• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Review and Measurements of RME ADI-2 DAC

Blumlein 88

Major Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
11,637
Likes
17,224
#61
What are your interfaces? What do they record? Sounds interesting for me.
A Focusrite Forte. An Antelope Audio Zen Tour.

Both have microphone preamps, line level inputs, ADCs, and of course headphone out, monitor out and DACs.

Leaving off a number of other functions.
 
Joined
Apr 7, 2018
Messages
8
Likes
13
Location
DE
#62
I don't know how it can insert any filter if you use bit-exact output using ASIO.

That selection also has to do processor priority and not any kind of audio filtering. I searched but could not find any documentation on it.

The whole fidelizer thing is without value in my opinion. I ran all of my tests on my everyday laptop with tons of stuff running and as you see, performance was superb. This is what you pay for. High quality DACs are immune to computer activities. So you don't need anything that fidelizer does.
Its a real "misnomer" processor prio has nothing to do with an audio filter.

Normally its the process schedulers task in an operating system to distribute processes and threads across all active cores.

One guy from Microsoft claimed years ago in a big thread on gearsluts, that Microsoft will perform audio optimizations in Win10, proper priorization of audio threads and so on. For the professionals this was vapourware because all this was only available for the Windows sound system, but not for the Pro's using ASIO drivers for their equipment. And also for home use I pretty much prefer ASIO in the Windows world. With Apple its different, they don't have ASIO they have core audio.

If you want to get a much better performance on Windows with a lesser likelyness for audio drops then you can perform several things on your own without getting a software where you do not know what it is all doing to your system:

  1. disable energy saving in your BIOS, to park or wake up a core from a higher sleep state (i.e. C6) can last over 250 microseconds, this is a lot from process scheduler perspective. C-States to C0/C1 or off (depending on the BIOS)
  2. disable also all T- and P-States if present
  3. also to be turned off: C1E, also creates little latency inside of the system
  4. disable Spread Spectrum as it alternates the clock instantly which is usually not needed, browse the internet for what purpose this is
  5. Set the process scheduler to priotize background jobs, not the applications in the foreground, by this background processes participate which is important for audio applications
  6. EIST and Turbo I would keep although you can also nail the CPU to a fix frequency by turning this off. I keep it enabled because then I can control CPU clock by the power profiles of windows. There are nice Gadgets where you can choose then the Windows Energy Profile per Mouse click, also available for Win10 if you install classic shell to get the old startmenue back and Gadget support.
  7. disable CPU core parking, which BTW is also being performed by Steinbergs Cubase/Nuendo, when you use their fine tuned Energy Profile. You can download a tool from Bitsum which is called parkcontrol, which makes it easy to customize it. I one of my blog articles I described, how I customized this for my recording workstation (which is also for video editing, gaming, office, Internet etc). Written in German though, but I think the pictures and overviews speak for themselves, so that I hope that its useful for you too https://www.tonstudio-forum.de/blog/index.php/Entry/52-BIOS-Optimierung-Supermicro-X10SRi-F/
You can also put into consideration to use Bitsums tool ProcessLasso which is eventually the most advanced product to have a better customizebale process scheduler for Windows. I use this tool to automatically switch to energy saving mode, if I am for 60sec not active on the PC or to switch to a high performance mode once backups or special programs are running like steam (games) or audio applications.

Keep also in mind that Windows 10 had some defects .... up to Creators Fall there was a big issue in the memory subsystem makig Win10 much slower compared to Win7. Since creators fall this flaw has been solved. But still there is one bug present which blocks audio for ~200ms if certain Windows functions are not called with a certain option. Unluckily Microsoft didn't inform Devs about this change in their library functions and they even did not document. In the next early preview they wanted to have it fixed, but I fear for us folks it might last until perhaps Q3 until this might be fixed also for the normal Win10 update cycle. I put together this information in this article: https://www.tonstudio-forum.de/blog/index.php/Entry/84-Windows-10-better-than-Windows-7-EN-DE/
So .. if you have an audio application or drivers which do not honour this required programming, then you also will have this blocked audio behaviour. Companies like RME react on such things very quick, but there might be other companies who have not yet added this to their Win10 ASIO drivers or audio applications.
 

Rod

Addicted to Fun and Learning
Joined
Mar 24, 2018
Messages
707
Likes
302
#63
Its a real "misnomer" processor prio has nothing to do with an audio filter.

Normally its the process schedulers task in an operating system to distribute processes and threads across all active cores.

