Do you already have the 2020 M1? Or if not, any spare computer on which you have tried or can try running Audirvana?
The issue is not really CPU core performance in the case of the Apple Inc. M series platforms⌠it is all about available RAM and the memory bus bandwidth in the context of Audirvana functionality.
I currently have it in a macbook pro 2012 with 8 gb of ram. Havenât bought any other yet.
Hi, thanks but what is that? Where should I see that info? And how much ram would be enough (meaning anything over it not adding in SQ)?
Look at the comparison chart that I posted previously and view the technical details regarding the M series Chip memory bandwidth specifications⌠Larger bandwidth is better⌠more RAM is better⌠Fundamentally, if you are looking to upgrade, configure your Mac to meet these requirements as budget will allow⌠M1 is fine, but for the future more than 16GB is going be useful if not today, very soon, as applications like Audirvana evolve to make use of the M series platforms.
The future Mac platforms are already moving into the M4 world.
Is there anything about the current sound quality or function that is unsatisfactory to you?
Thanks, thatâs helpful.
Not like that. But I have compared my macbook pro 2012 with a macbook mini 2012 using audirvana and there was a big difference. Hence my question. Besides, I need a new computer anywayâŚ
Either of the choices you gave me will run Audirvana just fine as long as youâve got at least 16GB of RAM (you may not need even that much with 24/192 files maximum). And I doubt youâll hear an SQ difference.
To me the decision would be more along the lines of how long you want to keep the new computer. Look at price difference over years of use. Plan to keep it until itâs 8 years old? Then youâve already used 4 of that on the 2020 model. I have to check, but I think some new models may have just come out, which would make the refurbished previous model a little cheaper on the Apple Store.
Ok, thatâs also helpful. Thank you.
Back to SQ, I ran top with the minimal Arch and Ubuntu Server installs just to have a very broad look at Audirvanaâs resource use.
At the beginning of the first selected track, while Audirvana is upsampling that track and the next to my DSD512 setting, CPU usage is approximately 230-280%. (With 6 cores, thatâs an average of a bit less than 40% to a bit less than 50% per core.) Then it settles way down, to a maximum of 3.7% on Ubuntu Server, and a maximum of 2.3% on Arch. Whether this has any bearing on my subjective preference for Arch, I canât say - weâre talking very minimal resource usage in both cases. When switching to the second track thereâs a brief blip up to 40-50%, then back down to minimal territory again.
Regarding memory, with total RAM of 32 GB and Audirvanaâs memory buffer set at a maximum of 15.6GB, RAM usage climbed to about 7% in both Arch and Ubuntu Server. So even with upsampling to DSD512, RAM usage wasnât heavy, though the two tracks I played werenât terribly long; there would be more RAM used for longer tracks.
@Jud Have you been able to play AS as a host (on Ubuntu server) with Gentooplayer based Diretta protocol as target?
Apparently a new linux AS is released but not sure they addressed the Dirretta issue. AS on Ubuntu using MPD/Upmpdcli works fine but I want to stay with Diretta.
Thanks
Hi, have never used Gentooplayer or Diretta protocol.
It seems to me the philosophies of Diretta and Audirvana are similar (reduce electrical noise through software), so perhaps there is a way to make them work nicely together. But my knowledge of software and Linux isnât nearly deep enough to do that on my own.
I confirm,
unfortunately, Audirvana still doesnât work with Diretta ALSA driver in the latest version of Audirvana for Linux. 
I hope we will have some good news after Munich High End. 
I suspect different ffmpeg library distributions are used, depending on the os Audirvana runs on. That will explain sonic differences. Not all, but it is an important aspect.
ArcoLinux matches Arch Linux, and I have both, so at least subjectively and with broad measurements of resource use I could see whether I would notice a difference not due to different versions of libraries.
Listened to both, and subjectively thought I liked the minimal Arch install with no desktop better. I felt imaging, soundstage, and clarity of instruments and vocals were better.
Resource use for the Audirvana process on top didnât seem hugely different, with a non-oversampling max of 4.3% with the desktop, 3.3% without. (Memory use was about the same between the two as one would expect.) Given this small difference, itâs certainly possible that my subjective impression of differences is all in my head.
Which is the only place it matters!
3 Likes
Do you normally upsample in Audirvana? I found it to be a game changer, but only up to 24/192 as thatâs the limit of my Rega DAC-R.
DSD512, the limit of my DAC. Measurements by the person who provided the DSD modulators for Audirvana, and also by the developer of HQPlayer, indicate DSD256 or DSD512 are the âsweet spotâ of lowest noise and distortion for DACs that can accept these rates. (I havenât read much about DSD1024 vs these other rates because my DAC isnât capable of that input rate.) Edit: I canât say whether weâd be capable of hearing the measured differences that exist.
24/192 should be helpful for your DAC, since Audirvana will substitute for the DACâs internal upsampling to some extent.
1 Like