|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
209.130.140.130
In Reply to: RE: Not Really posted by Dynobot on March 23, 2012 at 07:54:35
The analogy is not really appropriate because you are assuming that somehow getting the data to the DAC faster and through a less complicated path will improve things. This isn't true; your comparison is the equivalent to saying a concert will sound better if your delivery of the sheet music to the musicians doesn't require moving you through lots of traffic, and it arrives early. That is nice, but it isn't going to change the sound that the musicians make based on the notes they see on the paper...
Follow Ups:
Not faster, maybe you get there at the same time, but you get there with little to no effort. With all that traffic to contend with you might have to change lanes, speed up and maneuver more...but will a clear shot you can just take your time and cruise without worry.
The casual observer might say, so what, he got there at the same time so no difference. But your s/o will clearly notice a difference.
In this case, the listener is the s/o, and we all can hear a difference when our data gets to our Dacs with less effort.
Dynobots Audio
Music is the Bridge between Heaven and Earth - 音楽は天国と地球のかけ橋
=======
we all can hear a difference when our data gets to our Dacs with less effort.
=======
Can we indeed? Where can I read about a listening test supporting this statement? So far I have not been lucky to find any.
The data will get to your dac will using same path and operation regardless of other processes, provided they are stored in RAM for DMA well ahead. How does software influence the timing of the data being transferred to the DAC by DMA?
> > > Can we indeed? Where can I read about a listening test supporting this statement?
Every post or article you have read about getting less jitter out of a source is the result of just that....
The more you obstruct anything from getting from point-A to point-B the more jitter you induce on that thing.
Here is an Abstract of an article on OS/tasks Jitter
Abstract. This paper deals with real-time task jitter measurement under
RTLinux operating system. In the first part, it describes methods and tools developed to measure jitter in the RTLinux environment. In the second part, it is focused on discussion of results, obtained on PC hardware, and their interpretation.
Dynobots Audio
Music is the Bridge between Heaven and Earth - 音楽は天国と地球のかけ橋
==========
The more you obstruct anything from getting from point-A to point-B the more jitter you induce on that thing.
===========
No doubt about that. Just that it is not how computer audio works. The soundcard/USB controller does not care when that thing gets into point B, as long as it gets there on time. In computer audio the point B is almost always a memory location where the data waits (this is called the playback latency), until the soundcard/USB controller reads it via DMA, independently of the CPU (i.e. tasks/scheduler). How does the jitter in process scheduling as analysed by that paper you have linked fit into the picture? I do not see it being involved in the second stage - delivery to the actual DAC. It may affect the spectrum of power supply noise but that is about all.
You are using linux, right? A typical player would be MPD. For PCI cards, in its default configuration the data waits in RAM for the DMA to read for almost half a second since the buffer size is 500ms and period size (IRQ period) 5ms. It means every 5ms MPD makes sure there is approx. 495ms of fresh audio samples ready in RAM for the DMA to read. USB uses lower values since there are hard-coded limits in the USB driver. Raise them (for adaptive mode) and you will get the same values. And I have not ever seen anyone here discussing the buffer_time and period_time parameters in mpd.conf, meaning the defaults are being widely used.
==========
It means every 5ms MPD makes sure there is approx. 495ms of fresh audio samples ready in RAM for the DMA to read.
==========And the process scheduling jitter means whether those 5ms are precisely 5.00ms or 5.1ms or even 7 ms. It could even be hundreds of ms without any impact on the DAC operation, if the buffer time is still significantly larger to give the OS enough time to re-fill the part of memory buffer with old data. The computer audio was intentionally designed this way not to give CPU the last word in data transfer. The engineers would have had to be completely dumb to allow that.
Edits: 03/25/12
"The computer audio was intentionally designed this way not to give CPU the last word in data transfer. The engineers would have had to be completely dumb to allow that."
