I’m using a 2019 macbook pro 32GB ram with Sonoma dedicated to streaming. I have tried recently Audirvana Studio with it and for the first time the difference in SQ compared to ROON was very significant.
I have two questions:
What is the correct setting for the ram allocation? If I use it on Max., the sound quality is not as good as if I used it half way. But I would like to understand what’s the principle here…
Secondly, there seems to be an issue with the way audirvana connects with the USB converter. I have tried several converters and the issue is the same: I have to re-start the computer a few times, disconnecting the USB converter and refreshing the network option on audirvana to get right sound quality. Otherwise, the sound is not as good, as if there’s a bad synchrony between the software and the converters.
I am no expert on Macbook but allocating half of the RAM is sufficient on a 8Gb device.
So I suppose that in your case somewhere around 8Gb should be more than enough.
For your questions on the USB converter maybe some other users can be of help.
It is always good to copy and paste your debug info here. That gives the users more insight.
So folks can provide as much insight as possible, please paste your debug information report here…
It is found here: Settings → My Account → Help → Debug info
udirvana Studio 2.9.5 (20906)
macOS 14.7.2 x86_64 with 32GB physical RAM
Connected account of : hg
NETWORK
Status: available
SIGNAL PROCESSING:
Polarity Inversion:
Globally: OFF
Per track: OFF
Effects plugins NOT ACTIVE
UPSAMPLING:
r8brain not in use
r8brain filter parameters
Bandwidth = 99.5%
Stop band attenuation 218dB
Phase linear
AUDIO VOLUME:
Max allowed volume: 100
Replay Gain: None
SW volume control: OFF
LIBRARY SETTINGS:
Sync list: 0 folders
iTunes/Music library synchronization: not synchronized
Sort and display order: en
Library database path: /Users/marcelo/Library/Application Support/Audirvana/AudirvanaDatabase.sqlite
Local audio files fingerprinting
Tracks with no MBID: 11
Remote Control server:
Listening on on port 49181
ACTIVE STREAMING SERVICES
Qobuz: Connected as Qobuz Society
APPEARANCE SETTINGS:
UI theme: dark
Font size: regular
Language: System language
Show album covers in tracks list: no
Source list sorted:
My Music
Radios
Podcasts
Streaming
Local
Startup view: My Music: Albums
Show local extended in source list: no
Use media keys: no
Use media keys for volume control: yes
Use Apple Remote: no
Number of paired remotes: 0
Remote pairing code required: no
Screen saver disabled: no
=================== AUDIO DEVICE ========================
Active method: Local
Max. memory for audio buffers: 21822MB
Local Audio Engine: CoreAudio
Exclusive access: ON
Integer mode: ON
Use max I/O buffer size: ON
Actual I/O buffer frame size: 512
Preferred device:
NuPrime Audio Hi mDAC
Model UID:Hi mDAC:2FC6:F015
UID:AppleUSBAudioEngine:NuPrime Audio:Hi mDAC:CTUA191107:1
Active Sample Rate: 44.1kHz
Hog Mode is off
Bridge settings:
Sample rate limitation: 44.1kHz
Sample rate switching latency: 5s
Limit bitdepth to 24bit: ON
Mute during sample rate change: OFF
Selected device:Hi mDAC
Manufacturer: NuPrime Audio
Model name: NuPrime Audio Hi mDAC
Model UID: Hi mDAC:2FC6:F015
UID: AppleUSBAudioEngine:NuPrime Audio:Hi mDAC:CTUA191107:1
USB Vendor ID: 0x2fc6
USB Product ID: 0xf015
ID 0x57
8 available sample rates up to 384000Hz
44100
48000
88200
96000
176400
192000
352800
384000
Audio buffer frame size : 14 to 4096 frames
Current I/O buffer frame size : 512
Volume Control
Physical: Yes
Virtual: Yes
