Windows 11 vs 10.

[MarcelNL]

looking back the discussion was about a difference in sonic fingerprint between W10 and W11, so around differences in OS not about USB data transmission, or any other data transmission since USB wasn IMO not mentioned anywhere. 

 

Basic truth in computer systems? eerrr who's defining what is truth and about what?  Restating that something cannot make a difference makes it a truth?

 

There are various bit perfect audio players, using one and the same OS they sound different, OS's sound different too, it cannot be true it should not be true yet various people have drawn the same conclusion based on their observations.

 

your argument boils down to it can not be....I agree, yet it is 

[bodiebill]

43 minutes ago, Archimago said:

[Needless to say, I hear / see no issues with modern computers/DACs.]

 

Trying to understand -- all theoretical notions aside...

Is it true that you do not hear any differences between digital gear (DACs, transports, network streamers etc.) above a certain price point? And if yes, which approx. price point?

[firedog]

On 6/26/2022 at 1:55 AM, MarcelNL said:

your ears and brain are a pretty good measurement system, it evolved over a couple of aeons...

 

Fundamental misunderstanding. 

Your ears and brain are a perception system, not a measurement system. They engage in interpretation of perceived data, not in measurement.

If they were measurement systems, science wouldn't need to take measurements. 

 

As all human perception is based on interpretation, it often is "wrong" or fooled. 

This has been shown countless times, for all of our senses. It's part of how humans are built. We are designed to to draw conclusions based on perceived data, even when the data doesn't actually support the particular conclusion. This clearly has many evoultionary and practical benefits, but means our perception of the outside world is both informed by, and limited by, prior experience. 

Our senses/perception are often "correct", but also are often "fooled" into incorrect conclusions.

 

Audio perception of a hifi system isn't some special exception to this situation.

[firedog]

On 6/26/2022 at 1:55 AM, MarcelNL said:

 

multiple post

[firedog]

sorry.

[firedog]

sorry.

[MarcelNL]

you must be right, since you keep repeating yourself

 

our hearing and brain are fine tuned to asess or approximate a.k.a measure (not in microns but still, comparing something to a known value is what most measurments are) distance and location of sounds...trust me it's proven and the accuracy may make you wonder. 

 

but you'll repeat your message since you are right because you know you are, rule nr 1

[MarcelNL]

in case I misunderstood the repetition of posts, my apology...I now see it may have been a benign mishap!

 

[firedog]

25 minutes ago, MarcelNL said:

looking back the discussion was about a difference in sonic fingerprint between W10 and W11, so around differences in OS not about USB data transmission, or any other data transmission since USB wasn IMO not mentioned anywhere. 

 

Basic truth in computer systems? eerrr who's defining what is truth and about what?  Restating that something cannot make a difference makes it a truth?

 

There are various bit perfect audio players, using one and the same OS they sound different, OS's sound different too, it cannot be true it should not be true yet various people have drawn the same conclusion based on their observations.

 

your argument boils down to it can not be....I agree, yet it is 

None of us have seen any evidence that the sonic differences exist outside of any individual's subjective impressions, arrived at under bias conditions. You haven't shown the differences exist. Stating they do doesn't make it so.

 

You are more than welcome to enjoy/believe in listening however you want.

 

That's not the same as stating that the Win 10/11 differences etc. are an established fact/truth.

[MarcelNL]

I agree, those differences are not  a given!

[firedog]

17 minutes ago, MarcelNL said:

 

 

our hearing and brain are fine tuned to asess or approximate a.k.a measure (not in microns but still, comparing something to a known value is what most measurments are) distance and location of sounds...trust me it's proven and the accuracy may make you wonder. 

 

 

Which is irrelevant to my previous point and the discussion as a whole. None of the above means human perception/interpretation of what's coming out of your speakers is "accurate" and immune from the vagaries of how or why our brain/ear complex interprets sound. What you wrote can be 100% correct and not prevent us from having "mistaken" perceptions of what we hear coming from our systems. 

 

If we are honest, we all know that we've been fooled at times when listening -

we thought some phenomenon was occuring (thought we heard some difference) and realized afterwards it didn't actually exist. 

[fas42]

8 hours ago, Archimago said:

 

Yes, latency can be an issue when it causes noticeable sync problems in audio production, but this itself does NOT change the accuracy of the data nor the sound quality.

 

 

This is where you went off the rails ... you equated the integrity of data processing with the integrity of an audio reproduction chain - you see, analogue rears its ugly head in the business as some point, because unfortunately the ear/brain can't digest numbers fed straight to the mind as being music. Tell you what, I'll organise a banking transaction system to have an D to A converter coupled to a A to D converter inserted at some point in the, er, chain ... you don't have a problem with that, do you? Numbers go in, numbers come out - current converters are excellent we're told; there should be no problems with guaranteeing correct operation - correct?

