We all fell in the W7 pitfall

[PeterSt]

W2K would be an option though ...

Criss, no, no Linux. Maybe sometime (when this small part is really finished), but it will really need a complete rewrite.
I've heard Linux implementations too, and didn't like them either.

Peter

That's interesting as, purely in theory, due to its infinite tweakability Linux could be made superior to Windows (including current 'gold standard' Vista): Are you at liberty disclosing what were those Linux implementations you heard?

It has been a longer time ago now (over two years ?) and at that time things were "measured" by achievable latency - hence how real-time is that OS really ? As you will know MS systems claim not to be better than 20ms in general which is a total laugh if you look at the 1ms I use from the start (WASAPI) which is a kind of virtual because timer resolution is wacky near 1ms (and "normal" officially can't go lower), but this was before KS. And as you know we can now reach 1/88200 at 32 bits easily which is over 100 times lower than 1ms. So, how real time do we want real time to be ?
But as with W7 (where that numbers are reached), what does it tell ? not much about sound quality ...
Of course, I start about "real time", but that at least always *was* the argument for Linux, including "tweakablility". But next ... this implies more tweaking than low latency or anything. That we know now too. On the other hand, be careful, because at watching closely there aren't all *that* much really working out tweaks for SQ. The most we apply ourselves and is outside OS related stuff (like RAMDisk, copying at the right time, blahblah), and the tweaks that really should work out for SQ ... I don't know.

Even saying that Vista is better for SQ because it was more made for it ... I don't know. What I do know though that it works reliably, and at least I can do what I want, and it behaves how I expect. Not so with W7, and it is exactly there where it goes (and went) wrong. If I say (in the program) I want A to have priority over B, this doesn't work in W7 because it seems to have its own rules. Now *that* is important to me and SQ, because I'm in the blind, as is SQ because of that.

Btw, I don't have the experience on XP here (I never really used it for playing). Not really true, because at fist there was Engine#1 (which was Direct Sound) and already that sounded better to most people than any KS/ASIO from that time. But XP has other nasty things like limited buffer sizes here and there, which went away per Vista.

Peter

[Josef]

> And as you know we can now reach 1/88200 at 32 bits easily which is over 100 times lower than 1ms. So, how real time do we want real time to be ?

Excellent question!

It's pretty impressive that you can get 11 microseconds latency with your DAC - I am not able to even get close to those numbers
Could it be because of USB (my case) vs PCI (your case)?

If such low latencies are indeed important then due to way it works seems USB would be doomed in this regard?
 
In other words:

- what is your position on need for low latency?
- and do you think that for 384kHz playback a latency of 1/384 i.e. 2-3 microseconds might be needed for absolute maximum performance?

I have read quite a few discussions on this subject and people generally seem to fall into two camps: Either they believe low latency is absolutely necessary for high-quality audio playback or they completely dismiss the idea and even go in opposite direction by _increasing_ latency ('just get bigger buffers' camp).

Also, on this forum while pretty much everyone has moved to KS it seems that there is significant amount of people using Adaptive vs Special mode. Wouldn't it be interesting to find out if people with USB statistically tend to prefer Adaptive and vice versa? (Would you consider making a poll on the forum? )

(Please note this post is not a red herring: I've been reading around and it seems 'low latency playback' is a super-controversial topic that spilled a lot of blood in some internet forums. I'd completely understand if you'd rather choose not to answer and risk being vilified in some of those other forums. However, as someone who has spent years on getting 44.1 to 384kHz and making it work in both SW (ArcPrediction) and HW (your NOS DAC) it obviously makes you uniquely (and much more) qualified to voice your personal observations). 

[Josef]

> If I say (in the program) I want A to have priority over B, this doesn't work in W7 because it seems to have its own rules.

Wow - I didn't know it was that bad I thought it's just design changes in kernel which made W7 lose it's luster over Vista....
But if you are seeing such bad behavior then maybe it's just a bug and there is still hope for W7 - Perhaps it's an idea to try SP1 (think it's in RC stage now so pretty much done) and see if maybe MS fixed it?

[PeterSt]

I don't have much time right now, but the subject is interesting, I'll try to spend a few lines on it for now;

First off, *that* buffer sizes influence is clear to everyone I think (by now).
Those claiming that low latency does nothing, often go along with the statement "it isn't necessary for audio playback" (while it is for real time (midi etc.) live playing).

