Forget storage comparison. Even the same storage can sound different with different file copying methods

[plissken]

Without casting aspersions on what anyone believes -- I am trying to give rational explanations of:

 

a) electronic reasons why copies of files may sound different -- not magic

b) ways to reduce that

 

Your reason that you gave happens at the MR head of the drive in your Wiki. The data in the DAC buffer went from MR head to RAM on the HD itself, to SATA Buffer>RAM via DMA>RAM set aside by the Playback Application>To USB host buffer, bitstreamed over a USB cable, and re-clocked on the other end into the DAC's buffer.

 

If they are coming out with the same CRC they are going to sound identical. Same as if I make 10 copies of a 264 encoded movie.

 

Computers are there to insure that the experience is 100% the same every single time barring gross failure at some point.

 

It's the entire freaking point behind computers.

[esldude]

snip

 

Computers are there to insure that the experience is 100% the same every single time barring gross failure at some point.

 

It's the entire freaking point behind computers.

 

Oh you must be mistaken. At least in the audiophile world not only is that point wrong, it isn't even possible. Everything matters or did you miss that somewhere. The color of the capacitors on the motherboard effect the sound. You just have to get everything else ship shape first before it becomes aurally obvious. The effect is easily explained too. The quality of the power supply will effect the vibration of those capacitors. That vibration effects how photons leaving those capacitors happens and obviously color is related to photonic activity. As changing color changes photoic activation via vibration one need not even look at the quantum effects to see one is hopeless to think computers can provide the same experience 100% of the time just because some gross level of bits line up more or less all the time.

 

This indicates you are forgetting the primary freeking point behind all things audiophile.

[Sal1950]

Oh you must be mistaken. At least in the audiophile world not only is that point wrong, it isn't even possible. Everything matters or did you miss that somewhere. The color of the capacitors on the motherboard effect the sound. You just have to get everything else ship shape first before it becomes aurally obvious. The effect is easily explained too. The quality of the power supply will effect the vibration of those capacitors. That vibration effects how photons leaving those capacitors happens and obviously color is related to photonic activity. As changing color changes photoic activation via vibration one need not even look at the quantum effects to see one is hopeless to think computers can provide the same experience 100% of the time just because some gross level of bits line up more or less all the time.

 

This indicates you are forgetting the primary freeking point behind all things audiophile.

Dennis, I grade you a A+ on the above paper and suggest you present it to the AES for publication.

[davide256]

HUMMM, OK that's just ridiculous.

 

It's not if you are educated in the physical sciences and electronics engineering...

[jabbr]

Thank you.

 

I think a correct understanding of what happens in a computer at the physical (hardware) level and the data (software) level is important so that we focus our time and money on the things that really matter when it comes to improving sound quality.

 

The reality is that NOT everything matters and when it comes to things that do matter, some things matter a lot more than others.

 

Seems we have two groups : everything matters and nothing matters

 

I don't focus my attention --at all-- on tweaking my application server. I like to keep an open mind.

[kumakuma]

Seems we have two groups : everything matters and nothing matters

 

I said "NOT everything matters", not "nothing matters." Not quite the same thing.

[plissken]

Seems we have two groups : everything matters and nothing matters

 

I don't focus my attention --at all-- on tweaking my application server. I like to keep an open mind.

 

Data the ends up intact and copied over and over and over and over and over without error from HD platter>drive buffer>SATA controller buffer>DMA to a register set aside by the CPU>Audio playback application buffer set aside the the CPU>USB controller buffer> DAC buffer matters.

 

What matters it the CRC is correct throughout the process.

 

What doesn't matter is your flat refusal to understand how this works.

[Sal1950]

Seems we have two groups : and nothing matters

I don't think there's a single person on this site that believes that. But some here like to use that accusation to sidestep

science and facts.

[Paul R]

I'm making a point about a thread much like this one. One that Jim was kind enough to give me a thumbs up and free JRiver licensing. It was for being 100% correct.