[kumakuma]

6 minutes ago, fas42 said:

Tell you what, I'll organise a banking transaction system to have an D to A converter coupled to a A to D converter inserted at some point in the, er, chain ... you don't have a problem with that, do you? Numbers go in, numbers come out - current converters are excellent we're told; there should be no problems with guaranteeing correct operation - correct?

 

Data travelled over phone lines for years as analog signals. Worked fine.

 

 

 

[fas42]

49 minutes ago, firedog said:

Which is irrelevant to my previous point and the discussion as a whole. None of the above means human perception/interpretation of what's coming out of your speakers is "accurate" and immune from the vagaries of how or why our brain/ear complex interprets sound. What you wrote can be 100% correct and not prevent us from having "mistaken" perceptions of what we hear coming from our systems. 

 

 

There is a level of accuracy which is possible from audio systems, which not all people may respond to - this would be a function of how their brains are wired, and can't be worked around. However, a good percentage would, and therefore it's a worthwhile goal.

 

And that accuracy means that the speaker drivers can't be located as the source of the sonic image in front of them - even if, literally, their lives depended upon it. Not playing games here ... a sheet is suspended in front of the speakers, people can get as close to the sheet as they like, move around as they want, but they won't be able to reliably point to where the speakers are - it will be guesswork, a guesstimate.

 

This is possible. It happened to me over 3 decades ago, and others have reported the identical thing happening for them, on other systems. You don't "mistake this" - it's an illusion, just like all the various visual ones people are familiar with; a mirage doesn't evaporate because you know it's not real; your brain is wired to see it - and see it you will .

 

In audio this can be exploited to:

 

a) Be a marker for the accuracy of the presentation

 

and

 

b) Ensure the playback of any recording is a highly satisfying experience

[fas42]

25 minutes ago, kumakuma said:

 

Data travelled over phone lines for years as analog signals. Worked fine.

 

 

 

 

Sharp response, 👍 ... but that works because a protocol is involved; the transmission data is encoded in such a manner that feedback mechanisms can guarantee error-free transfers - in my scenario, no such exists; it's raw, simple as it gets, conversion.

[pkane2001]

 

41 minutes ago, fas42 said:

This is where you went off the rails ... you equated the integrity of data processing with the integrity of an audio reproduction chain - you see, analogue rears its ugly head in the business as some point, because unfortunately the ear/brain can't digest numbers fed straight to the mind as being music. Tell you what, I'll organise a banking transaction system to have an D to A converter coupled to a A to D converter inserted at some point in the, er, chain ... you don't have a problem with that, do you? Numbers go in, numbers come out - current converters are excellent we're told; there should be no problems with guaranteeing correct operation - correct?

 

Sorry to break it to you, Frank, but all digital has already gone through many D/A and A/D translation with many 1 bit conversions both ways. Including every one of your banking transactions. Sounds like you might have a ton of errors in your bank statements. How do you deal with such uncertainty? ;)

 

[yamamoto2002]

If you are carefully prepare the experiment setup, by sending and receiving 24bit PCM on one analog cable, top 5 or 6 bit can be recovered and lower 18 to 19 bits becomes garbage. There are many methods to improve it, one example is forward error correction

[fas42]

8 minutes ago, yamamoto2002 said:

If you are carefully prepare the experiment setup, by sending and receiving 24bit PCM on one analog cable, top 5 or 6 bit can be recovered and lower 18 to 19 bits becomes garbage. There are many methods to improve it, one example is forward error correction

 

Yes, the "Get Out of Jail Free" bit is built-in error correction - now, we do those transactions with no such error checking - and your retirement fund depends upon that ... any takers?

[samotc]

The human ear detects differences in the range of milliseconds, in this way we differentiate the directionality of sounds, hence the importance of system latency in sound quality. Another important aspect is the number of hard pagefaults that occur during listening, this causes clicks and cracks during it, and I have verified that they depend on the OS and its optimization and they occur more in some than in others, that would explain the possible differences among them.

 

I attach 2 files, one from Server Core 2019 filesdisk and another from Server Core 2019 ramdisk, where you can see differences in the number of hard pagefaults, as well as others in latency.
In any case, the fact that we do not find differences between the measurements, perhaps indicates that we are not measuring the correct parameters, not that there are no such differences.