Back in the 1960's when transistors were very expensive and data rates were very slow some engineers designed a multiline teletype interface for a minicomputer to work just like you suggest. This worked OK until a salesman tried to sell it to a PTT (Post, Telgraph and Telephone) authority down under. (AU or NZ, I don't recall which.) It turns out that they required all equipment to meet a tight jitter specification. (Tight because it required jitter in the microsecond region for telegraph lines running at 50 bits per second or slower.) As it turns out the software could not meet this number when it was handling a large number of lines and the product was too expensive if the computer was restricted to a small number of lines. So, you could say the engineers were dumb. But I knew them and they weren't completely dumb. They designed a new version that provided a bit buffer and solved the jitter problem.
One might wonder why a 50 bit per second telegraph interface needs jitter in the microsecond region. It has to do with telegraph distortion which starts at the transmitter, goes through the wires (large L, C and R) and ends up at the receiver which has to sort out the mess. Tight specifications reduced the chance a marginal telegraph line might garble the signal. (The lines would blow around in the wind and the dialectric would vary according to the weather, hence inconsistent distortion.)
Not much has really changed in over 40 years. Transistors cost much less, and time is now measured in units of picoseconds instead of microseconds. Bits weren't just bits back then and they aren't just bits today. (Time resolution in the analog domain is much smaller than time resolution in the digital domain according to the signal to noise ratio.)
BTW, telegraph distortion was very old hat in the 1960's.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
=========
They designed a new version that provided a bit buffer and solved the jitter problem.
=========
Exactly. Last few decades this buffer is called a circular DMA buffer. And through IRQs it offers more sophistication than a pure FIFO. It lets the writing process to define "breakpoints" (period boundaries). When the soundcard crosses this boundary, it sends an IRQ saying "I was here". The IRQ wakes up the until then sleeping writing process. The awoken writing process asks the driver/the soundcard about its current reading pointer position (which is different to the period boundary which caused the IRQ since some time has passed now), refills the already read portion of DMA buffer with new data and goes to sleep again to wait for another IRQ "I was here" call.
In MPD 44100kHz terms: wakeup every 5ms, the whole buffer 500ms, initial fill 500ms, after that 5ms of audio data processed and copied to RAM every 5ms (i.e. a few 100s of samples, nothing for modern CPU), the writing pointer approx. 495ms isch ahead of the reading pointer by the soundcard. If the IRQ gets processed with 100ms delay (quite a large delay in computer terms) - the MPD writing process will provide 105ms of data instead of 5ms, a few tens of kBytes of data.
Of course the writing process is usually a dedicated writing thread, independent of the GUI and control interface, reading over network, reading from storage etc.
==========
In MPD 44100kHz terms: wakeup every 5ms, the whole buffer 500ms, initial fill 500ms, after that 5ms of audio data processed and copied to RAM every 5ms (i.e. a few 100s of samples, nothing for modern CPU), the writing pointer approx. 495ms isch ahead of the reading pointer by the soundcard. If the IRQ gets processed with 100ms delay (quite a large delay in computer terms) - the MPD writing process will provide 105ms of data instead of 5ms, a few tens of kBytes of data.
==========
So in your estimation reducing the number of processes, changing priorities, and allocating the music process to a single core to its own does not or should not improve sound?
"The data will get to your dac will using same path and operation regardless of other processes, provided they are stored in RAM for DMA well ahead. How does software influence the timing of the data being transferred to the DAC by DMA?
Then why is it that so many people actually do [believe] they hear a difference in sound? I suppose even the Windows Fiderlizer should not effect sound either. I know you probably think your logic is solid and should therefore rule over reality, but its just not so...
Dynobots Audio
Music is the Bridge between Heaven and Earth - 音楽は天国と地球のかけ橋
=========
So in your estimation reducing the number of processes, changing priorities, and allocating the music process to a single core to its own does not or should not improve sound?
=========
You are mixing two aspects - noise (cpu load) and software timing (changing priorities, and allocating the music process to a single core to its own). Noise (supply lines, data lines) will affect the sound negatively. Software timing may affect resultant noise but who knows how? Why should changing rt priority change noise in a positive way, i.e. at best reduce it? It does not make any sense.
There is no direct link between the software timing and the DAC timing. The only linking element is the noise. Yes, I do not believe software timing will improve the sound. I have never experienced the effect and I did two DBT tests (hi carcass93, do not forget to jump in :-) ). One changing RT priority, another changing playback latency.