 

The positions about Data, Block Storage, Multiple copy stacks, people posting PDF's, slide shows, links to articles that they are incapable of using to make their point is the real shitting on the table and Alfe/Sandyk/Jabbr are the ones doing it.

 

That you haven't figured that out leaves me with the idea that everything you say is tinged with the idea of you being a sycophant to the subjective crowd here.

 

If you can't beat them with fact just dazzle them with bullshit. Right Paul?

 

Anyone that wants to push me off my data driven position feel free to do so or just shut up.

 

Boy, you sure are putting your foot in your mouth.

 

So, you are here being nasty to almost everyone in the hopes of gaining free software or other goodies from people. What a tosser.

 

-Paul

[Paul R]

Oh you must be mistaken. At least in the audiophile world not only is that point wrong, it isn't even possible. Everything matters or did you miss that somewhere. The color of the capacitors on the motherboard effect the sound. You just have to get everything else ship shape first before it becomes aurally obvious. The effect is easily explained too. The quality of the power supply will effect the vibration of those capacitors. That vibration effects how photons leaving those capacitors happens and obviously color is related to photonic activity. As changing color changes photoic activation via vibration one need not even look at the quantum effects to see one is hopeless to think computers can provide the same experience 100% of the time just because some gross level of bits line up more or less all the time.

 

This indicates you are forgetting the primary freeking point behind all things audiophile.

 

Ah common Dennis - that little pest is just trying to aggravate people so he can claim a right to fame and garner free stuff from unsuspecting vendors.

 

You know as well as I do that some things matter, some do not, some things matter more than others, and what matters can change depending upon the system, the environment, and the source material.

 

-Paul

[davide256]

HUMMM, OK that's just ridiculous.

 

SATA controller and drive is an electrical perturbation to the computer system...removing it parallels what the microRendu offers with SD card boot. Every device added to a computer bus makes the system design more complex for an engineer to control system stability and avoid the "law of unintended consequences".

[Sal1950]

SATA controller and drive is an electrical perturbation to the computer system...removing it parallels what the microRendu offers with SD card boot. Every device added to a computer bus makes the system design more complex for an engineer to control system stability and avoid the "law of unintended consequences".

 

Yea OK, micro$endu, magic dust noise, and all that. Your right. LOL

[jabbr]

Data the ends up intact and copied over and over and over and over and over without error from HD platter>drive buffer>SATA controller buffer>DMA to a register set aside by the CPU>Audio playback application buffer set aside the the CPU>USB controller buffer> DAC buffer matters.

 

What matters it the CRC is correct throughout the process.

 

What doesn't matter is your flat refusal to understand how this works.

 

You keep making the same buffer argument. I you read back, I've made the same argument on numerous occasions. I understand how it works very well.

 

You fail to take into account the possibility that the HDD may create electrical noise. While your bits from last year or whenever are sitting in memory, or on the USB card, the electrical noise created by the current readout is flowing through the ground, power supply, data lines whatever. Is this always significant? Not saying that but in certain circumstances elongated memory bus lines can cause increased noise.

 

To be clear I am not suggesting that there is any issue with the integrity of your precious bits, rather noise generated by the analog readout circuitry (for example) there are other issues with high speed digital design, PCB layout, transmission line issues etc etc. (I own both of Johnson & Graham's books on these topics.)

 

There is a lot more electrical engineering work that goes into designing the motherboards and PCIe cards that you use without any worries than you give credit for.

[jabbr]

I said "NOT everything matters", not "nothing matters." Not quite the same thing.

 

Sorry, I mean't to preface that by saying that I agree with what you said. I don't mean to suggest that everyone hear falls into one of the two extremes. I certainly don't. I agree that "NOT everything matters". But before blindly dismissing people, I try to understand if what they are reporting might be real. As I've said many times: I can see, from an electronics point of view, how two copies of a bitwise identical file might sound different on a particular system, but these differences aren't preserved by file copy operations (duh: or proper buffering). Point being that something that is heard for real on one occasion on one system does not generalize to all occasions for all systems. I am genuinely surprised that I am getting such pushback here.