Max volume alert: Disabled
MQA capability
Auto-detect MQA devices: No
Not automatically detected, user set to not MQA
DSD capability
Unhandled (PCM conversion) with boost gain of 6dB
Device audio channels
Preferred stereo channels L:1 R:2
Channel bitmap: Ox3, layout:
Channel 0 mapped to 0
Channel 1 mapped to 1
Audio channels in use
Number of channels: 2
Use as stereo device only: No
Simple stereo device: Yes
1 output streams:
Number of active channels: 2, in 1 stream(s)
Channel #0 :Stream 0 channel 0
Channel #1 :Stream 0 channel 1
Stream ID 0x58 2 channels starting at 1
32 virtual formats:
2 ch Mixable linear PCM Interleaved 32 little endian Signed Float 384kHz
2 ch Mixable linear PCM Interleaved 32 little endian Signed Float 352.8kHz
2 ch Mixable linear PCM Interleaved 32 little endian Signed Float 192kHz
2 ch Mixable linear PCM Interleaved 32 little endian Signed Float 176.4kHz
2 ch Mixable linear PCM Interleaved 32 little endian Signed Float 96kHz
2 ch Mixable linear PCM Interleaved 32 little endian Signed Float 88.2kHz
2 ch Mixable linear PCM Interleaved 32 little endian Signed Float 48kHz
2 ch Mixable linear PCM Interleaved 32 little endian Signed Float 44.1kHz
2 ch Non-mixable linear PCM Interleaved 32 little endian Signed Integer 384kHz
2 ch Non-mixable linear PCM Interleaved 32 little endian Signed Integer 352.8kHz
2 ch Non-mixable linear PCM Interleaved 32 little endian Signed Integer 192kHz
2 ch Non-mixable linear PCM Interleaved 32 little endian Signed Integer 176.4kHz
2 ch Non-mixable linear PCM Interleaved 32 little endian Signed Integer 96kHz
2 ch Non-mixable linear PCM Interleaved 32 little endian Signed Integer 88.2kHz
2 ch Non-mixable linear PCM Interleaved 32 little endian Signed Integer 48kHz
2 ch Non-mixable linear PCM Interleaved 32 little endian Signed Integer 44.1kHz
2 ch Non-mixable linear PCM Interleaved 24 little endian Signed Integer 384kHz
2 ch Non-mixable linear PCM Interleaved 24 little endian Signed Integer 352.8kHz
2 ch Non-mixable linear PCM Interleaved 24 little endian Signed Integer 192kHz
2 ch Non-mixable linear PCM Interleaved 24 little endian Signed Integer 176.4kHz
2 ch Non-mixable linear PCM Interleaved 24 little endian Signed Integer 96kHz
2 ch Non-mixable linear PCM Interleaved 24 little endian Signed Integer 88.2kHz
2 ch Non-mixable linear PCM Interleaved 24 little endian Signed Integer 48kHz
2 ch Non-mixable linear PCM Interleaved 24 little endian Signed Integer 44.1kHz
2 ch Non-mixable linear PCM Interleaved 16 little endian Signed Integer 384kHz
2 ch Non-mixable linear PCM Interleaved 16 little endian Signed Integer 352.8kHz
2 ch Non-mixable linear PCM Interleaved 16 little endian Signed Integer 192kHz
2 ch Non-mixable linear PCM Interleaved 16 little endian Signed Integer 176.4kHz
2 ch Non-mixable linear PCM Interleaved 16 little endian Signed Integer 96kHz
2 ch Non-mixable linear PCM Interleaved 16 little endian Signed Integer 88.2kHz
2 ch Non-mixable linear PCM Interleaved 16 little endian Signed Integer 48kHz
2 ch Non-mixable linear PCM Interleaved 16 little endian Signed Integer 44.1kHz
48 physical formats
2 ch Mixable linear PCM Interleaved 32 little endian Signed Integer 384kHz
2 ch Mixable linear PCM Interleaved 32 little endian Signed Integer 352.8kHz
2 ch Mixable linear PCM Interleaved 32 little endian Signed Integer 192kHz
2 ch Mixable linear PCM Interleaved 32 little endian Signed Integer 176.4kHz
2 ch Mixable linear PCM Interleaved 32 little endian Signed Integer 96kHz
2 ch Mixable linear PCM Interleaved 32 little endian Signed Integer 88.2kHz
2 ch Mixable linear PCM Interleaved 32 little endian Signed Integer 48kHz
2 ch Mixable linear PCM Interleaved 32 little endian Signed Integer 44.1kHz
2 ch Mixable linear PCM Interleaved 24 little endian Signed Integer 384kHz
2 ch Mixable linear PCM Interleaved 24 little endian Signed Integer 352.8kHz
2 ch Mixable linear PCM Interleaved 24 little endian Signed Integer 192kHz
2 ch Mixable linear PCM Interleaved 24 little endian Signed Integer 176.4kHz
2 ch Mixable linear PCM Interleaved 24 little endian Signed Integer 96kHz
2 ch Mixable linear PCM Interleaved 24 little endian Signed Integer 88.2kHz
2 ch Mixable linear PCM Interleaved 24 little endian Signed Integer 48kHz
2 ch Mixable linear PCM Interleaved 24 little endian Signed Integer 44.1kHz
2 ch Mixable linear PCM Interleaved 16 little endian Signed Integer 384kHz
2 ch Mixable linear PCM Interleaved 16 little endian Signed Integer 352.8kHz
2 ch Mixable linear PCM Interleaved 16 little endian Signed Integer 192kHz
2 ch Mixable linear PCM Interleaved 16 little endian Signed Integer 176.4kHz
2 ch Mixable linear PCM Interleaved 16 little endian Signed Integer 96kHz
2 ch Mixable linear PCM Interleaved 16 little endian Signed Integer 88.2kHz
2 ch Mixable linear PCM Interleaved 16 little endian Signed Integer 48kHz
2 ch Mixable linear PCM Interleaved 16 little endian Signed Integer 44.1kHz
2 ch Non-mixable linear PCM Interleaved 32 little endian Signed Integer 384kHz
2 ch Non-mixable linear PCM Interleaved 32 little endian Signed Integer 352.8kHz
2 ch Non-mixable linear PCM Interleaved 32 little endian Signed Integer 192kHz
2 ch Non-mixable linear PCM Interleaved 32 little endian Signed Integer 176.4kHz
2 ch Non-mixable linear PCM Interleaved 32 little endian Signed Integer 96kHz
2 ch Non-mixable linear PCM Interleaved 32 little endian Signed Integer 88.2kHz
2 ch Non-mixable linear PCM Interleaved 32 little endian Signed Integer 48kHz
2 ch Non-mixable linear PCM Interleaved 32 little endian Signed Integer 44.1kHz
2 ch Non-mixable linear PCM Interleaved 24 little endian Signed Integer 384kHz
2 ch Non-mixable linear PCM Interleaved 24 little endian Signed Integer 352.8kHz
2 ch Non-mixable linear PCM Interleaved 24 little endian Signed Integer 192kHz
2 ch Non-mixable linear PCM Interleaved 24 little endian Signed Integer 176.4kHz
2 ch Non-mixable linear PCM Interleaved 24 little endian Signed Integer 96kHz
2 ch Non-mixable linear PCM Interleaved 24 little endian Signed Integer 88.2kHz
2 ch Non-mixable linear PCM Interleaved 24 little endian Signed Integer 48kHz
2 ch Non-mixable linear PCM Interleaved 24 little endian Signed Integer 44.1kHz
2 ch Non-mixable linear PCM Interleaved 16 little endian Signed Integer 384kHz
2 ch Non-mixable linear PCM Interleaved 16 little endian Signed Integer 352.8kHz
2 ch Non-mixable linear PCM Interleaved 16 little endian Signed Integer 192kHz
2 ch Non-mixable linear PCM Interleaved 16 little endian Signed Integer 176.4kHz
2 ch Non-mixable linear PCM Interleaved 16 little endian Signed Integer 96kHz
2 ch Non-mixable linear PCM Interleaved 16 little endian Signed Integer 88.2kHz
2 ch Non-mixable linear PCM Interleaved 16 little endian Signed Integer 48kHz
2 ch Non-mixable linear PCM Interleaved 16 little endian Signed Integer 44.1kHz
Local devices found : 2
Device #0: ID 0x57 Hi mDAC
Manufacturer: NuPrime Audio
Model UID: Hi mDAC:2FC6:F015
UID: AppleUSBAudioEngine:NuPrime Audio:Hi mDAC:CTUA191107:1
USB Vendor ID: 0x2fc6
USB Product ID: 0xf015
Model name: NuPrime Audio Hi mDAC
Device #1: ID 0x65 MacBook Pro Speakers
Manufacturer: Apple Inc.
Model UID: Speaker
UID: BuiltInSpeakerDevice
Model name: Apple Inc. MacBook Pro Speakers
UPnP
UPnP network interface
Available Network interfaces:
Apple USB Ethernet Adapter
UPnP devices found : 0
Chromecast
Chromecast devices found : 0
I suggest lowering your playback pre-load memory allocation to 4-6GB… You are currently allocating 22GB of your 32GB of System RAM for playback pre-load memory… The synergy of available RAM for System CPU operations in concert with Audirvāna operations, is intrinsic to sound-quality as you can hear… This is a balance… these operations are tied to memory throughput across the memory bus and the speed of the CPU, memory speed, and in your case, latencies due to broadband throughput intrinsic to your streaming service and the DSP required to unpack the streaming data as well as router performance.
When possible employ USB 3 transmission protocol on the shortest possible decent-quality cable to your DAC. This can be done by employing something like the iFi-Audio iPurifier3. USB 3 protocol removes the signaling interrupts from the data lines.
I suggest in concert with a lowered pre-load memory allocation, to experiment with ‘Power of Two’ up-sampling of PCM files, this will up-sample all PCM to the maximum sample-rates of your DAC (352.8kHz and 384kHz).
Note: lower your sample-rate switching latency to either zero or maximum one-second (1s), five-seconds (5s) is an abnormal setting that might be used in some really crazy playback scenario…
Thanks for the reply.
If I understand you, I should make available the “extra” ram to the non-audirvana processes, and just allocate 4-8gb to audirvana, correct?
If that’s the reasoning I was doing the opposite. It’s something that should be clearer on the help menu…
Regarding the 5s latency, I’ve been using it believing it would allow the system to have a better sync with all the clocks involved when changing the resolution. Switching resolution has always been a cause of small sound differences and inconsistencies for me.
May I ask you to consider also my second question? Sometimes when starting the computer and opening audirvana the SQ is a bit adulterated. I have to turn it off, unplug and plug the usb converter , and turn the mbp on again, then the SQ is correct. I suspect it’s a minor clock or network glitch. I don’t expect you to have a solution as there are too many variables, but I wonder if this is a common connectivity issue. Incidentally, I have tried four different usb converters and it’s all the same. The OS has been recently installed, so as the audirvana app. It’s an odd issue that I don’t have with ROON and had also before on my previous computer with Audirvana.
also, the ram allocation is set to almost all the way up when we install audirvana. That’s why I think I was also under the impression that the larger portion of the memory should be dedicated to the app. Any reason for that?
When I look at the activity monitor, the total ram used by the computer with audirvana playing a regular 44.1k stream is about 4gb…
Simply put… Yes
Sample-rate switching latency is going to be relative to available RAM for all System and Audirvāna operations… Start with zero latency and move up incrementally in .5ms delay… Based on what I see, you are not playing DSD files, which presents more potential for noise when switching sample-rates by nature of the 1-bit PDM encoding format, so a very short delay will serve you best… The fundamental RAM required for both macOS and Audirvāna is a constant (Sonoma and Sequoia are large)… the pre-load memory allocation is a controllable variable and is interdependent with System/CPU topology performance/Memory speed and Memory bus speed operations.
Turn on ‘Hog Mode’ this disables all System sounds/alerts from interrupting the Audirvāna audio-engine… Check your System Exclusive settings to give Audirvāna maximum control and disable all other functions during playback.
Not sure what this means, as it is a very subjective assertion…
Your USB output from the computer should not be of any concern (unless on a long interconnect… It should be very short)… What is imperative, is how you are managing your power/ground relationships with all of the components connected to your computer… Your playback components which include your router, will induce noise on the signal-path from grounding/earthing differentials, which precipitate ‘jitter’ on the digital-audio signal… Best practice is to keep all of your components on a common power/ground/earth circuit from a single socket from the wall outlet, unless you know for sure that both sockets of the outlet are on the same circuit and make sure there are no non-audio related devices on that circuit (lights, fans, motors, etc.) … My system is completely isolated and resides on its own ‘power/ground island’…
It is possible to isolate the DAC from the computer USB bus via the iFi-Audio iPurifier 3, which as I previously described, will enable the use of USB 3.0 protocol cable for the primary interconnect from the computer. I also isolate my computer Ethernet feed from my router using an iFi-Audio LAN iSilencer.
The object is to mitigate the accumulation of noise related jitter, “Jitter Stew”… The only true digital-audio bits are the ones and zeros encoded on the storage device before being lifted from it, for transmission of the code signals as electrical voltages to and through your computer topology and amalgamation of playback components… All of these factors play into your perceived quality of sound…
Just for reference, my fundamental playback system…
Audirvāna 2.5.9 is running on macOS 15.2, running on a M2 Max Mac Studio with 64GB RAM, 8GB of playback pre-load memory allocated…
I am headphone-centric and modulate (up-sample) all PCM files to DSD128 (5.6MHz) after HRTF processing in the plug-in module. My USB 3.0 interface resides in a Thunderbolt 3 PCIe expansion chassis which hosts the USB interface and an extensive PCIe-power filtering system… USB 3 cable to iGalvanic 3 which is connected to a 4" USB 2.0 ‘cable’ pc-board (discontinued UpTone Audio USPCB) that feeds my DAC… today the iPurifier 3 will replace this short transition from USB 3.0 to USB 2.0.
Thanks again. Where is the setting for hog mode? Never seen it…
By adulterated sq I meant only that it was different from when it is at its best.
Other than ground noise and bit perfection there’s also small clock mismatch issues that cause noise, no?
Clock synchronization is related to noise on the data signal which carries the meta-data about timing etc, Gross noise or poor interconnect quality can affect signal integrity and clock synchronization in the component architecture topologies, and worse-case you would most likely get digital ‘hash’… The phase relationships of the L and R signals as encoded in the master recording can be skewed by noise related jitter… If you are streaming digital-audio files from a streaming service, it is entirely possible the transmission is jittery and will induce latency that ripples through the entire playback chain.
Analysis of jitter component is found in the Texas Instruments technical article linked below:
“Skew definition and jitter analysis”
By Steve Corrigan
System Specialist, Data Transmission
I have the system optimization on at extreme, but the spotlight and Time Machine off options are unchecked as I have already taken care of turning them off on Terminal. Didn’t want to add redundancy.
I have noticed in the past that when the spotlight off option was activated, the computer would enter in a very active mds mode after quitting audirvana, which was unbearable
Why is it called hog mode if there’s no reference to that term? Confusing…
I have 32 gb on my mbp ram, which I thought was an overkill for audirvana. But I see you have 64… Does that make a difference if we compare with a 16 gb ram computer? I mean, audirvana needs about 8 gb and all the other processes combined would never , according to activity monitor, get to more than 8 gb ( if os is optimized and only audirvana is used). I just wonder if by accident I bought the right machine, as I wasn’t really worrying about more than 16 gb…
My previous platform was a 2016 MacBook Pro, i7 2.7GHz. with 16GB RAM (4gb being allocated for pre-load memory) before it died… And under the same playback scenario, Audirvāna provided a high-performance playback experience… My current platform, is looking into the future of digital-audio file playback… The premise for all of my previous Apple computer systems has been developed over 40 years of experience with digital-audio recording and playback system architectures, and being a Apple platform user since the Apple IIe, among other contextual technical engineering acumen.
