WTF with memory buffer allocation? Origin or Studio

Origin
1

Your Max Memory for Audio Buffering
is set at 62 GB out of 64 GB ram, reduce this to ± 4GB

Removed that quote…

That amount of memory in Audirvana preferences is for BUFFERING the song into memory for Audirvana then to play it in memory (meaning the software then go in sleeping mode, not working)… that as nothing to do with how much memory an application use it like Photoshop asking to use 60% of your memory inside your computer… it is an application that use a lot of ram on NOW… not Audirvana, that preference is only for BUFFERING the song… it uses 12gb of your 16 real GB of RAM to buffering the song… it took like 2 seconds to buff a song at 24/96 for a 10 minutes song… after 2 seconds, Audirvana is sleeping again… Easily seeing that in activity monitor… it is NOT still using that number of GB to use the software… JUST buffering.

The FASTEST the song is BUFFERED,
the faster the software is in sleeping mode after…
Damien set this feature by default not for nothing…

YES, maybe if you are using Studio and UPSAMPLING or the 2 combined, you should set it a little lower but not to less then too much…

SO EASY, to demonstrate that… record your screen if you can’t remember…

Select a 10 minutes song at a higher bit rates like 24/96 or better 24/192 if you can… if you have 8gb of memory set the buffer to 5gb, if you have 16gb set the buffer to 12gb, if you are like me, have 20gb, set it to 17gb (my number :slight_smile: ) … if you have 64gb of memory, man! you think you should care?? set it at 50gb, if you are scared :slight_smile:

Open Activity Monitor or something like it on PC…

Play that long 24/96 or 24/192 and see with lots of memory like Damien default settings what happens… You see the song is buffed rapidly while just starting to play… GOOD!, no? Yes, then Audirvana after 2 seconds is playing from memory and Activity monitor is at zero or less then 2% on a Mac Mini 2018…

Then play the same song with the buffer memory at 4gb instead of 12gb out of 16 as an example… what you see? the line for buffering the song is slower to reach the end… meaning? it plays, it buff, it plays, it buff, it plays it buff… you think your computer is at is BEST now, no, compare to the time it takes before… 2 seconds then sleep…

Then try again the same song, stay at 4gb out of 16 ram memory buffed in Audirvana and upsampled that same song to DSD128… now you see your computer is playing the song, upsampling the song, still buffering the song, playing the song, upsampling the song, still buffering the song, playing the song, upsampling the song, still buffering the song, till it reaches the end of the 10 minutes song (not even talking about Audirvana using 300% of your CPU compared to another software while doing all that)…

So much better then seeing at first your computer software Audirvana sleeping after 2 seconds??… YOU decide.

YES, if you use Studio, with internet playing a song at 24/192, using downloading while praying you are REALLY listening to a 24/192 from Internet (yes it show that bitrate in Audirvana, then you are safe :slight_smile: ) you’ll see that in 10 years you were listening to mp3 :slight_smile: all that time… it can use more memory and maybe you should lower that buffer a little more then if you are using only your own files with Origin, but never to see the buffering line while playing taking too much time…

The less time it takes to buffered a song, The better you computer (Audirvana software) is resting fast and playing from memory with the Lowest computer activity…

My 2 cents.

2

Depends on the files you are playing and their size. With my buffer set at 4GB Audirvana in the loading decoding tells me that is sufficient ro preload 2hrs 14 m of music at 44.1 kHz and nearly 31 minutes at 192 kHz. Obviously playing at high rate DSD files will use more. The goal is to be able to let Audirvana pre-load the track into RAM before playback as such even with high rate DSD files or upsampling to 768 kHz 6 or 8GB is enough. (See screenshot of loading decoding setting for AS 2.7.0)

Audirvana by default on my Mac sets the max memory allocation to 125 GB out of 128 GB total Ram. That figure is nonsense… I have no music passages, even on classical DSD files that would use more than 10GB to load the longest track.

Screenshot 2024-03-09 at 13.39.50
Screenshot 2024-03-09 at 13.39.50838×458 42.1 KB

3

FOR BUFFERING, not using that ram always…

with 128gb of real ram… i wonder why you replied to that thread??
You have bug? your music jump?
you can’t have problems for that kind of much RAM…
must be something else if you have buggy software for sure…

4

Nope, I never have issues playing DSD128 classical tracks at 30/40 minutes length. Apologies if I miss interpreted your post but I was under the impression you were suggesting that you should leave the buffer at what AS assigns by default? My bad.

5

PLEASE don’t show me anything that i told anyone to do first…

Do the 3 things i have shown first…

real memory buffer set as Audirvana default
same song with lower buffer
same song with same lower buffer with upsampling…
same lower buffer song with upsampling, then with studio, with internet adding slowest adding

then, tell me you are right…

6

YES i was telling to let the buffer as it was… unless you were with Studio or…

7

OK apologies for miss interpreting.