Next the subject has to be divided into two parts :
1. The buffer size of the device;
2. The buffer size of the software preceeding that.

#1 already has been audible to me always, and as it always was : the lower the better. But let's skip this one for now, because it's a real rough one, compared to what #2 can do.

#2, before, played a role just the same, but I think it was over-shadowed by #1. This was before low latency KS (Special Mode), because generally the software buffers were larger than the device buffer. It sure did something, but #1 ruled over it.

Today, with (actually ultra) low latency KS things have become very different. Now, what's in my mind is that the software is operating at such low latency level in communication with the driver, that actually this operates at (near) the same level as jitter. That is, it mathematically should be, if only that part of the process would be subject to jitter, which it officially is not. But :
While everything within the PC hence software officially is not (up to and including the driver), it has been proven easily that whatever it is, it influences. Not jitter, than something else (noise that incurs for jitter at the other end, whatever). However, no matter what it is, if it influences, now it influences at the jitter level (because it operates at the (near) sample level).

I come to this (and after reading your post Josef), because I suddenly seem to recognize plain jitter at using KS Special Mode. So, it doesn't work out, but *does* imply more detail. I think everybody would agree with the latter. As said more often, for special "ambient" music this is superb, because almost always this is about normal detail (but detail you won't find in e.g. Jazz) while nothing is there to disturb like a woman voice or real violin. It's synth details, often added with drums and cymbals, may they be sampled or not. It can't go wrong, and the detail is "needed" because it's there (especially on the synth level). Now, why wouldn't this work with a woman voice etc. ?

Let's say that the detailed communication is able to separate the samples better at that level of the communication (like in : use 8 samples for buffer size, and per 8 samples you'll have additional detail). Again, *how* it works we don't know yet, and we only know that it *does* influence.
In the mean time, and just *because* we operate at this "sample" level, the timing becomes outrageously important. So, the buffer runs near empty, and just in time it's refilled again, and this happens each 8 samples. Now, the message "buffer near empty" is not synchronized with the filling of it (both will run in different threads to begin with -> driver and playback software) so this already may create a huge amount of jitter (noise jitter, haha). So, the message as well as the filling both raise current draw in the system, and both create noise, that by itself influencing the DAC ...

Suppose this holds true, there will be an additional power going on (and this is a very strong power) : the power of resonance. Again, the timing of the message (buffer runs out) will have no real relation to the filling of it, and you can bet that it can happen that at one time the buffer is refilled while 2 samples are still there to play, while another time this will be one only. In the long term (but think under ms here) this will be a repeating pattern, and the current draw (hence noise) will show a pattern too. And at the other end this emerges as a jitter pattern.

This jitter pattern will not be a commonly known pattern, already because it was created by two sources. So, not a sine, but a more nasty pattern. So, theoretically it will produce an "unknown" (jittery) sound to us.

I shall admit that most (if not all) of this is based upon a perceived *real* better detail to begin with, which is a dangerous assumption. I mean, for the same it could be about a more squary sound, which incurs for faked detail. But I don't think it is, and I never heard someone perceiving it like that. But well listening through this all is the most difficult, and possibly I can't do that.

Peter

[bushpilot]

W2K would be an option though ...

Criss, no, no Linux. Maybe sometime (when this small part is really finished), but it will really need a complete rewrite.
I've heard Linux implementations too, and didn't like them either.

Peter

That's interesting as, purely in theory, due to its infinite tweakability Linux could be made superior to Windows (including current 'gold standard' Vista): Are you at liberty disclosing what were those Linux implementations you heard?

It has been a longer time ago now (over two years ?) and at that time things were "measured" by achievable latency - hence how real-time is that OS really ? As you will know MS systems claim not to be better than 20ms in general which is a total laugh if you look at the 1ms I use from the start (WASAPI) which is a kind of virtual because timer resolution is wacky near 1ms (and "normal" officially can't go lower), but this was before KS. And as you know we can now reach 1/88200 at 32 bits easily which is over 100 times lower than 1ms. So, how real time do we want real time to be ?
But as with W7 (where that numbers are reached), what does it tell ? not much about sound quality ...
Of course, I start about "real time", but that at least always *was* the argument for Linux, including "tweakablility". But next ... this implies more tweaking than low latency or anything. That we know now too. On the other hand, be careful, because at watching closely there aren't all *that* much really working out tweaks for SQ. The most we apply ourselves and is outside OS related stuff (like RAMDisk, copying at the right time, blahblah), and the tweaks that really should work out for SQ ... I don't know.

