0.9y-5-05

[manisandher]

Peter, I get a 'YES' all the way down to Q1=-4. However, I get distortion (the exact same that I described before) at Q1=-2. I therefore have Q1 set to -1.

(I started at Q1=+14, then +5, 0, -1, -2, -3 and -4. Should I have come down from +14 in single increments?)

Mani.

[PeterSt]

Actually, that you won't get the special mode is logic for you, because you could go so low in the first place.
Maybe you can incur for it at highering your soundcard's buffer ?

Anyway, don't make too much fuzz about it right now. First thing is that all keeps on playing (at any Split File size btw).

Peter

[manisandher]

There seems to be an inconsistency:

Highest buffer setting - exactly when the distortion sets in, I stop getting a 'YES' (Q1=+2).

Lowest buffer setting - when the distortion sets in, I still get a 'YES' (Q1=-2).

I thought perhaps it wasn't working right. But if it's not important, that's

Mani.

PS. Still pleased with Engine#4

[Eric]

Hello Peter,
I followed your instructions. All tracks play without a problem.

In my case, when Q1= 6, I am getting a YES and normal sound. But if I go 1 step lower to Q1=5, the YES is gone and the sound gets very much distorted.
What is the point you are trying to make here?

Cheers,
Eric.

[PeterSt]

That this doesn't work for you ?

Actually it is to weird to work, but here is does.
Never mind if it doesn't. Ok ?

[Eric]

Okay.
By the way, I just played a bit with the FireFace400 buffersize.
Setting it to the max of 1024 samples, I could go as low as Q1=3 (and a YES). Lower than that, distortion occurs.
Cheers,
Eric.

[DannyD]

For me Q15 to Q14 is the transition point.  I am using a Lynx AES16 and a BADA with no oversampling.  I like "abnormal" mode, and as a result I finally like Engine 4.  I used to find the sound lacked stability, especially with vocals.  Now it has the sonic stability of Engine 3 with the frequency extension and finesse of the previous Engine 4. I think I even like QAP now, but I need to try a broader range of music before I draw any conclusions.  Good stuff!

[DannyD]

I find the transition point between normal and not normal is different when XXHE is set to no oversampling and to QAP. 

With no oversampling it is consistently between Q1=14 and Q1=15. 

With QAP it is variable.  I played the same song multiple times at Q1=6 and the first time it was Not Normal (blank), the second time was Normal (Yes), and the third time it was also Normal.  At Q1=5, the results were No, No, Yes.

I guess I'll stick with no oversampllng.

[Gerard]

The other q settings influence the Q1 as well.

(ps does not work here)

 

[Eric]

Peter,
re. Engine#4.
Although all tracks are now being played, the transition between the tracks is audible like a short and little plop sound. It is very subtile, but it is there. I am talking only about albums where there is no space between the tracks, e.g. live concerts or classical albums.

With Engine#3 this doesn't happen, the transition between tracks is smooth.

I am sure you will be able to determine what causes it and fix it.
Cheers,
Eric.

[manisandher]

I am using a Lynx AES16 and a BADA with no oversampling.

Hi Dan,

When you talk about 'oversampling', are you refering to what BAD call 'interpolation'?

Mani.

[DannyD]

Sorry, Mani.  I am referring to XXHE's upsampling feature, as in QAP.  When I have it set to No Upsampling the breaking point between between Normal and Not Normal is consistently Q1=14/15.  When I use Arc Prediction at Quad setting, the breaking point is inconsistent. 

Dan

[DannyD]

Peter,

Thanks for the name change.  No longer "not normal", just special.  I think there's a lot to like in the new setting, even though you now call it wrong.  Are you going to tell us what's going on under the covers?

Dan

[manisandher]

I am referring to XXHE's upsampling feature, as in QAP.

OK, I've got it now!

Peter, I'm having problems with 0.9y-5-06 (and 5-05). I've got gaps appearing intermittently in the music. Also, Unattended seems less 'responsive' than before. It takes ages for the wallpaper to come up and for the Alt-X command to work.

None of this happens with 5-04. But the 5-04 log file is interesting. I seem to be running out of buffer just as the music starts - there is a spike in latency exactly at this point also. But nothing is really audible - there may be a tiny, tiny hiccup heard sometimes, but nothing to worry about. Once the music has started, it continues without any gaps. I still get the odd very, very faint crack or pop coming through from time to time though...

On a final note, there is a big difference in CPU usage between Q1=0 and Q1=-1. The former is quite constant, the latter is full of periodic peaks and troughs...

Mani.

[Calibrator]

On a final note, there is a big difference in CPU usage between Q1=0 and Q1=-1. The former is quite constant, the latter is full of periodic peaks and troughs...

I've found that to be true also with 0.9y-5-06 ( didn't get around to trying 5-05 ). In fact any Q1 setting below +2 effectively removes any meaningful control of the system until the queued tracks have completed. Mouse and keyboard are dead weights. Lowering Thread Priority helps a little but then I get a lot of stutters and distortion when it first starts playing the queue. After the initial processing stage things then sound OK.

Although 5-06 nows plays seamlessly from one track to the next, an Alt-N while in UNattended mode will bring forth an "XXEngine3.exe has stopped working" message.

Will continue playing.

Cheers,

Russ

[PeterSt]

Yeah, I now realize that the way it is setup sucks all the life out of everything else. I forgot about that because I lowered the priority - for the same reason and never saw it anymore. I will try to let it work differently.

Russ, the Alt-n is known to me, and I guess I should have solved that earlier.

Peter

[Leo]

With the Cantatis card I get the Special mode working at Q1 = 4. Below that a horrible distortion. I like the SQ a lot in the special mode. So much in fact that I had to pacify my neighbours with a bottle of wine after playing too loud too long

[PeterSt]

Nice ...

[PeterSt]

On a final note, there is a big difference in CPU usage between Q1=0 and Q1=-1. The former is quite constant, the latter is full of periodic peaks and troughs...

Mani, would you care to show the cpu graph for these both settings ? 5-06 only please.
Thank you,
Peter

[manisandher]

On a final note, there is a big difference in CPU usage between Q1=0 and Q1=-1. The former is quite constant, the latter is full of periodic peaks and troughs...

Mani, would you care to show the cpu graph for these both settings ? 5-06 only please.
Thank you,
Peter

I was still talking about 0.9y-5-04, which is the only stable version of KS on my system. Sorry for any confusion.

But FWIW, here are the CPU graphs for both 5-04 and 5-06. On '506 Q1_-1 with gap', you can see two peaks in CPU usage around the middle of a track. The first is a periodic peak which doesn't interrupt playback. The second peak does interrupt playback for a 2-3 seconds, after which time playback resumes as normal until the peak happens again a few minutes later.

[EDIT: All graphs except '506 Q1_-1 with gap' are taken at the very beginning of the music - do not include pre-processing.]

Apart from a peak at the very beginning of playback, latency (not shown) remains totally flat at the 50μs level throughout.

Mani.

[manisandher]

Please ignore any references to gaps in playback in the above post. This has been easily solved by changing my firewire driver.

0.9y-5-06 now works perfectly down to Q1=-1. For Q1=-2 to -4, I get still get 'normal mode' showing in the log file... but it sounds anything but normal - distortion plus half-speed effect. No 'special mode' for me

Mani.

[Telstar]

No special mode for me either