC's haven't worked out Thunderbolt quite right yet unfortunately.
Molten Music Technology I fully understand what your saying. The only thing is every other piece of thunderbolt 3 hardware from monitors to hubs to GPU's works flawless in Windows 10. Its just the apollo thats buggy. Maybe its because of the Thunderbolt 2 to 3 conversion but, the guys at apollo say it should work problemless thats why I was wondering if you could do a tour on the apollo see if it works for you. Its great hardware maybe one of the best and there' s a ton of musicians out there either owning it or thinking about buying it. Anyway thanks for the response 👍
Ok then, send yours into me and I'll check it out
Molten Music Technology haha I wish I could but, we're based in the Netherlands and the other guys in the studio use it with their macs on a daily basis.
Thank you
Does the heat create problems after month of use?
Nope, still hot, still fine
if you are on windows XPto10 check out the Odeus asio link driver adon..all my interfaces are have now new and more fuctionality and are much more stable now!!!i dont now how they made that,but i never lookt back!??!Now i just look for roundtrip latency and connectivity!!!!!wood be awsome if you Rewiew that Killer app!