==========
Then why is it that so many people actually do [believe] they hear a difference in sound? I suppose even the Windows Fiderlizer should not effect sound either. I know you probably think your logic is solid and should therefore rule over reality, but its just not so...
==========
Show me results of a credible DBT result which confirms it is not only about believing. I have asked for them countless times, to no avail. Setting up a blind test in linux is simple since everything can be easily scripted. A trivial script of mine to start from is freely available at http://blog.ivitera.com/pavel/linux-audio/tool-for-ab-testing .
> > > > Show me results of a credible DBT result which confirms it is not only about believing.
Its just this simple.
If I say I hear a difference when I do "anything", then I hear a difference. I don't need a DBT or a White-Paper to explain the scientific proof. Try to focus in on reality, you nor anyone else can say anyone did or did not experience anything with their senses.
Sensory perception needs no proof because its a personal experience.
If I say I hear it, then I do [period].
Dynobots Audio
Music is the Bridge between Heaven and Earth - 音楽は天国と地球のかけ橋
========
Sensory perception needs no proof because its a personal experience.
If I say I hear it, then I do [period].
========
Well, I am not your psychoanalyst to discuss your impressions and beliefs. Should you want to delve into technical issues, I will be glad to participate.
You mean to say that everything you have experienced in life can be proven with a DBT??
A simple yes or no answer will do fine.
Dynobots Audio
Music is the Bridge between Heaven and Earth - 音楽は天国と地球のかけ橋
No reason to continue with this level of discussion...
Don't be a sore looser.At least find a better way to concede than to just run away.
Your answer would have been "NO everything that I have experienced in life can not be proven with a DBT"
Its ok just say it....I promiss you will feel better afterwards, of course I can't prove it but I believe it. ~wink~
Dynobots Audio
Music is the Bridge between Heaven and Earth - 音楽は天国と地球のかけ橋
Edits: 03/27/12 03/27/12
Your response is exactly the reason why I have no interest in further discussion on this level with you.
Unlike you, I do not care if I win or if you win. What I care about is I want to finally find out if people can REALLY distinguish between RT kernel, low latency, i.e. strict and precise playback timing vs. regular timing. I do not care about subjective claims, they have no value for me. Internet is full of useless claims with no evidence. I do not trust my ears just as much as I do not trust your ears. Show me you can distinguish the difference in an objective scenario where your brain cannot play games with you and I will work hard to try to find the cause.
But I have asked for objective results I am looking for many many times. The result - none. Even though it is so simple... provided you can really hear it.
"What I care about is I want to finally find out if people can REALLY distinguish between RT kernel, low latency, i.e. strict and precise playback timing vs. regular timing. I do not care about subjective claims, they have no value for me."
You are asking for absolute truth. You are asking for the unattainable. We are all subject to the limitations of time, place and person. "Reality" is a thought projection of our minds. We each see a limited projection of reality in our mind. Our projection varies from moment to moment. This is why there is no general answer to your question.
If you get down to specific questions concerning what a particular listener hears at a single point in time, you still won't get an answer. Even if you are the experimenter and the listener so there can be no one else deceiving you there will still be no real answer. What one hears depends on the internal state of one's mind as well as external impressions on one's senses. Intentions, knowledge and experience are at least as important as sensory impressions. You may know what you heard, no doubt about that, but you won't know for sure why you heard what you heard. You will not be able to resolve causation without making a large number of assumptions.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
===========
You are asking for absolute truth. You are asking for the unattainable. We are all subject to the limitations of time, place and person. "Reality" is a thought projection of our minds. We each see a limited projection of reality in our mind. Our projection varies from moment to moment. This is why there is no general answer to your question.
============
You know that we have had this discussion several times. I am not asking for absolute truth. I am asking those who say they do hear the difference clearly (and that DO has been emphasized by them - either "day difference", or "if I say, then I do") to confirm their firm belief with an objective test. No more, no less. I am not asking anyone who says that the diffeerence is subtle. But if you are convinced, you should be able to tell the difference even if you do not know whether there is any or not. If not, if the firm conviction depends on a number of other aspects, then the "firm" cannot be so "firm" anymore... Or not?