Even saying that Vista is better for SQ because it was more made for it ... I don't know. What I do know though that it works reliably, and at least I can do what I want, and it behaves how I expect. Not so with W7, and it is exactly there where it goes (and went) wrong. If I say (in the program) I want A to have priority over B, this doesn't work in W7 because it seems to have its own rules. Now *that* is important to me and SQ, because I'm in the blind, as is SQ because of that.

Btw, I don't have the experience on XP here (I never really used it for playing). Not really true, because at fist there was Engine#1 (which was Direct Sound) and already that sounded better to most people than any KS/ASIO from that time. But XP has other nasty things like limited buffer sizes here and there, which went away per Vista.

Peter

I don't compare my technical knowledge with yours. I am simply a long time audiophile. I tried W XP and W7 with J River software. Was not happy with the sound with either and decided to try Linux. At 75 years of age, I am not really literate with computer operating systems so this was a real challenge for me.

Tried 8 different Linux distro's and settled on Ubuntu 10.04 as I was able to upgrade it to the latest alsa drivers. Once this was accomplished, I have been very satisfied with the sound coming from my Benchmark HDR coupled to my McIntosh MC-275 amp. Sound quality is simply the best I have heard.

Happy that you have had success with Vista and hope you find may followers with your software.

Bill

[ivo]

I don't compare my technical knowledge with yours. I am simply a long time audiophile. I tried W XP and W7 with J River software. Was not happy with the sound with either and decided to try Linux. At 75 years of age, I am not really literate with computer operating systems so this was a real challenge for me.

Tried 8 different Linux distro's and settled on Ubuntu 10.04 as I was able to upgrade it to the latest alsa drivers. Once this was accomplished, I have been very satisfied with the sound coming from my Benchmark HDR coupled to my McIntosh MC-275 amp. Sound quality is simply the best I have heard.

Happy that you have had success with Vista and hope you find may followers with your software.

Bill,

I am almost 2 times younger than you

Are you saying that Linux+ALSA is better SQ than any Windows OS? Can you describe what is the difference like? How huge is that?

Ivo

[PeterSt]

Hey Bill, that's quite some thing you've achieved there. I mean your age in combination with all this hocus-pocus. But then you had quite some years to warm up of course. Anyway, hats off.

Ivo, Bill comes from JRiver area and it is not "SQ" related at all, but adviced for its useability. And so, many do.

I can't be sure, but I don't think Bill ever got XX running which is not his fault (merely mine of course).

I think it goes too far to say an OS "sounds" as such, but it will be the players for that OS making something of it (within that OS). Maybe W7 is an exception, haha.
It would be great If Bill can and wants to answer the question, but keep in mind the context.

Peter

[Chriss]

Hello Bill
regards for started up Linux+Asla it's not an easy job for any age! Well I tried it 4 months ago and I must say that Foobar+ASIO on UNtweaked Win7 sounds better to me. And there is HUGE difference between Tweaky Vista and 'Basic' one. Plus RamDisc and SSD (I'm playing from 8gb Flash and it's  ) Keep the experiments and try our little digital eden
Regards, Criss.

[Josef]

>Tried 8 different Linux distro's and settled on Ubuntu 10.04 as I was able to upgrade it to the latest alsa drivers. Once this was accomplished, I have been very satisfied with the sound coming from my Benchmark HDR coupled to my McIntosh MC-275 amp. Sound quality is simply the best I have heard.

Wait: 75 years young, playing music via computer (after testing EIGHT! _Linux_ distros!) and listening on McIntosh 275???  

Bill - Don't tell me you actually got your McIntosh in 60's when they originally came out? That would be too much

We need to put this into perspective guys: McIntosh 275 is still made today based on original design from, what,  1961?!!! (gosh, I personally converted from transistors to tubes after hearing that same amp....)
I mean, can you imagine Apple selling Apple II in 2027 and some kid even looking at it?????

Wow, I wonder when I'm 75 if I'd still be up to try, hmmmm, Windows 17 or 8 Linux 203x distro's
 
Bill - Respect!
Post of the month!

PS Peter - May I suggest something gets done about that GUI (a default 'simple' mode perhaps?) if for no other reason than to get Bill to still another level: I mean, where on Earth can you find audiophiles like this? No such forum anywhere.....