 

latency CORE2019 AUDIO FILEDISK_reposo_BIOS++.txt test_tracksRAM_BIOS++.txt

[firedog]

38 minutes ago, samotc said:

The human ear detects differences in the range of milliseconds, in this way we differentiate the directionality of sounds, hence the importance of system latency in sound quality

I don't think you can show a cause and effect relationship between system latency in the ranges we are talking about and appeared directionality of sounds. A fraction of a second of system latency doesn't translate to a "delay" in the audio. The system doesn't work that way.

 

Clicks and cracks? I don't have them in my systems. Not Win10, not Win11, and not anything else. 

If you hear them, apparently you have a different problem, not directly related to which OS, etc.

 

Sorry, but it sounds like you don't understand much about how computer systems work, and are creating cause and effect relationships where none exist.

[yamamoto2002]

1 hour ago, samotc said:

The human ear detects differences in the range of milliseconds, in this way we differentiate the directionality of sounds, hence the importance of system latency in sound quality. Another important aspect is the number of hard pagefaults that occur during listening, this causes clicks and cracks during it, and I have verified that they depend on the OS and its optimization and they occur more in some than in others, that would explain the possible differences among them.

 

I attach 2 files, one from Server Core 2019 filesdisk and another from Server Core 2019 ramdisk, where you can see differences in the number of hard pagefaults, as well as others in latency.
In any case, the fact that we do not find differences between the measurements, perhaps indicates that we are not measuring the correct parameters, not that there are no such differences.

 

latency CORE2019 AUDIO FILEDISK_reposo_BIOS++.txt 8.28 kB · 0 downloads test_tracksRAM_BIOS++.txt 8.07 kB · 1 download

 

It seems you are using 13 years old i5 750 CPU, due to spectre/meltdown workarounds the processing performance is more impacted than recent CPUs, and it does not support TPM2.0 and not upgradable to Windows 11, approx. 8.2 times slower than current i5 12500 CPU, I think it is time to upgrade your computer if it is possible.

[PeterSt]

11 hours ago, MarcelNL said:

luckily the financial world does understand that jitter matters, the Solarflare X2522 NIC card that is used a lot in that sector can be adjusted to 'ultra low jitter' and it can be 'modded' for even lower jitter by installing the software that makes that possible...for those who do ultra short trading. 

 

Wait wait wait. Now we're talking.

 

 

 

11 hours ago, MarcelNL said:

Shall we open a collector thread where folks who just want to say 'nay' not adding to the topic at hand can refer to a thread and say Nay+1 to keep the objectivist obfuscation out of all threads with topics that do not match a chapter in a text book? 

 

Scalp'm !

[PeterSt]

9 hours ago, Archimago said:

I understand how these things work

 

[PeterSt]

2 hours ago, samotc said:

Another important aspect is the number of hard pagefaults that occur during listening, this causes clicks and cracks during it,

 

It now seems that some people dedicate differences in perceived SQ are due to glitches and all (I'm only quoting the last occurrence of that). But this is rubbish of course. FIRST all needs to measure bit perfect. Thus, play two differently sounding files, capture them both digitally and compare. Not 100% equal ? Then you have an issue (probably also with your bank).

[StreamFidelity]

1 hour ago, firedog said:

A fraction of a second of system latency doesn't translate to a "delay" in the audio.

 

I agree that the latencies in human hearing have nothing to do with the latencies at the OS level. Nevertheless, low latencies in the audio computer are an advantage. Why?

 

The music file is digitally in bits (0,1). To play it, the file must be rendered and sent to the DAC. In the process, high latency can cause clicks and cracks. Latency is the time between action and reaction. For example, suppose the checksum protocol finds no match on a packet requests the packet again. There is not enough time because of slow processing. Then this information is irrevocably lost. The result is a crackle.

 

1 hour ago, firedog said:

Clicks and cracks? I don't have them in my systems. Not Win10, not Win11, and not anything else. 

If you hear them, apparently you have a different problem, not directly related to which OS, etc.

 

This is an obviously audible error. And it is true, then there are fundamental problems. No matter whether it is Windows 10 or 11. 

 

Let's move on to the unsecured knowledge. The thesis is that lowest latency in computer is good for audio. Why? I suspect that speedy processing with low latency leads to fewer errors. For example, a packet error can still be corrected quickly. I suspect that low latency allows better processing with high quality algorithms, for example with HQPlayer. I suspect improved efficiency and thus less electronic noise. 

 

Windows 11 may have an advantage over Windows 10 in core allocation. See the new Intel Thread Director. I haven't done an A/B comparison. Windows 11 works very well with latencies. That's why I use it. The image shows the latencies at the driver level. The highest latency is with the TCP/IP driver at 0.03 ms (30 µs). This very low value is achieved with a Solarflare X2522 NIC and W11 Pro.