USB Audio Driver Registry Buffer tweak

[manisandher]

I have nothing really to contribute to this thread other than... only a brave man/woman would take Peter on when it comes to computer audio.

Those of us who've been here for a while have seen many 'contenders' come and go, all defeated IMO.

But there's always a first!

Mani.

[PeterSt]

At last ...
Btw, here's the one and only official link, for others in other countries : http://support.microsoft.com/kb/953228/en-us

Now to the point again;
Since this is a patch to a misbehaving system, which includes a new version of some usb**.sys file, you won't be knowing whether it works without this file. It may, but it just as well may not, and therefore you have to proove it (somehow).
The *fact* that there's exactly no real reference to this "KB" anywhere, tells me (theoretically !) that it doesn't exist. The link I mentioned though, *is* real, but it isn't for any other country or language. Notice that all the others (650) are stupid auto translations, and the "standard" link you provided will in no way lead to the english text. This tells me not all, but something.
For me this goes as far as that this is some leftover from MS and possibly some "accident" in the first place. Btw, the patch can be downloaded just the same, but you won't be able to tell to what (OS version) it should be applied (because the data given is not consistent regarding this).
We could think that by now it is an official thing (from off Vista-SP2 or something but this is the inconsistent data, or otherwise at least for W7 RTM and W2008 R2.
Sadly (for you) the Registry Attribute is not there by default, while it defaults to 10ms, which is what you set it to, and now you have better sound.
GREAT.
But if you don't mind, I won't believe that. You should at least have set it to *other* than 10ms to have things in your favour. And then still we won't know.

Of course we now are believed to think that the default has changed since whenever this came about (2007 I think), and that is why it works for you. But even more "of course" I won't believe any of such a thing, especially because you are not the first to come up with something like this, and the most often people just don't know what it is about, but read something somewhere. This is similar to what you call a buffer length in time, while it is not at all (yes I know, you see ms).

Here you have a few other links to play with. Can't find the registry entries but somewhere they exist (in 1000 other links ). It's about a similar thing and most probably the registry key is the same, but the attribute name is different (if memory serves me well). Just apply it to your USB (DAC) device ...

http://www.soldierx.com/tutorials/USB-ControllerAdapter-polling-rate-tweaking
http://forums.epicgames.com/showthread.php?t=581256
http://www.ngohq.com/news/15043-how-to-increase-usb-sample-rate-in-windows-vista-7-a.html

All I want to say with these links at hand : there is more that acts upon the same thing, and while we won't even know which has prevalence, we also don't know whether things could be made inconsistent because of this. And THAT is what I am about.
What next will happen is that 80% of people have audio stuttering or whatever at a next XX version, while 0% of those people will come up with a "hey, I applied that link from Blaine !". No no, that is for me to find out.

Do you now get my message ?
Nobody won, but you lost. Almost.

Still ...
Still it was smart of you to find this KB, because I sure don't see how ever you could get there. Also, the "tweak" itself most clearly is totally unknown (ok, not anymore  ). So, it sure might do something, and it is to be investigated at some stage. But this is very different than just applying it, without knowing what it will cause to others. I know, you may not care much, but as sufficiently said / implied by now, I do. I have a player to support and I don't want users to end up with "stuttering" all over, just because something by accident worked for you. That is why I gave that WASAPI example right away, but apparently it needed quite some more posts to make that clear to you. Well, maybe it doesn't still.

Also notice (for some mindsetting perhaps) that actually no such thing exists as a "device buffer" which would be able to change for USB devices. The USB devices themselves have these buffers, BUT in the OS these buffers exist just the same. And oh, for XP it is "1ms" (which again is no ms at all).
This link of yours may be able to change that, and back at the time I would have been the most happy with that. But since I didn't know about this link, I solved it differently.
Hey, what did I just say ?
Ah, oh. Now what.

So, good link, and you didn't loose either just because of that. Nobody wins, nobody looses, just don't get lost in counterproductive tweaks you won't recall yourself when things stop working.

Eh, thanks ?

[blaine78]

Sadly (for you) the Registry Attribute is not there by default, while it defaults to 10ms, which is what you set it to, and now you have better sound.
GREAT.
But if you don't mind, I won't believe that. You should at least have set it to *other* than 10ms to have things in your favour. And then still we won't know.

Pete, please read carefully what i posted from MS site.
The default for XP is 10ms and 4ms for Vista. Unknown for Win 7. you ,yourself are guessing. i don't use XP, i use win 7. it is only defaulted to 10ms after installing a patch, which i have not done and this patch is for vista not win7.

[blaine78]

http://www.soldierx.com/tutorials/USB-ControllerAdapter-polling-rate-tweaking
http://forums.epicgames.com/showthread.php?t=581256
http://www.ngohq.com/news/15043-how-to-increase-usb-sample-rate-in-windows-vista-7-a.html

Thanks, i've already seen these links before, and didn't work. USB IdleEnable that another posted on here does work.

[PeterSt]

you ,yourself are guessing.

Haha, careful now. You are the guessing one, and I *know*.

Default (actually fixed as far as I know or knew) for XP is 1ms (which is not ms), for Vista I forgot whether I ever tested that, and for W7 it is infinity. I am GUESSING that for Vista it's infinity too.

Remember, I can see, measure, try, write code to try, write the audio software anyway.
So who is guessing ? not me.

Maybe your means of interpretation are different from mine, but really nothing like entering a new Registry entry will apply a "patch". This is to be read as : without applying that patch you can add Registry entries you want, but nothing will happen. So, it needs the patch first.
Next, the patch disappears, and instead it becomes a standard for the next OS version. This will include W7, unless it's taken out again anyway. Now which way do you want to go ? ... Taken out ? then it does nothing. Still in there ? then it defaults to 10ms and can next (and only now) be overwritten.
If you want to read this differently, all stops here.

What it (once again) all leads to, is that you won't know what you are doing; not because your are dumn or anything, but because you can't know what the OS does for - or to you. Remember, I *know* how it is, because I check things.

Now think further ... this beautiful tweak of yours could change your infinity into 10ms. So, you applied a change indeed. Sadly, XXHighEnd can (and will) work with this infinity figure. Now what ? You molested things.
I am not saying this *is* so, but chances are fair, if only your "tweak" still works.

Shall we now cut it out ? apply it to JR and be happy.

I will tell you one last thing, to maybe show you at last that you don't know what you are doing ("to");
Try to find latency figures from MS. This is how fast a response from buffers etc. can be, and what the best response for e.g. MIDI stuff can be.
What will you find ? 10ms.
You already know by now this actually can be 0.5ms, right ?
Or do you maybe accept that it is really (but now REALLY because *I* say so) is 1 stereo sample of 32/384 ? take your calculator. And not only because I say so, but because people use it (to be honest at 192KHz).
You want to project it onto external latency ? also fine. Multiply what you found by 48. Or by 32 for those with (RME) soundcards which can do that.
Remember, 32/384, may you get around what that actually says (referring again to your "1ms" which is not about time at all).

Now tell me again what you'd want with your xxms which I overrule beyond all MS articles.
Yeah, that you can molest your own PC with it, because I'm really not in control of everything (especially the "tweaks" I don't know about).

I really hope this is enough of it !

[blaine78]

you ,yourself are guessing.

Haha, careful now. You are the guessing one, and I *know*.

Default (actually fixed as far as I know or knew) for XP is 1ms (which is not ms), for Vista I forgot whether I ever tested that, and for W7 it is infinity. I am GUESSING that for Vista it's infinity too.

Remember, I can see, measure, try, write code to try, write the audio software anyway.
So who is guessing ? not me.

Maybe your means of interpretation are different from mine, but really nothing like entering a new Registry entry will apply a "patch". This is to be read as : without applying that patch you can add Registry entries you want, but nothing will happen. So, it needs the patch first.
Next, the patch disappears, and instead it becomes a standard for the next OS version. This will include W7, unless it's taken out again anyway. Now which way do you want to go ? ... Taken out ? then it does nothing. Still in there ? then it defaults to 10ms and can next (and only now) be overwritten.
If you want to read this differently, all stops here.

What it (once again) all leads to, is that you won't know what you are doing; not because your are dumn or anything, but because you can't know what the OS does for - or to you. Remember, I *know* how it is, because I check things.

Now think further ... this beautiful tweak of yours could change your infinity into 10ms. So, you applied a change indeed. Sadly, XXHighEnd can (and will) work with this infinity figure. Now what ? You molested things.
I am not saying this *is* so, but chances are fair, if only your "tweak" still works.

Shall we now cut it out ? apply it to JR and be happy.

I will tell you one last thing, to maybe show you at last that you don't know what you are doing ("to");
Try to find latency figures from MS. This is how fast a response from buffers etc. can be, and what the best response for e.g. MIDI stuff can be.
What will you find ? 10ms.
You already know by now this actually can be 0.5ms, right ?
Or do you maybe accept that it is really (but now REALLY because *I* say so) is 1 stereo sample of 32/384 ? take your calculator. And not only because I say so, but because people use it (to be honest at 192KHz).
You want to project it onto external latency ? also fine. Multiply what you found by 48. Or by 32 for those with (RME) soundcards which can do that.
Remember, 32/384, may you get around what that actually says (referring again to your "1ms" which is not about time at all).

Now tell me again what you'd want with your xxms which I overrule beyond all MS articles.
Yeah, that you can molest your own PC with it, because I'm really not in control of everything (especially the "tweaks" I don't know about).

I really hope this is enough of it !

All that matters Pete, is that it sounds better.