One guy from Microsoft claimed years ago in a big thread on gearsluts, that Microsoft will perform audio optimizations in Win10, proper priorization of audio threads and so on. For the professionals this was vapourware because all this was only available for the Windows sound system, but not for the Pro's using ASIO drivers for their equipment. And also for home use I pretty much prefer ASIO in the Windows world. With Apple its different, they don't have ASIO they have core audio.

If you want to get a much better performance on Windows with a lesser likelyness for audio drops then you can perform several things on your own without getting a software where you do not know what it is all doing to your system:

  1. disable energy saving in your BIOS, to park or wake up a core from a higher sleep state (i.e. C6) can last over 250 microseconds, this is a lot from process scheduler perspective. C-States to C0/C1 or off (depending on the BIOS)
  2. disable also all T- and P-States if present
  3. also to be turned off: C1E, also creates little latency inside of the system
  4. disable Spread Spectrum as it alternates the clock instantly which is usually not needed, browse the internet for what purpose this is
  5. Set the process scheduler to priotize background jobs, not the applications in the foreground, by this background processes participate which is important for audio applications
  6. EIST and Turbo I would keep although you can also nail the CPU to a fix frequency by turning this off. I keep it enabled because then I can control CPU clock by the power profiles of windows. There are nice Gadgets where you can choose then the Windows Energy Profile per Mouse click, also available for Win10 if you install classic shell to get the old startmenue back and Gadget support.
  7. disable CPU core parking, which BTW is also being performed by Steinbergs Cubase/Nuendo, when you use their fine tuned Energy Profile. You can download a tool from Bitsum which is called parkcontrol, which makes it easy to customize it. I one of my blog articles I described, how I customized this for my recording workstation (which is also for video editing, gaming, office, Internet etc). Written in German though, but I think the pictures and overviews speak for themselves, so that I hope that its useful for you too https://www.tonstudio-forum.de/blog/index.php/Entry/52-BIOS-Optimierung-Supermicro-X10SRi-F/
You can also put into consideration to use Bitsums tool ProcessLasso which is eventually the most advanced product to have a better customizebale process scheduler for Windows. I use this tool to automatically switch to energy saving mode, if I am for 60sec not active on the PC or to switch to a high performance mode once backups or special programs are running like steam (games) or audio applications.

Keep also in mind that Windows 10 had some defects .... up to Creators Fall there was a big issue in the memory subsystem makig Win10 much slower compared to Win7. Since creators fall this flaw has been solved. But still there is one bug present which blocks audio for ~200ms if certain Windows functions are not called with a certain option. Unluckily Microsoft didn't inform Devs about this change in their library functions and they even did not document. In the next early preview they wanted to have it fixed, but I fear for us folks it might last until perhaps Q3 until this might be fixed also for the normal Win10 update cycle. I put together this information in this article: https://www.tonstudio-forum.de/blog/index.php/Entry/84-Windows-10-better-than-Windows-7-EN-DE/
So .. if you have an audio application or drivers which do not honour this required programming, then you also will have this blocked audio behaviour. Companies like RME react on such things very quick, but there might be other companies who have not yet added this to their Win10 ASIO drivers or audio applications.
Thank you. I will do this for the RME dac.
 

Wombat

Major Contributor
Joined
Nov 5, 2017
Messages
6,726
Likes
6,141
Location
Australia
#64
Its a real "misnomer" processor prio has nothing to do with an audio filter.

Normally its the process schedulers task in an operating system to distribute processes and threads across all active cores.

One guy from Microsoft claimed years ago in a big thread on gearsluts, that Microsoft will perform audio optimizations in Win10, proper priorization of audio threads and so on. For the professionals this was vapourware because all this was only available for the Windows sound system, but not for the Pro's using ASIO drivers for their equipment. And also for home use I pretty much prefer ASIO in the Windows world. With Apple its different, they don't have ASIO they have core audio.

If you want to get a much better performance on Windows with a lesser likelyness for audio drops then you can perform several things on your own without getting a software where you do not know what it is all doing to your system:

  1. disable energy saving in your BIOS, to park or wake up a core from a higher sleep state (i.e. C6) can last over 250 microseconds, this is a lot from process scheduler perspective. C-States to C0/C1 or off (depending on the BIOS)
  2. disable also all T- and P-States if present
  3. also to be turned off: C1E, also creates little latency inside of the system
  4. disable Spread Spectrum as it alternates the clock instantly which is usually not needed, browse the internet for what purpose this is
  5. Set the process scheduler to priotize background jobs, not the applications in the foreground, by this background processes participate which is important for audio applications
  6. EIST and Turbo I would keep although you can also nail the CPU to a fix frequency by turning this off. I keep it enabled because then I can control CPU clock by the power profiles of windows. There are nice Gadgets where you can choose then the Windows Energy Profile per Mouse click, also available for Win10 if you install classic shell to get the old startmenue back and Gadget support.
  7. disable CPU core parking, which BTW is also being performed by Steinbergs Cubase/Nuendo, when you use their fine tuned Energy Profile. You can download a tool from Bitsum which is called parkcontrol, which makes it easy to customize it. I one of my blog articles I described, how I customized this for my recording workstation (which is also for video editing, gaming, office, Internet etc). Written in German though, but I think the pictures and overviews speak for themselves, so that I hope that its useful for you too https://www.tonstudio-forum.de/blog/index.php/Entry/52-BIOS-Optimierung-Supermicro-X10SRi-F/
You can also put into consideration to use Bitsums tool ProcessLasso which is eventually the most advanced product to have a better customizebale process scheduler for Windows. I use this tool to automatically switch to energy saving mode, if I am for 60sec not active on the PC or to switch to a high performance mode once backups or special programs are running like steam (games) or audio applications.

Keep also in mind that Windows 10 had some defects .... up to Creators Fall there was a big issue in the memory subsystem makig Win10 much slower compared to Win7. Since creators fall this flaw has been solved. But still there is one bug present which blocks audio for ~200ms if certain Windows functions are not called with a certain option. Unluckily Microsoft didn't inform Devs about this change in their library functions and they even did not document. In the next early preview they wanted to have it fixed, but I fear for us folks it might last until perhaps Q3 until this might be fixed also for the normal Win10 update cycle. I put together this information in this article: https://www.tonstudio-forum.de/blog/index.php/Entry/84-Windows-10-better-than-Windows-7-EN-DE/
So .. if you have an audio application or drivers which do not honour this required programming, then you also will have this blocked audio behaviour. Companies like RME react on such things very quick, but there might be other companies who have not yet added this to their Win10 ASIO drivers or audio applications.

That is a lot of advice. Since you are new here, can you give us some creds. to back it up?
 

Rod

Addicted to Fun and Learning
Joined
Mar 24, 2018
Messages
707
Likes
302
#66
Its a real "misnomer" processor prio has nothing to do with an audio filter.

Normally its the process schedulers task in an operating system to distribute processes and threads across all active cores.

One guy from Microsoft claimed years ago in a big thread on gearsluts, that Microsoft will perform audio optimizations in Win10, proper priorization of audio threads and so on. For the professionals this was vapourware because all this was only available for the Windows sound system, but not for the Pro's using ASIO drivers for their equipment. And also for home use I pretty much prefer ASIO in the Windows world. With Apple its different, they don't have ASIO they have core audio.

If you want to get a much better performance on Windows with a lesser likelyness for audio drops then you can perform several things on your own without getting a software where you do not know what it is all doing to your system:

  1. disable energy saving in your BIOS, to park or wake up a core from a higher sleep state (i.e. C6) can last over 250 microseconds, this is a lot from process scheduler perspective. C-States to C0/C1 or off (depending on the BIOS)
  2. disable also all T- and P-States if present
  3. also to be turned off: C1E, also creates little latency inside of the system
  4. disable Spread Spectrum as it alternates the clock instantly which is usually not needed, browse the internet for what purpose this is
  5. Set the process scheduler to priotize background jobs, not the applications in the foreground, by this background processes participate which is important for audio applications
  6. EIST and Turbo I would keep although you can also nail the CPU to a fix frequency by turning this off. I keep it enabled because then I can control CPU clock by the power profiles of windows. There are nice Gadgets where you can choose then the Windows Energy Profile per Mouse click, also available for Win10 if you install classic shell to get the old startmenue back and Gadget support.
  7. disable CPU core parking, which BTW is also being performed by Steinbergs Cubase/Nuendo, when you use their fine tuned Energy Profile. You can download a tool from Bitsum which is called parkcontrol, which makes it easy to customize it. I one of my blog articles I described, how I customized this for my recording workstation (which is also for video editing, gaming, office, Internet etc). Written in German though, but I think the pictures and overviews speak for themselves, so that I hope that its useful for you too https://www.tonstudio-forum.de/blog/index.php/Entry/52-BIOS-Optimierung-Supermicro-X10SRi-F/
You can also put into consideration to use Bitsums tool ProcessLasso which is eventually the most advanced product to have a better customizebale process scheduler for Windows. I use this tool to automatically switch to energy saving mode, if I am for 60sec not active on the PC or to switch to a high performance mode once backups or special programs are running like steam (games) or audio applications.

