• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Review and Measurements of Benchmark DAC3

OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,596
Likes
239,661
Location
Seattle Area
IMHO, the manufaturer provided driver is part of the product and it certainly makes sense to test with it as that is how the manufacturer intended the product to be used. Testing the product with WASAPI also makes sense as that is the standard to which we expect manufaturers to comply as well, while testing with ASIO4ALL is eh, well..
There are manufacturers like Schiit that don't provide any ASIO drivers. To test and compare them to anything else I would have to use ASIO4ALL. If I used native drivers for the others, then people would complain that the two were not tested the same.

We are a science and objective forum. Let's see the objective data that ASIO4ALL does something to the bits. I will then use that evidence to get Audio Precision to support WASAPI. Without it, they tell me to use ASIO4ALL and it is hard for me to argue otherwise if the measurement are the same and they don't have to do extra development.
 

gvl

Major Contributor
Joined
Mar 16, 2018
Messages
3,474
Likes
4,072
Location
SoCal
Last time I installed a Schiit driver it included an ASIO interface, it used some weird name though so it was not immediately obvious what it was when looking at the list of ASIO devices, if memory serves, it has been a while...
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,067
Location
Zg, Cro
There are manufacturers like Schiit that don't provide any ASIO drivers. To test and compare them to anything else I would have to use ASIO4ALL. If I used native drivers for the others, then people would complain that the two were not tested the same.

I believe that the fact that some manufacturers don't provide ASIO driver shouldn't be used as an argument not to use ASIO driver provided by the other manufacturer. I think all product should be used according to the manufacturers instructions, so use the driver if one has been provided or go without it if not.

We are a science and objective forum. Let's see the objective data that ASIO4ALL does something to the bits. I will then use that evidence to get Audio Precision to support WASAPI. Without it, they tell me to use ASIO4ALL and it is hard for me to argue otherwise if the measurement are the same and they don't have to do extra development.

If I remember correctly ASIO4ALL is an ASIO wrapper for a kernel streaming, so it basically provides ASIO command set which are then translated and passed to WDM audio driver. I don't see how that could affect say linearity of the DAC.

I believe usage of manufacturer provided ASIO driver is also advised by companies who developed players (JRriver included).

https://wiki.jriver.com/index.php/Audio_Output_Modes

Choosing the Audio Output Mode
On OS X MC uses Core Audio to access your audio device, which provides bit-perfect playback, and requires no configuration. On Windows, Media Center will typically choose the best Output Mode available for your device, based on the drivers installed. If given the choice between multiple modes or drivers, choose the best output for your hardware in this order:

1) If your hardware has a native, well-behaved, ASIO driver, use ASIO.

2) Otherwise, on Windows Vista, Windows 7, or newer, use WASAPI Exclusive Access.

3) Otherwise, use Kernel Streaming if it works.

4) If none of the above are possible, use DirectSound or WaveOut. Neither of these provide hardware direct output, so choose based on performance.

If you have problems, start with DirectSound and get MC working. Once you have audio working and are comfortable using MC, then experiment with other audio output choices.

ASIO
See also: ASIO
ASIO is a sound card communication system created by Steinberg.

If your soundcard has a native ASIO driver, this is the most direct (and normally best) way to communicate with it.

Please note that ASIO4All is basically Kernel Streaming with an ASIO wrapper. There's no good reason to use ASIO + ASIO4All vs just using Kernel Streaming directly in Media Center.

WASAPI (Windows Audio Session API)
See also: WASAPI
WASAPI is a hardware communication system in Vista, Windows 7, and newer. When used in exclusive mode, the Windows system mixer is bypassed, but other applications cannot simultaneously use the audio device.

Kernel Streaming
Kernel Streaming is a hardware direct way to speak directly to a WDM audio driver. It works on XP, so is useful when the above two options are not possible.
 
Last edited:

Sal1950

Grand Contributor
The Chicago Crusher
Forum Donor
Joined
Mar 1, 2016
Messages
14,169
Likes
16,879
Location
Central Fl
IMHO, the manufaturer provided driver is part of the product and it certainly makes sense to test with it as that is how the manufacturer intended the product to be used. Testing the product with WASAPI also makes sense as that is the standard to which we expect manufaturers to comply as well, while testing with ASIO4ALL is eh, well..

Installing so many drivers will eventually make a horrible mess of your PC but maybe creating a restore point before installing a driver and performing a system restore after you're finished can keep the situation under control.
What a mess!
Friends don't let friends do Windoz. ;)
 

SIY

Grand Contributor
Technical Expert
Joined
Apr 6, 2018
Messages
10,483
Likes
25,238
Location
Alfred, NY
So far, with one exception (CEntrance), I haven't had any issues with multiple ASIO drivers. And the CEntrance, to be fair, just crashes test software but plays fine with foobar, Audition, and JRiver.
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,067
Location
Zg, Cro
What a mess!
Friends don't let friends do Windoz. ;)

LOOL :D

I'm on a thin ice here as Amir worked for Microsoft and i worked for Oracle, so I'm trying to be helpfull here as much as I can, although my audio chain now uses Raspberry running Volumio instead of Windows. ;)
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,596
Likes
239,661
Location
Seattle Area
Last time I installed a Schiit driver it included an ASIO interface, it used some weird name though so it was not immediately obvious what it was when looking at the list of ASIO devices, if memory serves, it has been a while...
That was the case before. With Windows 10 Creators Edition, you can no longer install those drivers. I tried and it didn't work. And researching the same answer pops up that you must use in-box class drivers.
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,067
Location
Zg, Cro
What a mess!
Friends don't let friends do Windoz. ;)

I guess this is one of the reasons why you said that, right? :D

"With Windows 10 Creators Edition, you can no longer install those drivers. "
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,596
Likes
239,661
Location
Seattle Area
I should say that the reliability of the latest versions of ASIO4ALL has surprised me across so many devices I test. In the past, I hated it because it would rarely work when I needed it.