8

I HAVE NEVER had a problem with buffer at 17gb out of my 20gb real memory
playing any files at any upsampling rate to my DAC max DSD128, NEVER it stop or reepeeeaaatt itself…

9

just try the 3 steps i told and see for yourself…
what takes more resources form your Mac or PC, then decide…

YOU with that ram never should think about anything…
even with the MAX filters ressources of HQPlayer :slight_smile:

10

What you perceive is not always reality,… Your interpretation of ‘never having a problem’ is a matter of subjective interpretation… You may not perceive the contextual qualitative elements that I do on my system and the behavior of my system when allocating more System RAM than is needed to play a track or tracks loaded into the pre-load memory…

“Buffering” can be interpreted in a couple of different ways… one way is how you are interpreting the actions in the “Bufferized” module of Audirvana… The other way is to see these as separate actions requiring buffers or ‘registers’ for the process DSP calculations…

These simple yet important contextual elements of the playback engine, produce audible influences on the sound-quality of the file playback scenario… You may not be able to discern these subtleties on your playback system and therefore, you have been captured by NaÏve Realism that has you believing what you perceive is what everybody else perceives…

:notes: :eye: :headphones: :eye: :notes:

11

Yes, i quote you here for last… call the Internet police…

The Guy/Girl that converted his/her RAW DSD files to PCM and tell me that after applying is plugin counts on Audirvana to Upsampling is PCM now tracks to DSD 128 will be as is if was not downsampled first to PCM… ARF!!!

Oh, i forget you have that DSD chip on your DAC that separate the venue that makes PCM track as regular RAW DSD tracks… Clap, clap…

i wish my other thread was seen before i could reply here…

Music is the Best - Frank Zappa

12

What is your point? :thinking:

13

Bye, wish you the best…

14

For your edification…
I have never, ever, in any of my responses or related topics, expressed that the DSD files decimated to DXD (24/352.8kHz) that I apply HRTF DSP to, and then modulate again, to DSD128, are the same or “as if” it were identical to the ‘mother’ native DSD file…

I am fully aware of the qualitative sound-quality differences of these two playback sources (DSD ‘mother’ and DXD ‘child’) in juxtaposition…
I discern those qualitative differences… The application of HTRF in the context of the DXD ‘child’, is facilitative in nature, and does present an audible and subjective difference in qualitative sound, as juxtaposed to the DSD ‘mother’ where HRTF DSP cannot be applied,

So, I still don’t get your point…
:notes: :eye: :headphones: :eye: :notes:

15

Should i respond again… Mmm, getting late, OK just before going to bed :slight_smile:

It is unbelievable that we can’t resist to respond to you :slight_smile:
here, you are the best of :slight_smile:

For your edification…
I have never, ever , in any of my responses or related topics, expressed that the DSD files decimated to DXD (24/352.8kHz) that I apply HRTF DSP to, and then modulate again, to DSD128, are the same or “as if” it were identical to the ‘mother’ native DSD file…

No you never, ever tell that in those exactly words, you are too bright for that.

BUT it is what you do in fact with your RAW DSD tracks, you do transfer them in PCM to apply something to them, then, you trust Audirvana to bring back them to DSD128 with your DAC chip that separated the path to a most sophisticate path compare to other DAC chip… you know, that bring so much more sounded files… no?

I don’t care what you do to your files, or hear, YOU take real DSD files that you own and not pirate… then turn them for real into PCM to apply something to them, no? Then, they are no more DSD files and you try to believe me that Audirvana with its upsampling abilities throughout with your wonderful DAC chip path is restoring the DSD they were first…

NO, you just hear a big bit rated PCM track played as DSD128 through a software.

And to conclude to that thread, you don’t need to lower the buffer memory for better sound, again my 2 cents, who i’m i to trust me, do you own work people…

16

Again, for your edification…

I could just playback the 24/352.8kHz ‘child’ file with HRTF DSP applied… However, I prefer the sound of this PCM ‘child iteration’ being modulated back to 1-bit 5.6MHz PDM (DSD128) via r8Brain for output from my 1-bit DSD-centric DAC circuitry, which is also my preference for all PCM file playback, where HRTF DSP is applied before being modulated to 1-bit 5.6MHz PDM (DSD128).

Apparently you do not understand the signal differences between PDM (Pulse Density Modulation) and PCM (Pulse Code Modulation) and the ramifications of how these ‘digital’ signals are converted to analog signals at output from a given DAC…

AND… again for your edification…
I play DSD 2.8MHz (DSD64) and DSD 5.6MHZ (DSD128) BINAURAL recordings natively (no plug-in DSP can be applied even if I wanted to apply some processing)… The only processing is in the DoP 1.1 packetizing for transmission via USB.