Nice to open your mind....my little Circle friend.
Dynobots Audio
Music is the Bridge between Heaven and Earth - 音楽は天国と地球のかけ橋
If they follow their suggestions there are two likely outcomes:
1. the effect disappears and/or they become confused.
2. they report their results and get flamed for conducting an invalid experiment.
So what's in it for them?
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
=======
If they follow their suggestions there are two likely outcomes:
1. the effect disappears and/or they become confused.
=========
How can an effect disappear when they were so certain it exists? Like you are sure you know stuff, yet you fail the exam since the knowledge disappeared? Well, a teacher would know the answer... Or you really know the stuff and do not fail the exam.
=======
2. they report their results and get flamed for conducting an invalid experiment.
======
But they will know for themselves if they cheated on the results or just others are giving them hard time for their honesty. But not doing any testing for myself and still being very vocal publicly... definitely not my cup of tee, sorry.
"How can an effect disappear when they were so certain it exists?"It would appear that you haven't much experience doing double blind tests that attempt to discriminate between marginally different sounds. After this has happened once or twice one can begin to lose one's trust in one's own senses and mind. There are a number of theories that try to explain why this happens, but I'll leave that to others with more expertise in the "why" of human perception. It does happen, however, a case of "what" being more important than "why".
These tests are definitely educational. But they are more educational about ourselves than about the audio components being tested.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
Edits: 03/27/12
Perhaps the only way for mankind to accept new realities is to first knock down old Dogmas.This new frontier of knowledge may not have black and white explanations, yet the reality persists.
Dynobots Audio
Music is the Bridge between Heaven and Earth - 音楽は天国と地球のかけ橋
Edits: 03/27/12
- the last refuge of an audiophile.
Bill
Dynobot was a participant in this thread I started on the "Isolation" ward. Other may not have seen it, as it didn't get wide participation.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
is an age old question.Religious Zealots would love to be able to prove their perception of a higher Being.
But then I don't think a DBT would be an appropriate measuring tool.
Dynobots Audio
Music is the Bridge between Heaven and Earth - 音楽は天国と地球のかけ橋
Edits: 03/27/12
I said I can distinguish a difference with the RT-Kernel. You did not believe me and asked for a DBT test to prove my perception.Fact is you can not use a DBT to prove anyones perceptions...because they are personal experiences.
You didn't want to face that truth so you ran.
Why ask a question that you are not prepared to accept the answer too, UNLESS it reinforces your own belief?
You will never find what you are looking for simply because you are not prepared to accept the truth.
The above sentence is a Dynobot NUGGET-o-Wisdom, remember it and apply it to your life, happiness will follow.
Dynobots Audio
Music is the Bridge between Heaven and Earth - 音楽は天国と地球のかけ橋
Edits: 03/27/12 03/27/12 03/27/12
Discussion with you makes no sense now. Good luck.
Yes, its possible that everything, including the alignment of the planets, affects the sound in your computer audio system...IF THE DAC IS DEPENDENT ON THE TIMING OF INCOMING DATA. Hence my reminder...if all this crap is making a difference in your system, you probably need a new DAC.
Then why did you buy your Dac???
Its main feature "Asynch" is suppose to address the issue of data timing between the source and the Dac.
Odd that you would buy a Dac and not believe in its merits.
Is this a habit for you, do you not believe cough syrup works but buy it for your cold. Kind of like being an Athiest and going to church every Sunday.
Dynobots Audio
Music is the Bridge between Heaven and Earth - 音楽は天国と地球のかけ橋
Hahaha
Need a new Dac.
Dynobots Audio
Music is the Bridge between Heaven and Earth - 音楽は天国と地球のかけ橋
Somehow I don't see the flip-flops in the registers worrying that the next clock pulse won't arrive in time. :-)
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
You never know....
Dynobots Audio
Music is the Bridge between Heaven and Earth - 音楽は天国と地球のかけ橋
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: