Audiophiles lack of respect.

[thesurfingalien]

Hi Alex,

 

I assure you I do not have an axe to grind. Although being skeptical to the extreme regarding this topic, I think my efforts to recreate similar results show that I at least keep an open mind.

 

But what you do not have you can not share.

 

Regards,

Peter

 

[thesurfingalien]

Hi Mel,

 

I think the whole idea behind buffering an entire song is to prevent disk-access and conversion-activity during playback. These actions could generate audible noise when done during playback.

 

I use Foobar2000 that does not do buffering, but I can not measure nor hear any negative influence of either disk-activity or FLAC to WAV conversion during playback.

 

 

You are correct in what you write, but if I understand correctly , buffering an entire song is not to prevent hick-ups in playback.

 

I could be wrong as I never looked into that too much.

 

Regards,

Peter

 

[elcorso]

Peter & Mel,

 

Regarding "...And it would also help if you are using an application like A+ that pre-caches a song in memory before playing...

 

I have two DACs one is USB and the FireWire, bot playing under A+, Integer Mode and from memory playing exactly the same music track.

 

The USB CPU one load is ?5%

 

The FireWire one load is ?25%

 

Memory usage is the same under both DACs.

 

I guess under the FireWire we can get more noise, thanks to the SPSU in the PC increased power consume (from ripple), even if "the the transfer of the data from the application to the DAC is extremely fast".

 

That' why it is important (to me) to know if it could be fragmentation in the memory buffer. Because microseconds delay in transmission times could affect some way music time and phase, where I found (or think I found) differences in the SQ.

 

This is, by no way, a catch question. But to try to understand what I listen what I listen, and it is not from my imagination.

 

Regards,

 

Roch

 

[thesurfingalien]

Hi Roch,

 

Interesting... I can not think of a reason why the CPU-usage is so much higher when using FW...

 

Can you check if there are additional processes running or CPU-usage of Audirvana (or some other process) is higher when using FW?

 

It could not be something like software-upsampling to 24/192 to accommodate the FW DAC right? (Highly unlikely, and that should not require 20% CPU-time)

 

I am not a Mac-user, so I could be missing something here, but I don't believe fragmentation in the memory-buffer of the player is the case or even matters if it did.

 

Also, your player sends the data to the memory-buffer of the USB and FW controller. Usually, a user will receive error-messages when the data is not arriving in time (causing buffer under-runs), and the USB of FW controller is not capable of sending data to the DAC in time.

 

So, for now I can not see a possibility for a delay in transmission.

 

Regards,

Peter

 

[Mel Hi ReZ]

I agree. There should not be any delay in transmission. The CPU can perform the memory fetch operation much faster than what the DAC needs, regardless of where in the memory bank it resides.

 

I did a very simple test with my configuration, which uses USB to Vlink192 to DacMini DAC, and the CPU utilization in my MacMini is very light. And I have my music files coded in Apple lossless, therefore the CPU has to perform the decompression before playing through the Audirvana+ iTunes configuration.

 

1. Very light load after startup:

CPU: %User-0.4, %System-1.25, %Idle-98.5

Memory: Free-2.82GB, Wired-747MB, Active-395MB, Inactive: 1.15GB

 

2. Audirvana+iTunes playing 16/44.1 song:

CPU: %User-1, %System-3, %Idle-96

Memory: Free-2GB, Wired-925MB, Active-927MB, Inactive: 188MB

 

3. Audirvana+iTunes playing 24/192 song:

CPU: %User-2, %System-3, %Idle-95

Memory: Free-704MB, Wired-800MB, Active-1.39MB, Inactive: 1.14GB

 

There were no page faults (as it should). CPU utilization goes slightly higher for just a few seconds at the transition point between songs. In order to play a new song, as expected as the new song is loaded into the application/memory, and other conversion and miscellaneous tasks are performed.

 

I agree with Peter that there must be something else going on in Roch's system and configuration to explain why the CPU utilization is so much higher using FW.

 

[elcorso]

No other process are running under both cases, but the required by the system and the app.

 

No upsampling, conversion, etc. in both cases.

 

In both cases music already loaded in the memory buffer.

 

I'm not talking extrictly about data 'not arriving on time', since it causes drop-out. But in one case, under fragmentation on memory (I don't know if possible) could cause a lesser speed than with no fragmentation at all, then some differences in the SQ (if they are).

 

Please understand that I do believe in 'mother nature laws' the same as Bill believes. If this laws wouldn't exist, all the time, the whole universe will crash. But we are all the time trying to understand and measure this laws.

 

My big doubt is, if we have now all the necessary tools under the scientific method with the right and ultra precise instruments to measure it.

 

For example, I was following with interest the experiments in the UE regarding they try to reproduce the Big Bang Theory. The first one achieved that neutrines are faster than light, but the second one (by yesterday publication) said NOT. Then Albert Einstein was right, there isn't nothing faster than light.

But, who knows, next experiment could (again) proof the contrary, even if I believe in Einstein extraordinary genius.

 

Thanks and regards to answer me,

 

Roch

 

[thesurfingalien]

Hi Roch,

 

We are here to help eachother! Or, at least try :-)

 

I am not sure under what circumstances you can hear differences. Can you tell me a bit more about it?

 

 

For CPU-load, is there a way for you to see the individual CPU-usage on all processes? If there are no differences in the list of processes running, than at least one should take more CPU-time to account for the difference...

 

Anyway, maybe a MAC-user can chip in here?

 

 

Regards,

Peter

 

 

 

[orgel]

Peter wrote:

 

"Anyway, maybe a MAC-user can chip in here?"

 

I can't say that I experience this with my FireWire DAC (the Mytek DSD), but I've heard second-hand accounts that some of the proprietary FireWire drivers (e.g., DICE) can consume a bunch of CPU cycles. Not sure what to make of this, since my CPU load with my mid-2011 Mac mini with the Mytek/FW/DICE driver is typically 5%-6% with music playing, including on-the-fly upsampling in Pure Music.

 

If, though, Roch is using a third-party FireWire driver, I guess that's a possibility to account for the increased load. There has to be some process eating up the cycles, and it should be reasonably easy to figure out which process(es) it is by using Activity Monitor. If it were me, I'd make sure all obvious extraneous processes were shut down, fire up Activity Monitor, click on the "CPU" tab, and click on the "% CPU" column head to sort by that property.

 

--David