Do you use audirvana on your main system as well?
Also, do you have a link for a tutorial (can be retarded level) oon the minimal linux install for the audirvana ? Iā would like to try to do it on a sd card on my macbook.
What Iāve described above is a setup where Audirvana runs on the main computer in my office, then via UPnP over optical Ethernet to the Fitlet3, which is connected to the DAC in the main system.
There are 5 operating systems on the main computer - Windows 11 and 4 Linux distributions, Xubuntu, ArcoLinux, Arch Linux, and Ubuntu Server - and Audirvana runs on all of them. I also have a MacBook Pro in my office where Audirvana for MacOS is running, then via UPnP over Wi-Fi to the Fitlet3 in the main system.
So everything I have described is running Audirvana to the main audio system.
I havenāt tried running Linux from an SD card and donāt know how snappy the response would be. What Iām doing is running Audirvana on reasonably fast hardware, then sending the result to a UPnP endpoint on which there isnāt much happening, in order to keep the endpoint as electrically quiet as possible.
Your research would be as informed as mine regarding how to install a running Linux system on an SD card (probably better, because youāre aware of what you donāt know, and Iām not), and whether it would be possible to run it from a MacBook. Do you already have Linux installed on the MacBook?
Iāve done it 10 years ago on a macmini. It worked, but I canāt remember a thing
1 Like
If you had to buy a macbook to use it just for music with audirvana, what would you do? Mine is very old and I need another machine anyway. I wonder whether say an M3 pro would be better , the same or worse than an M1 as far as noise goes. I know from experience that more ram memory can really improve things, but thatās all.
I think the problem if you wanted to use Linux is that it was much easier to install a solid working Linux on Intel Macs. On the newer ones with ARM chips it can be done, but that Linux distro (Asahi Linux) is still in the alpha/beta stage with random hardware not working.
If you simply wanted to run MacOS, thatās easy. Plenty of RAM (16 or even better 32GB) is better, and my M1 Max chip has no problem sending DSD512 over UPnP. (If the connection to the DAC is direct USB itās limited to half that, DSD256. MacOS via USB is limited to DoP which gives you half the ānative DSDā rate.)
I have an old macmini that can be used as endpoint. Should I use a unbutu version or the one youāre using is really preferable?
I donāt care about dsd as I use mostly qobuz. Besides, most of my local files are pcm. I suspect the more powerful the chip, the better, as the extra power will in theory take better care of whatever else is going on in the machine. However, maybe with the level of processing I need and audirvana to minimize extra activity, that becomes a moot issue (?).
Parenthetically, Iām so disgusted with Appleās decision to make the machines un-upgradable, that I donāt think I would spend more than $1k, unless I knew I was getting a major sonic improvement.
RAM and memory bandwidth and connectivity are going to be the primary considerations if you want the Mac to serve well into the future, in the context of digital-audio playback performance.
@Jonas This is a hyperbolic statement by @Jud based on a subjective interpretationā¦ Nothing is limitedā¦ There are no ānativeā DSD512 recordingsā¦ These DSD512 files are modulated (upsampled) iterations of lower sample-rate PCM and DSD recordingsā¦ His rationalization is that the higher-sample rate (22.4MHz) of DSD512 produces a subjectively perceivable tangible difference than that of the same file played at the sample-rate of(11.2MHz) DSD256ā¦
Since there are no objective assessments of subjective interpretation of perceivable difference in a native 11.2MHz DSD256 recording and the 22.4MHz DSD512 iteration of that recording, his statement regarding DoP transmission of a DSD256 file over USB versus a raw DSD512 iteration transmitted via Ethernet, is subjective hyperbole influenced by cognitive-bias that has him believing he is reducing some level of SDM filter (upsampling) related noise, that he believes he can detect. 
If you are modulating PCM at any sample-rate, the Nyquist cut-off Fs (frequency of sampling) is codified in the original recording A/D encodingā¦ The only thing a DSD512 iteration will have, is more zeroās stuffed into the fileā¦ For all intents and purposes these two files will be indistinquisable from one and otherā¦ The Nyquist cut-off Fs of a 11.2MHz DSD256 recording is 5.6MHz! ā¦ There are a myriad of other factors that will have greater influence on the sound-quality of either file.
If youāre not upsampling to DSD and you currently have an optical connection youāre happy with, I canāt see that running UPnP with a Mac Mini as an endpoint is going to do a lot for you. It might even wind up sounding worse to you. The major disadvantage of Toslink is limited bandwidth, but thereās enough for what youāre doing.
Yep, once theyāve got you, youāre got.
Thanks, youāre confirming my suspicion. Perhaps on a linux minimal setup installed in a small computer powered by battery the difference could be big enough to justify the hassle. The truth is that with audirvana in a dedicated macbook (remember that the battery is crucial with regular ethernet connection) with a basic config streaming via usb to spdif/ optical , the dac sees very little noise from the computer, on levels that probably compare to more sophisticated alternatives. I say that based on a comparison to a rpi with pi2 board used as roon end point, using the same optical connection.
Yes Jitter from optical cables is a real big shit that must be addressed, but hereās not the place for that discussion.
Thank you. I see what youāre saying, but thatās a discussion that doesnāt concern or interest me. Iām not being blasĆ©, I just prefer to spare everyone here my views on dsd and related matters. I also donāt do ETās/ufology, perverted sexuality and climate crap. A matter of taste. 
2 Likes
Yeah, thatās why the 10G/SFP+ interfaces, because the spec requires them to reduce jitter thatās above spec if they encounter it.
I would also say a high quality LPS might even outdo a battery with regard to sound quality, but in your case you already have battery power, and I wouldnāt see the added expense of the high quality LPS being worth it.
Yes, well power supplies can always make a big difference. But in my experience the best battery supply (difficult to make to eliminate battery noise, needs excellent āimpossibleā to find voltage regulators) surpasses the best linear ac supply. But even that generalization is hard to make, too many variables. In the case of the macbook, the battery is crucial to avoid ethernet noise leaking to the ac outlet where the system is connected. Huge difference, but only if the dac is electrically isolated from the sourceā¦
1 Like
Ethernet noise is really terrible. No matter how good a linear supply you use, with regular ethernet and un-isolated connection to the dac, thereās no music.
1 Like
Yep, before 10G optical throughout, I used an EtherRegen that I was very pleased with.
One further note: Since I thought things sounded better when using the main computer with a minimal Arch install, and that computer is electrically isolated from the DAC by multiple optical links, I thought to try a minimal Arch install in place of a minimal Ubuntu Server install for my UPnP endpoint, which is electrically isolated from the DAC by a single optical link. And indeed I think I like the sound even better now. 
Hi, so if one is not doing dsd, upsampling,etc, do you think it would still make sense to invest in something like a MB M1max, or a 2020 M1 machine would be good enough? I plan to have it dedicated for audio. I ask because the cash differece is hugeā¦