Keep also in mind that Windows 10 had some defects .... up to Creators Fall there was a big issue in the memory subsystem makig Win10 much slower compared to Win7. Since creators fall this flaw has been solved. But still there is one bug present which blocks audio for ~200ms if certain Windows functions are not called with a certain option. Unluckily Microsoft didn't inform Devs about this change in their library functions and they even did not document. In the next early preview they wanted to have it fixed, but I fear for us folks it might last until perhaps Q3 until this might be fixed also for the normal Win10 update cycle. I put together this information in this article: https://www.tonstudio-forum.de/blog/index.php/Entry/84-Windows-10-better-than-Windows-7-EN-DE/
So .. if you have an audio application or drivers which do not honour this required programming, then you also will have this blocked audio behaviour. Companies like RME react on such things very quick, but there might be other companies who have not yet added this to their Win10 ASIO drivers or audio applications.
Well the only thing for me to change was background app processes. Spread spectrum in the bios is set to auto and greyed out so I cant change it. All 4 cores are active and in parkcontrol shows 100%. C states was off already. Thanks again. I uninstalled Fidelizer Pro.
 
Joined
Apr 7, 2018
Messages
8
Likes
13
Location
DE
#67
"C-States to C0/C1 or off" this is literally bro-science
Sorry, how can I help ? I don't understand what you mean. Dissatisfied with my proposal or are you not used to fine tune the BIOS where maybe applicable ?
 

Purité Audio

Major Contributor
Industry Insider
Barrowmaster
Forum Donor
Joined
Feb 29, 2016
Messages
4,601
Likes
3,319
Location
London
#68

Veri

Major Contributor
Joined
Feb 6, 2018
Messages
7,439
Likes
8,811
#69
Sorry, how can I help ? I don't understand what you mean. Dissatisfied with my proposal or are you not used to fine tune the BIOS where maybe applicable ?
Turning off C-states will not magically make your processor better at handling processing loads, audio or not. Utilization goes up as the CPU is stressed, running at a static state is pointless, will consume more power and will not change anything in regard to stuttering or bugs.

And before you tell me to actually try it out, I did tinker with C-state before and did not notice any "benefit" whatsoever.
 
Joined
Apr 7, 2018
Messages
8
Likes
13
Location
DE
#70
That is a lot of advice. Since you are new here, can you give us some creds. to back it up?
Hi Wombat, maybe sorry for this "wall of text" as a new member ;) If you ask for references, well in my current job I am advanced network technician, but am also very experienced with servers / computers in the Unix and Windows area (so to say since Microsoft Xenix). In my younger days I programmed myself and supported one of the free Unix projects, where Apple's "userland programs" have its roots from). I am also hobby musician and recording/playing guitar. This brought me to the area of computer optimization for recording, because with my former combination of computer and recording interface (Focusrite/Firewire) I had issues with audio drops. So I spend a lot of time with optimization of Windows PCs for recording projects up to >100 channels for recording and playback at lowest latencies (my 400 tracks test project). My Windows 7 recording workstation I fine tuned up to a point where I can reach lowest DPC latencies down to only a few microseconds, like you can see here in this screenshot.
 
Joined
Apr 7, 2018
Messages
8
Likes
13
Location
DE
#71
Turning off C-states will not magically make your processor better at handling processing loads, audio or not. Utilization goes up as the CPU is stressed, running at a static state is pointless, will consume more power and will not change anything in regard to stuttering or bugs.

And before you tell me to actually try it out, I did tinker with C-state before and did not notice any "benefit" whatsoever.
I understand your feelings, BIOS is not the area where we like to spend our time on and tbh it was also not my favourite in the past ;)

C-States are responsible to control, up to which point the CPU is allowed to enter sleep states. The higher the sleep state is, the more time it takes for the CPU cores to wake up and to process jobs (and this is the area where latency critial applications like audio can fail depening on some other surrounding conditions like i.e. system load, latency of drivers, etc). Most default BIOS settings allow the CPUs cores to reach C6, which is the highest sleep state. In the internet I found information, that the wakeup of a CPU core from C4E/C5 takes already around 250 microseconds time with C6 even more than 250 microseconds, see this table:


For a PC on process level this is already a lot of time. Some tuning tools being used in the recording area report alarm when latency spikes of around 1000 microseconds are being measured and report that a PC is not suitable for recording (-> LatencyMon, DPC Latency Checker). Depending on your system load (what is already installed and whats currently running in the background of your system) it can already make a significant difference if you fine tune in this area to disable everything which is related to energy saving in your BIOS. When avoiding such high sleep levels your system can react much quicker / agile on new processing load.