[kumakuma]

Sorry, I mean't to preface that by saying that I agree with what you said. I don't mean to suggest that everyone hear falls into one of the two extremes. I certainly don't. I agree that "NOT everything matters". But before blindly dismissing people, I try to understand if what they are reporting might be real. As I've said many times: I can see, from an electronics point of view, how two copies of a bitwise identical file might sound different on a particular system, but these differences aren't preserved by file copy operations (duh: or proper buffering). Point being that something that is heard for real on one occasion on one system does not generalize to all occasions for all systems. I am genuinely surprised that I am getting such pushback here.

 

Not from me. Pretty well everything you've said so far makes perfect sense.

 

I also think that you and plissken agree more than it may appear from your exchanges in this thread. The key issue is that you are focusing on the HW side of the equation and he (?) is focusing on the SW side of things.

[89reksal]

Yea OK, micro$endu, magic dust noise, and all that. Your right. LOL

 

No, "YOU'RE" right. There goes your credibility.

[mansr]

You keep making the same buffer argument. I you read back, I've made the same argument on numerous occasions. I understand how it works very well.

 

You fail to take into account the possibility that the HDD may create electrical noise. While your bits from last year or whenever are sitting in memory, or on the USB card, the electrical noise created by the current readout is flowing through the ground, power supply, data lines whatever. Is this always significant? Not saying that but in certain circumstances elongated memory bus lines can cause increased noise.

 

To be clear I am not suggesting that there is any issue with the integrity of your precious bits, rather noise generated by the analog readout circuitry (for example) there are other issues with high speed digital design, PCB layout, transmission line issues etc etc. (I own both of Johnson & Graham's books on these topics.)

 

There is a lot more electrical engineering work that goes into designing the motherboards and PCIe cards that you use without any worries than you give credit for.

 

Sure, a hard drive creates some noise, but no way in hell can slight differences in the microvolt squiggles representing two identical bit sequences cause audible, or even measurable, differences at the DAC.

[jabbr]

Sure, a hard drive creates some noise, but no way in hell can slight differences in the microvolt squiggles representing two identical bit sequences cause audible, or even measurable, differences at the DAC.

 

And you know this how? Because you have designed commercially successful storage media? You probably are smart enough to have an inkling of knowledge of the real complexity of some of the modern storage devices. So we have someone like "alfe" who is an actual R&D scientist with actual experience designing products for actual international corporations in this exact area, and I should believe your "no way in hell"? He has said that he is under strict NDA for details, so you can use some imagination to consider that he may be working on something in this field. He points to publicly available documents.

 

In any case it takes more than microvolts to fire a laser or step a stepper or seek a seeker e.g. Patent US7274529 - Disk drive with adaptive actuator braking upon unexpected power loss - Google Patente seems to suggest 12Volts ... or 5 or whatever... (and this is but one of numerous articles that many be helpful)

 

Not hard to imagine a two files having different block access patterns causing significantly different power spikes.

[Sal1950]

Sure, a hard drive creates some noise, but no way in hell can slight differences in the microvolt squiggles representing two identical bit sequences cause audible, or even measurable, differences at the DAC.

That in a nutshell is the beginning and end of the truth.

Folk believing otherwise need to reexamine their conclusions for the true reason behind what they are hearing.

"you have excluded the impossible, whatever remains, however improbable, must be the truth"

Something amiss in your testing process or expectation bias in sighted listening conclusions.

[plissken]

You keep making the same buffer argument. I you read back, I've made the same argument on numerous occasions. I understand how it works very well.

 

You fail to take into account the possibility that the HDD may create electrical noise. While your bits from last year or whenever are sitting in memory, or on the USB card, the electrical noise created by the current readout is flowing through the ground, power supply, data lines whatever. Is this always significant? Not saying that but in certain circumstances elongated memory bus lines can cause increased noise.

 

To be clear I am not suggesting that there is any issue with the integrity of your precious bits, rather noise generated by the analog readout circuitry (for example) there are other issues with high speed digital design, PCB layout, transmission line issues etc etc. (I own both of Johnson & Graham's books on these topics.)