One note: you must optimize the buffering to get correct results. Using stock parameters for kernel buffers (2), will generate unstable results due to buffer underruns and create tons and tons of distortion in the measurements. This shows up easily in the "Dashboard" measurements I use so I always catch it and set the buffer to 4. Once there, it is rock solid.
 

Sal1950

Grand Contributor
The Chicago Crusher
Forum Donor
Joined
Mar 1, 2016
Messages
14,169
Likes
16,879
Location
Central Fl
I guess this is one of the reasons why you said that, right? :D

"With Windows 10 Creators Edition, you can no longer install those drivers. "
I run Linux here, PCLinuxOS to be exact.
Purchased my first USB DAC 3 years ago.
Plugged it in and it worked. :D
 

gvl

Major Contributor
Joined
Mar 16, 2018
Messages
3,474
Likes
4,072
Location
SoCal
That was the case before. With Windows 10 Creators Edition, you can no longer install those drivers. I tried and it didn't work. And researching the same answer pops up that you must use in-box class drivers.

My experience even before W10 Schiit drivers required a prior voodoo dance before they install successfully, sounds like things went further south since then.
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,067
Location
Zg, Cro
I run Linux here, PCLinuxOS to be exact.
Purchased my first USB DAC 3 years ago.
Plugged it in and it worked. :D

Of course it did. And you probably don't have to worry if it will continue to work once your Linux upgrades itself. Btw, I'm hoping Thomas is also not using Windows otherwise will soon get warned for this Windows bitching. :D
 

Thomas savage

Grand Contributor
The Watchman
Forum Donor
Joined
Feb 24, 2016
Messages
10,260
Likes
16,305
Location
uk, taunton
Of course it did. And you probably don't have to worry if it will continue to work once your Linux upgrades itself. Btw, I'm hoping Thomas is also not using Windows otherwise will soon get warned for this Windows bitching. :D
Windows aptly named as using it makes you want to jump out a window..
 

HifiGuy

Member
Joined
Jun 25, 2018
Messages
20
Likes
4
Thats why i use Debian Linux. There is no need for drivers for ALSA und it provide a bitperfect playback system if it is configured right. I still think about all measurements of amir, wheather their right or not.
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,067
Location
Zg, Cro
Thats why i use Debian Linux. There is no need for drivers for ALSA und it provide a bitperfect playback system if it is configured right. I still think about all measurements of amir, wheather their right or not.

+1 for ALSA - Voluimo also uses it.
 

Sal1950

Grand Contributor
The Chicago Crusher
Forum Donor
Joined
Mar 1, 2016
Messages
14,169
Likes
16,879
Location
Central Fl
Thats why i use Debian Linux. There is no need for drivers for ALSA und it provide a bitperfect playback system if it is configured right. I still think about all measurements of amir, wheather their right or not.
What media player are you using?
 

rebbiputzmaker

Major Contributor
Joined
Jan 28, 2018
Messages
1,099
Likes
463
That was the case before. With Windows 10 Creators Edition, you can no longer install those drivers. I tried and it didn't work. And researching the same answer pops up that you must use in-box class drivers.
I hope it is ok to ask.

Why were you not able to install a driver? Surprising that a driver would be provided that cannot be installed.
 

John_Siau

Active Member
Technical Expert
Audio Company
Joined
Jul 17, 2018
Messages
186
Likes
1,420
Location
Syracuse, NY USA
Yes, the unit was returned to his owner. As Thomas says, I am happy to test another unit if Benchmark sends me one. I will test it with their drivers and ASIO4ALL and report. And of course any difference in jitter performance.

I looked at Benchmark Jitter test in the manual and unfortunately it is done completely different then mine. It uses a 10 kHz tone, bandpass filtered unlike my measurements that are done with J-test signal. What difference that can make, I don't know. Here is their results for reference:

View attachment 14004
Amir,

This test is actually done the same way that yours was. The only difference is that the test tone was 10 kHz instead of 12 kHz. The sidebands would show up equally well with either test. The notch filter is automatically inserted on the 555 and then a second FFT without the notch is spliced onto the first FFT so that the test tone can be displayed. The notch filter in the AP2722, AP2522 and the 555 is necessary in order to prevent distortion in the A/D converter at the front end of any of these three test stations. With the 555, the notch filter is just inserted automatically. I can show you FFT tests from an AP2722 where two FFTs are superimposed so that the entire spectrum can be viewed (just like the final results from a 555). But here is something better:

Jude from head-fi.org also has a 555 and a DAC3. He replicated your tests and saw no evidence of any jitter-induced sidebands. His results agree with ours. His results can be seen here:

https://www.head-fi.org/threads/schiit-yggdrasil-impressions-thread.766347/page-591#post-14366380

It is important to figure out what went wrong with your tests. The results you posted are very strange and do not agree with the results obtain by us or by other reviewers.
 
Top Bottom