0.9z-7-4: Unattended mode behaving like AB

[charliemb]

Here are the two prior.

Otherwise, did you answer my question about the proper keystrokes or mouse clicks for obtaining these log files?   Because I am running unattended, I have to hit alt-x and alt-s.  Does the order matter?  Then do I quit xxHE?   Obviously, I didn't do it right or I wouldn't have to be redoing it now.

[charliemb]

The previous was the repost you asked for.

This is for a new instance from tonight.  The song is track "07 What a little more rain" by artist Sara K, but actually the disc was Chesky's Greatest Audiophile Recordings or similar (not labeled as Sara K).

[PeterSt]

Otherwise, did you answer my question about the proper keystrokes or mouse clicks for obtaining these log files?   Because I am running unattended, I have to hit alt-x and alt-s.  Does the order matter?

Ah, I'm sorry. I thought I answered the question, but I guess I needed to elaborate as well.

Any shortcut command applied (when running in unattended) creates a log file. So, Alt-x does. But also Alt-s does. These are two log files *after* the real fact, because the one we'd need is the one which comes from the moment you started playing. So, that XXHighEnd instance (just visible on-screen) carries a log file as well. Thus, with Alt-s and Alt-x (or the other way around) two log files emerge we don't need. However :

When using Alt-x and press Stop, only one log file we don't need emerges.

This counts for the XX log files. The X3 emerges per "play" session and the X3PB emerges per Sound Engine start (normally one per boot session when you never kill it and it never fails).

Clear now ?

[PeterSt]

Again funny;
So, day before yesterday you coincidentally played that Beck two times which was the most helpful, and now you added an addtional XX log file to be on the safe side, which ... explained what I saw in the last log file.
You communicate well ...

So, I now sure found something. I don't know how it happens yet or what you may do to incur for it, but at least I should be able to prevent it.

Great work Charlie.
Peter

[charliemb]

Ahhhh,  this is great news.  Some progress somewhere.  Thanks.

To summarize the best procedure for gathering log files, it's best to hit alt-x and then click on Stop (not alt-s), and then send ONE extra xx file.

One last point I'd like to make is that in these examples the files were hiRes;  but it has also happened on 16/44.1 material, which I admit is strange.  Strange because if it is a miscalculation based on assuming 44.1/16 when it is really 24/88 or 24/96,  then what is being assumed when the source material is 16/44?   ..mp3?

[PeterSt]

Charlie,

It can happen it all kinds of directions and can happen to all. It depends on whatyou played previously (for a "Play" session).

MP3 is just 16/44.1.

To summarize the best procedure for gathering log files, it's best to hit alt-x and then click on Stop (not alt-s), and then send ONE extra xx file.

Correct. But generally we'd need the larger log file; the one where you see things really happening (just look into them and you will understand).

Peter

[PeterSt]

All right.
So below you'll find a new 0.9z-8-3e to hopefully - and finally - solve it all. Paste over your formal 0.9z-8-3a XX folder.

The whole thing is so complex that possibly you'll receive messages telling you that playback is not possible, slomo or highspeed sound comes from it, or other things I could not test in this short time - but it should work.

With the motto "an accident never comes alone" I found a stack of problems that can (or will) occur once Hires material is played, and which not many (in this commnuty) do. This includes myself, and for that reason I never could notice - as nobody reported anything of this kind. But things sure were wrong.

So, this should help Charlie. Finally ...
Peter

[charliemb]

Fantastic!  Thanks.      I'll try it right now  (keeping the volume low at first and keeping my finger near the DAC's mute button!).

[charliemb]

So far so good.  Three hours tonight, with mixture of resolutions, and ending with all hiRes for about 1hr.  That's not definitive, but it's a good start.   Keeping fingers crossed.

The only anomaly I encountered you already know about, and that is that alt-e, followed sometime later by an alt-p, did not continue from the same song, and instead started from the top of the playlist.  But that's another problem and minor in comparison.

Also not new with this version are some messages I've been getting since about z83a ...everytime I hit alt-c from Unattended the pointers set up correctly in Nice Stuff, but I get two messages.  One is something like "91: Object could not be set to instance of an object: c:xdata,"  and a second one about c:xdata but I don't remember.   Perhaps this is due to logging?   Still alt-c works.

In any case, thanks for this big step forward!

[PeterSt]

Ha, great feedback that is (two-folded).

I will look into alt-c. I never use that (only alt-h (whole album to Nice Stuff) and alt-z (one track to Demo)), but things sure have changed here, so I probably made a mistake somewhere. Thanks for mentioning.