 

There is a lot more electrical engineering work that goes into designing the motherboards and PCIe cards that you use without any worries than you give credit for.

 

Intel and AMD control the reference designs. They know and understand high speed signalling and grounding.

 

Keep throwing up the red herrings. Maybe they'll turn into Mackerel. It's the DAC's job to reject noise and properly designed DAC's do.

 

Benchmark ran 100 foot of cable and ran USB over it. Their DAC was 100% able to extract the data and play back perfect audio.

[plissken]

 

To be clear I am not suggesting that there is any issue with the integrity of your precious bits, rather noise generated by the analog readout circuitry (for example) there are other issues with high speed digital design, PCB layout, transmission line issues etc etc. (I own both of Johnson & Graham's books on these topics.)

 

Circling back to the original post: You can't have your cake and eat it too. Playing two identical files off of the same storage medium and stating there is an audible difference due to what utility wrote those two files, and it's all determined by the disk controller, into same size blocks.

 

My offer still stands of sending out a flash drive repeating the OP's process to put another stake through the heart of this particular undead creature.

[jabbr]

Circling back to the original post: You can't have your cake and eat it too. Playing two identical files off of the same storage medium and stating there is an audible difference due to what utility wrote those two files, and it's all determined by the disk controller, into same size blocks.

 

To be very clear, there is no issue with *me* having my cake and eating it to. You should carefully reread my posts. I never suggested that such file layout differences are under the control of a particular software utility. I only said that if such copies have audible differences, that this could be due to physical layout issues with the physical file.

 

To be clear: I would call that an artifact. ?

[plissken]

To be very clear, there is no issue with *me* having my cake and eating it to. You should carefully reread my posts. I never suggested that such file layout differences are under the control of a particular software utility. I only said that if such copies have audible differences, that this could be due to physical layout issues with the physical file.

 

To be clear: I would call that an artifact. ��

 

You are missing a point that you just made: The reading of those files are happening in bursts and they are happening ahead of when you are listening to them.

 

What you are listening to at any point in time was read and copied multiple times prior to hearing it. Think about the bandwidth available on the SATA bus an how quick it can transfer information.

 

It makes GB Ethernet at 90MB / Second downright plebeian.

 

You keep assigning real time attributes to non real time events.

[jabbr]

You are missing a point that you just made: The reading of those files are happening in bursts and they are happening ahead of when you are listening to them.

 

What you are listening to at any point in time was read and copied multiple times prior to hearing it.

 

You keep assigning real time attributes to non real time events.

 

1) Why don't you go back to Post #55 where I don't talk about sound at all and point out exactly what about that is "mass stupidity"

 

2) Who cares about real time? The beginning bits could be hitting the DAC as the readout power spikes from the middle of the file are being generated. There is no obligation that the power spikes from reading those bits have to affect the DAC process of those same bits. The DAC can be randomly affected by power spikes generated from many operations. Alfe was trying to tell you that this depends on the buffer sizes. Why do you assume that the entire file is read at once? I've called this an artifact, and artifacts can happen only sometimes and only in certain situations. I've never said otherwise. Ever.

[plissken]

1) What don't you go back to Post #55 where I don't talk about sound at all and point out exactly what about that is "mass stupidity"

 

2) Who cares about real time? The beginning bits could be hitting the DAC as the readout power spikes from the middle of the file are being generated. There is no obligation that the power spikes from reading those bits have to affect the DAC process of those same bits. The DAC can be randomly affected by power spikes generated from many operations. Alfe was trying to tell you that this depends on the buffer sizes. Why do you assume that the entire file is read at once?

 

Take a white noise file and grab a measurement mic. Start continuous play and start a file copy process. Watch the output stay the same in the spectral analysis, CSD, or other plotting.

 

How about this: we can do a remote control session where you listen to a favorite track. I'll at predetermined random interval copy data from the drive that the track is on to a usb drive and back so both read and write operations are performed.

 

You tell me when those events are taking place since this is the crux of the claim you are making.

 

I can screen cast the session. Of course this won't happen because you really don't believe what you are trying to sell.