So, if somebody should have already problems with audio drops, maybe especially when some other programs are running , then this kind of tuning can give to your system more headroom to proactively prevent audio loss, especially under load.

I hope this clarifies a little bit and again sorry if I post such PC details here and overwhelmed you with "my love" ;) I simply noticed how well and up to which detail the DAC measuring take place in this forum, that I maybe completely forgot that this is no PC hardware forum. I had only the best intentions to give you my support on this topic when I read about audio loss .. because I spend already so much time on this topic.
 
Last edited:

bennetng

Major Contributor
Joined
Nov 15, 2017
Messages
1,311
Likes
1,226
#72
Do notice that "performance" and "quality" is not quite the same. For example a DAW geek may desire lowest possible latency, maximum amount of tracks, plugins and virtual instruments but an audiophile may need for example, better looking graphs in AP.

Also, while some OS or BIOS tweaks may solve some problems in a problematic system, every computer system have platform specific stuff. Things work on this system may not work on another one. For example someone experienced audio drop-outs and some of us also suggested some latency/power/BIOS tweaks but none of these work until some hardware replacement.

https://hydrogenaud.io/index.php/topic,114138.0.html

In my own experience with the a C2D, 1st and 4th Gen Core i system, disabling higher C-State can sometimes improve DAW performance in some projects, but I never disable C1E. While it may be measurable in terms of nanoseconds but not a single time disabling C1E can make a originally stuttered project become smooth. What I did notice is the CPU became hotter and the fan spin faster.
 

Veri

Major Contributor
Joined
Feb 6, 2018
Messages
7,439
Likes
8,811
#73
Likewise, for audio-listening, and not DAW low-latency work, a regular OS install (and necessary drivers if they're needed) + a music player should be all you need.
 

Dro

Active Member
Joined
Apr 8, 2018
Messages
105
Likes
59
#74
Thank you for this review. Have you also measured THD+N vs frequency?

Sort of OT, but certainly related to this DAC: I am curious whether the AK4490 is a 1-bit DS DAC or multibit DS. I could not find anything about this in the spec sheet. It does however contain some information about the four available filters at 44.1 and 96 kHz.
 
Joined
Apr 7, 2018
Messages
22
Likes
5
#77
Listening impressions are in the original review. This loaned unit is brand new so I don't think the owner will give me permission to open it up. I will tear open my ADI-2 Pro though and see what is inside.
There is a normal ASIO driver from RME? :)Why do you need a poor-quality WASAPI built in WIN? Insert into your review for the sake of completeness and correct it in order not to confuse people.
ADI-2_Pro_on_Roon.PNG
 

Frank Dernie

Major Contributor
Forum Donor
Joined
Mar 24, 2016
Messages
5,467
Likes
12,205
Location
Oxfordshire
#78
There is a normal ASIO driver from RME? :)Why do you need a poor-quality WASAPI built in WIN? Insert into your review for the sake of completeness and correct it in order not to confuse people.
View attachment 11940
Amir has explained his reasoning frequently enough, whether one accepts it or not is another thing. Try checking what role Amir had as a senior engineer at Microsoft back in the day.
I gave my PC away when I retired and don't use computer music much myself.
 

Veri

Major Contributor
Joined
Feb 6, 2018
Messages
7,439
Likes
8,811
#79
There is a normal ASIO driver from RME? :)Why do you need a poor-quality WASAPI built in WIN? Insert into your review for the sake of completeness and correct it in order not to confuse people.
View attachment 11940
Excuse me? WASAPI is in general, flawless and bit-perfect. I don't get people's obsession with ASIO. That it doesn't work with all sample rates with the RME ADI-2 DAC is rather an exception as confirmed by their customer support and has nothing to do with "poor quality" so there's no reason to call it so (unless you have proof that ASIO is in fact superior?).

Secondly, Amir has in fact stated that ASIO drivers are necessary for flexible functioning.

In this case I had selected 88.2 kHz and that is the only sample rate Roon sees as a result. Playing anything then forced format conversion to this rate.

I contacted RME customer support in US and seemingly this is the way it works. This is quite disappointing as I don't understand why this way of operation is useful at all.

What this means is that practically speaking, you need to use ASIO drivers to have flexible playback under Windows. I don't know how it works under Linux of Mac. Hopefully it is better than in Windows and plug-and-play works.
Others have reported that under OSX that there are no problems.
 
Last edited:
Top Bottom