[PeterSt]

Josef, I must honestly say that when I wrote this :

I can't be sure, but I don't think Bill ever got XX running which is not his fault (merely mine of course).

I thought : gosh, hopefully it isn't true, but it just as well can. But I didn't think much of the GUI here, but merely about the problems people can encounter I just don't know about. On that matter, the more is posted about this (instead of leaving the game before it started), the better it will be. But who does that ? very few I'm afraid.

Btw, amongst us are more near 70 guys than you can guess. But I think 75 beats it all. Really really great to hear.

Peter

[manisandher]

Post of the month!

Agreed!

But surely anyone who can handle 8 Linux distros can handle little ol' XX... come on, it's not as difficult as that!

Mani.

[Josef]

But surely anyone who can handle 8 Linux distros can handle little ol' XX... come on, it's not as difficult as that!

While you do have a point there, as Peter said, we don't really know at the moment what was the actual turn-off reason. Hopefully, Bill will elaborate on specific issues.

But, now that you ask, (Peter, it's manis fault ) regardless of specific problem, I think an easy case can be made at least for default 'sizes': first, XX window is small hence everything feels pushed into an artificial straight-jacket (why is it small? to save space for something else? what? and making XX full-screen will leave everything as unreadable on 1024x768 as it is on 1920x1280 (even worse)?), font is too small hence hard to read (again, why? do most people really demand to see every song title in playlist on-screen even if they need magnifying glass otherwise they refuse to hit 'play'?) and what about those super-tiny buttons all over the place that even I (admittedly, a tiny bit under 75) have trouble hitting much less figuring out what they mean (icon for Quality Panel means.....what? And button to the right with 'X' means CoverArt Cache, obviously, right?) so I'm forced to hover the mouse over and read explanation in an even smaller font size? (and don't even get me started on opening/collapsing panels again seemingly saving space for ... what exactly?.....)

Heck - don't get me wrong, I really don't mean to be negative (rather: constructive criticism).
Of course JRiver a.k.a iTunes-like interface also has deficiencies & can be improved as well!
But wouldn't it make sense to do it the way most people are used to instead of re-inventing the wheel?
So most people can just 'use it' and enjoy the music which was the point anyway - wasn't it?

[PeterSt]

Hi Josef - The answer is fairly simple ... because people with 640x480 screens exist ...

But my opinion : if you work with the 1280x1024 standard (and its wide screen derivals), I really don't see a problem. But since you talk about such a derival (1024 x 768) - which is larger (unless you use that on an 8" screen), maybe I just don't understand.

Please notice that the only option is making the screen larger - and with that the buttons. Not only the buttons, because they won't fit. And I'm thinking touch screen here, where larger buttons my physically still fit, but can't be touched without touching the adjacent.

I said it before, if we think that 1920x1080 is normal, well, ... we've got something wrong, unless it's on the 32" and larger tV.

I really appreciate the input on this, and if people have problems on 1920x1080 it really *is* a problem, but not really my fault. It's unusable at normal display sizes (call that 20" at most), unless for photos / movies and CAD stuff which is *made* for that. All other is not, and you'd have the same problems.

If I let go of the 640x480 (which also is not normal of course !) then no "PC" inbuild screen (like OrigenAE) could be used.

Make two versions ? brrrr
Make it scaleable ? later.
Make it scaleable right from the start ? next time.

So, always my fault, but nothing much to do at this moment. I think.

Thanks a lot,
Peter

[LydMekk]

We have discussed the size problems before. I for one is staying with 09y-8c for now due to the visibility on my 1920x1080p displays.

The sound is very good on 09y-8c here.

[Flecko]

I tested Vista Ultimate x64 SP2 against Windows 7 x64
Both were booted in debug mode for best performance. In debug mode most services are shut down.
The better latency that could be achieved with vista was obvious. I got about 30us (best about 10us, worst about 60us). With W7 I got about 100us (best about 80us, worst about 160us).
I tried different settings with Vista but at the end I choose the same as I use with W7. Everything was set to the same paramter. registry was changed in the same way (usb IdleEnable=1). I restarted several times and listened to the same two tracks on each OS (the hunted-kodo, Vicente Amigo). It always took a while to reboot. But at the moment I would say, they sound almost the same. That is my first Impression and as always, I might change.