Clean power for your.... SSD/HDD...?

I'm afraid not. There's too much abstraction at work. At the lowest level, the HDD will buffer sequential reads into its cache; the music player will likely perform some degree of buffering as well, even if the whole song isn't read into memory at once.

Time domain doesn't come in to the equation until the final bitstream is going to the DAC.

 

Exactly, but I'm wondering where exactly that real-time bitstream starts. Presumably the reason why folks claim to hear differences with USB converters follows the same logic, in that now that our USB driver is outputting a real-time (that is, in the time domain) signal, any signal integrity issues will not be buffered and will flow through the system.

Also, I think we need to define "buffer" more precisely. An asynchronous FIFO vs something else?

 

It wouldn't be before the audio player writes to the output device, whereupon it is handed off to a low-level audio API e.g. WASAPI. At that point it should be a true stream with a clock rate. By then the effects of the storage medium would presumably be long gone; the music player as well, as stream creation is delegated and abstracted away to the audio API. Even at that point there will be some degree of buffering between the player and the low-level interface.

 

I need to better understand asynch USB, but in that case it seems the DAC pulls data from the host and takes over clocking if I am understanding correctly. The clock area is where I'm fuzzy with async.

 

I always assumed the supposed difference in audio performance had more to do with the power drain/ripple/voodoo of the devices leaking noise and junk into the power supply, not the method of data storage/retrieval.

 

The way I can tell if one is fixing imagined problems is two-fold...

1) The fix is applied not to correct any specific audible deficiency, but only to make the SQ "better".
2) The improvements thereafter described lead off with "seems to" or "sense of" and then a bunch of vague fluffy terms.

An example would be:

"When I taped the $100 bag of stones to my $2,500 8 gauge power cord it seemed to increase the focus and there was a sense of enhanced spaciousness to the sound"

 

If disk choice etc would actually make a difference, simply run stress test tool like http://www.passmark.com/products/bit.htm which stresses all of your cpu/mem/gpu/hdd to the max! Differences would be immediately obvious with it running, no?

 

Sometimes, the impossible happens. Sometimes, the possible, frustratingly, doesn't (especially with Microsoft software --- bias proudly paraded) How many times have those here who have dealt with computers for stuff other than sound, seen something, scratched their heads, and said, that should not happen; that cannot happen. Of course, there must be a reason for it: it is just that we are not aware of it. Fine: we know something happened, we don't know why. Assuming that it does not risk lives or threaten to close the business, those of us who are not engineers at a seriously deep level find a way to make the circumstances in which the thing we don't want to happen doesn't, and what we do want to happen does, with reasonable reliability, and in proportion to the skills what we are being paid to have.

This is what happens in the real world, outside management studies and audiophoolia. Inside those worlds, people make stuff up. Or pretend they know what happened. They have no basis for their claims, it is imagination with a few bits of jargon thrown in, or a few real-life effects or technological facts, which are true but irrelevant, thrown in.

Those computer audio sites began life by providing reasonable information. Even today, one can get video interfering with audio: way back then in the earlier versions of Windows such stuff more regularly needed tweeking.

I never was a programmer (or a hardware engineer) just a hands-on systems manager, with my hands on everything from the network cabling to the accounting database. I know one little C program (or did: even basic syntax escapes me these days) though. It goes something like main(){while(1){fork()}} and it means forever, go on creating new processes Each process actually does nothing, but it will soon bring the computer to its knees. I never tried running it on a machine playing audio. Would it affect it? My guess is Sure! It will stop. (Actually, I think it does little more than use up one resource, the process table, but it is more than a decade since I even had to think about such things).

So, anyway, what was I saying... oh yeah... the impossible happened, right? What next? jump to conclusions and post some assumptions on computer audiophile, which then get adopted into the lore of audiophilia? Or... throw a blind test at it. If a person can hear green ink on the CD edge, basic inquisitiveness demands that, at least, they get someone else to change the discs while they don't look. Wouldn't pass muster on hydrogen audio, but it is a start.