I still don’t get your point…

:notes: :eye: :headphones: :eye: :notes:

17

@RunHomeSlow
I want to see if I understood correctly.
This BUFFERING is used to load the music file and then stops being used until the next song, is this correct?

I have 16GB and I’m leaving 6GB and I use almost all types of files.
Today I use a Mac mini M2 with 16GB of RAM and Audirvana Studio

When I used a 2014 Mac mini with I5 and 4GB of RAM and Audirvana v3.5.5, if I left what the program adjusted automatically, on DSD256 there were glitches.
I left BUFFERING at 1.3GB and it worked normally.

18

In the past Audirvana was loading into memory 2 songs, play first load the other, might be still the same.

I have a mini 2018, with 20gb… any file or bitdepth plays easily even dsd128, my dac maximum with 17gb out of my 20gb for the buffer, it loads in few seconds, then judt play with zero ressources taken to the mac.

BUT if i upsampled 16/44 to DSD128, it still buff without any bug, but more slowly for sure you see the line advancing while soft is also playing the song with computing upsampling at the same time, 3 steps while playing against just playing the song from memory at regular bitdepth. And Audirvana is taking much more computer ressources with upsampling (for me, more then hqplayer that can stay at 100% compare to easily 250 to 300% cou usage for Audirvana till it rest after buffer).

But if you found lower your buffer was better, good, you tested it :grinning:

I use Origin, local files.
Imagine now you use Studio and listening music while streaming, and tell Studio to upsampling to the max of you dac… yes you could be in troubles :grinning:

1 Like
19

I find that when simply choosing a 16/44.1kHz file to playback or other bit-depth/sample-rate or DSD file, the computer activity jumps from 2-3% to sometimes over 50% until all of the album files are loaded into the playback queue and then settles back to a cruising speed… This CPU activity is a System level operation that is symbiotic, in order to load the playback pre-load memory ‘buffer’ and the CPU activity demands System memory… But this is not the only operation the CPU is handling in the Audirvana playback scenario as it pertains to available System RAM.

All of the processes like sample-rate conversion, etc, and transport protocol packetizing require System level memory, independent of the playback pre-load memory.

When the available System RAM is being diminished by a gross allocation of playback pre-load memory, the System is forced into disc-swapping ‘Virtual Memory’ and the number of System level interrupts increases, and latencies are created (wait times)… These interrupts have influence on the performance of Audirvana and will precipitate a variety of performance anomalies… some grossly obvious (depending on available System RAM) and some that subtly influence the perceived sound-quality of the subjective playback of the file.

There is no reason to allocate more pre-load memory for playback than is realistic given one’s configuration of Audirvana… There is a balance that must be found, so to get the best performance of Audirvana, given one’s System configuration…

From experience, I make the recommendation of staying at around 4-6GB (1-2 GB if the System RAM is 8GB) no matter how much System RAM is available… The availability of free System RAM will only better serve the synergy of OS and hardware and the performance of Audirvana…

:notes: :eye: :headphones: :eye: :notes:

20

@Agoldnear

The thing that you don’t want to look at or see yourself is your activity monitor. BUFFERING in Audirvana a CD that makes Audirvana to play from memory, TOOK around 5 seconds!!! maybe 10 if upsampling… then it is memory playing, not still using that memory buffer GB, CPU is sleeping, no ram is use by Audirvana, nothing works on the computer…. it is 95% idle most of the time. Just opening the activity monitor window takes more ressources on all the computer, than playing a CD….

So put your buffer at 4gb… look at the line loading while playing, i’ll let mine like at 17gb while i just listen to music with no loading while playing with a computer that it is sleeping… no cpu activity even with 1700 threads of thing running in the background… but Audirvana only is open :slight_smile:

You want to see a little a 2 minutes movie of an opening of Audirvana song, selecting a 24/88 file that i upsampling to 24/352 with 17gb of buffer ram… takes 5 seconds, then computer will mostly sleep till the end of the CD…except maybe at each start of a new song, 20% of activity for 2 more seconds.

So, WHY do you still want to see your buffering lines takes 1 minute or more while your song is playing?? that impact your computer ressources while playing and reducing the Quality or the hearing of the listener… no, i can’t talk like you :slight_smile:

Shit, i only have an old mini 2018 with no gizmo in the HDMI slot, no screen, that movie was filmed through an iMac 27’’ while screen sharing with the mini and playing that song upsampling to 24/352… i live dangerously.

Do the same thing… take a song at 24/88 or 24/96 upsample it to 24/352 or 384 not important… with your 4gb buffer and see what happen with your buffering line advancing when the song start… if it takes more time then mine, you impact the sound quality longer then me on your computer. The end.