That premise is to provide the OS and CPU/hardware platform topology with as much RAM that I can afford, so to create an environment where the system is capable of lowered latency and signal interrupts from request wait-times and unencumbered signal throughput, that subsystem operations depend upon…
The current system playback is flawless, especially when jumping around from track to track and from modulated PCM to DSD and pure binaural DSD playback.
I have sensed a lower noise-floor with the Mac Studio that I attribute to its design architecture and the newer macOS in comparison to the MacBook Pro platform employed in the same playback system/component architecture.
I see that the Hi mDAC utilizes a custom USB interface, which may ultimately play into your assessment of sound-quality, as this has been the folly of companies like Schiit Audio where early-on their USB implementation was sub-par… Since then, the designers have made great improvements in their USB interface implementation… Also the Hi mDAC is dependent on being powered by the USB bus controller, which will have some implications on noise related jitter.
I see, thanks for the feedback. I suppose that when you’re comparing the mbp to the desktop you have all settings exactly the same, as far as possible. They are very different machines… in my experience a proper configuration is far more important than different hardware, ram size on the side.
Did you see my reply regarding the hog mode thing ? I’ve tried checking the disable spotlight/ Time Machine / screen saver and now hog mode is on according to debug mode. But if I had those things already disabled, what’s the point of checking those boxes? Before installing audirvana the os was totally optimized for minimal extra processes.
Configuration is intrinsic to and synergistic with platform performance…
Question for the Audirvāna team 
All things being equal, would you say a macbook air (fanless…) m3 with 16gb with sound better than an intel i9 mbp with 64gb? Just for pcm, mostly streaming from qobuz
Macbook air m3 can also have 24gb , so either 16 or 24 compared to the intel with 64gb…
A very subjective construct…
You are asking me to presume something that is dripping with subjectivity in the assessment of the final audition, in a given playback system architecture and amalgamation of components… It is impossible for me to do, or anybody not able to experience this juxtaposition in your particular playback scenario, among other factors and biases that play into the cognitive biases that formulate a contextual and subjective certainty about what “sounds good”. 
Simply put… I suggest to everybody putting together a computer hosted digital-audio playback system, that they purchase the most powerful platform architecture with the largest available RAM that one can afford… However I will state here that an 8GB system is contextually useable, but not optimum… Maybe an ‘M’ series macOS platform with 16GB RAM where the system library is not stored on the System drive will perform very well… This is something that must be corroborated by users that have systems built around these platforms and available System RAM… However, the modus operandi of each playback scenario will not be exactly the same… You will need to get insights from users that have system architectures and playback scenarios like yours.
From this point it takes more than this simple conversation to build a compendium of experience and technical expertise, to get the most from the macOS platform architecture hosting the Audirvāna audio-engine… Audirvāna is a high-performance audio-engine and sometimes requires deeper insights into the functionality of the application so to support the integrity of the bit-perfect digital-audio signal presented to the platform output bus controller protocols, and beyond those controllers where Audirvāna has no control of the fundamental integrity of encoded digital-audio signal being transmitted to the DAC. Ultimately it is the synergy of all playback system components that we hear in the final audition… We do not hear the encoded digital-audio signals… We listen to the results of all of the system/component architecture(s) influences that the encoded digital-audio data signal passes through on its way to the analog output circuitry of the DAC.
I am not an Apple user, so I cannot possibly tell which devices would suit your needs best.
But experiences from users all over the world and research show that attention for every detail is necessary. Every part of the chain has or can possibly effect sound quality.
Which of the chosen solutions serve the personal taste of a user is dependent on personal taste.
Trial and error are part of the entire journey. I like the articles from Alpha Audio very much for insights.
@Agoldnear has brought a lot to the table already, so I think you have some directions to possible destinations on your journey
Last (best?) advice that I can give: remember to enjoy the music…
Thank you, but it has nothing to do with personal taste or subjectivity. To be clear I prefaced my question with “all being equal” namely, if the only differences were the computer processor and ram size. I’m just trying to assess which, either processor or ram size, would be more relevant. Simple question, perhaps with not a simple answer.
I understand that it would be speculative if one never compared the two situations, but I thought of asking it based on his extended experience with mac computers. No big deal.