Rubber chickens. I prefer the boneless ones. Boned chickens cause bit decay.

 

Hahahaha ohhh man. Thank you for the good laugh. Look up memory hierarchy on your computer. Music goes from your hard disk, loaded to RAM, then into your USB buffer through DMA (direct memory access), and then gets transferred to your USB DAC.

A better power supply for your computer will make an insignificant difference. Memory is transferred in large chunks at a time because access times are so long (in relative computer terms). The USB interface from your PC to the DAC is probably the only place the power supply noise will have ANY effect what so ever. Even then your talking about noise at about, and I'm being very generous here, -80dB. The air conditioning in your house produces louder "noise" then the effects of a "bad" power supply. Hell, I'm willing to bet the electrical resonance of your lightbulb is more audible then the noise that the power supply adds to the USB interface.

 

True - there is no reason to assume a linear relationship between computer resource usage and whatever effects might be going on. It could well be that the differences are non-linear, seeing as we have little understanding of the mechanisms at work, nor the nature of the supposed changes or optimisations.

In my experience, running stresstests tends to cause dropouts, but it was a long time ago when I did these "research" and can't rely on memory for any other more ephemeral differences. Still I would encourage experimentation, to their own standards, by anyone who is interested to find out for themselves with their own systems. A lot of the software and tweaks I used to think were doing good, now with my system in the current state, and my own evolution of listening skills or possible changes in hearing I can determine that they were actually doing harm.

Right now my philosophy is to keep things sensible and practical. There are still things that operate outside common sense, but the way I see it if one understands how a computer works, how a DAC works, how hifi works, to understand the nature of these things and work with them rather than against them. Generally software makers optimise for certain operating systems, timer settings, resource scheduler settings, so if you change these things you are probably working against the software!

Similarly a computer is designed in a certain way, and responds in performance to component selection in a certain way. If you are fighting the underlying nature of a computer eg by changing out clocks, bypassing caps, using unusual power supply and grounding schemes without understanding the underlying nature of computers, you could probably make things worse.

If you understand how a computer works, and how software works, and do things that work with the system rather than against it ie freeing up resources, disabling unneeded components (but not ones that are useful) using hardware that improves performance rather than hinders it, and develop good critical listening skills you may well be able to improve upon a general use workstation.

In my experience this generally means a minimalist audio endpoint, or a more substantial headless server which can be administered over a network. Multi purpose workstations in my experience can work decently well but produce a lot of noise, and can be improved upon. I speak from experience as I have both a high powered gaming workstation and a dedicated HTPC machine built to be a music server. I was comparing both a few months ago with a view to sell some of the gear I don't need, but the music server has stayed. I was even using the gaming workstation exclusively for a while, seeing how far I could take it, but ultimately it fell short of the dedicated machine. Other gear such as my USB/SPDIF converter and some cabling (and software) did not fare so well and is being sold. IF my HTPC machine did not present value to me I would have definitely sold it off.

I am generally wary of solutions which insert into the USB path and attempt to improve things after the fact. They way work if things are initially so bad that they can be improved. Generally though I prefer to try and fix the underlying problem directly, and use the best low noise (electrical and acoustic) computer hardware and software and also the best high bandwidth (machine made), well shielded USB cable (eg Wireworld or Light Harmonic). For many this may not be practical or affordable, and bandaid measures such as USB reclockers or isolation may be a better solution.

 

Absolutely. Nothing magic, no unicorns and certainly no pink mist. If bad noises happen when the screen is updated, there is an problem with interrupts. The fact that I may have struggled with that sort of thing for hours means that I don't understand it --- not that nobody does. Lots of people do. There's a guy called Rui Nuno Capela, for instance, who knows a thing or two about how to get Linux machines to work better in this respect. I'm glad he does!