Maybe alt-e needs a new topic for further workout; I use that regularly and don't see what you said at all. In that topic, start with mentioning whether this is with selected tracks (ctrl-click etc. in the Playlist).

Peter

[charliemb]

I also have the x3 and x3pb if you need them.

No problem with song length, at least apparently.  And not so fun anymore.  This time the problem is big loud gaussian noise, not music, going to only or mostly the left channel.

EXPLANATION
After randomly selecting HiRez tracks using the randomize function (first then n) and after playing for a while, say one hour, it was too many Sam Cooke tracks in a row and so I hit alt-x while track 6 - "Just for you" was playing and also coming close to the end of that track,  xxHE came up as expected, than while transitioning to track 9 - "Ill come Running back to you", instead of that track I got very loud / ear damaging / and if for a long time perhaps tweeter damaging gaussian or thermal noise like RANDOM fullscale data going out through the left channel.

Instantly I hit mute on DAC first thing, even before I clicked on Stop in xxHE; then I saved the log files;  then I told the OS to restart.  And while shutting down, I got the blue screen of death, luckily while DAC was muted.  The BSOD probably happened as the OS sent audio data out to the DAC.

Not fun.  We are into potential ear and tweeter damaging territory and so please check the code carefully and perhaps do some testing of your own.  I won't be able to test this all alone...  as these are new and expensive speakers we're talking about (and not to mention irreplaceable ears).

Later in the night, very late, I turned the volume down on the DAC, way down, and attempted to repeat the problem and it did not repeat.  Perhaps I entered alt-x too soon in the song and should have waited for the last minute or 30 sec (perhaps while the next track is loading is where the problem can be repeated).   ...or this could have to do with "vista can't do SFS=60" and I have it set to 60 per your last request.

[PeterSt]

Charlie,

Let's first say that it hardly can be a coincidence, after people for years now (really) not running into such an issue that I know of. Still it can be, and yet I think it is. But, this is NOT to turn away from any issue, so, just saying ...

In this case no transistion from one format to the other was in order. The log files nicely show all now, and I don't see anything wrong. But indeed, I now may need the other two log files to see whether now there something is odd. So, that's first I think.

The transition from one "about" track to the other at using Alt-x also won't do anything. It may when using Alt-u/d but only because I know that special things happen in that area, but it should all be covered for. Plus that you didn't do that of course.

Far more logical would be the sheer fact that you hammer on the system with bringing up XXHighEnd (which is quite heavy - see the time it takes) while at the same time a next track has to be read from disk. If *then* things go too far something like you experienced might happen. Btw, something like that happened to me with W8 and I considered it no coincidence, that OS dealing with all different than W7. Same for Vista, hence W7 compared to it. Still it happened once with W8 in a whole month, but it never happened for 3 years with W7.

One other indication of the OS possibly going wrong is this :
When I'd put out the most wrong byte order, the OS doesn't know about this. It just pushes through the bytes and it doesn't know about audio, nor does it know how the bytes should align. However, internally all has to stay in-line (in sync) or otherwise things get messay (like what you heard). Now, the fact that it came through one speaker already tells something (although I think I could let that happen too but must think about how really), but the mere fact that your OS BSODd (which is not normal for your system I assume) tells a lot more. Btw, notice that when I say "OS" it will merely be the driver, but this is part of the OS at the time.

Again, not to turn away from the issue ...
So please post those log files.

Peter


PS: As usual, a great recap of what happened. This is the most useful/important ...

[charliemb]

Here are the remainder of the recent log files from that session / date code.

And to confirm, no my system never BSODs.  The last time was when you introduced Q1factor and I tried Q1factor=2 and got a huge crack followed immediately be the BSOD.  This was for sure with SFS=60 on Vista. 

Thanks so much for being attentive to this.  Much appreciated.  We xxHE users are in a bind: Other players we've tried are unplayable / frustrating in comparison, and so if xxHE becomes frustrating for any reason, then there is only frustration left.

[PeterSt]

Can it be so that somehow you missed a track that was about to play ? So, all what I see is that your mentioned (to come up) "I'll come running back to you" was the one lastly playing, to be followed by "Lucky Bounce" from another album, and that one is 24/96 ...

Assumed this is correct (hard to imagine with your detailed outlay) all I see is a buffer error (but this is fairly normal UNLESS something can't cope with it because it *is* an error) and shortly (some 30 seconds) after that you pressed Stop.

Assumed I am not right at all (also hard to imagine) try to concentrate on 24/88.2 running into 24/96 and notice it is possible that your DAC just can't cope with this. But then it will be 100% repeatable I'd say.
Anyway, be careful ...

Regards,
Peter


PS: Since all is fairly explicitly shown in the log file now, I can tell that all really looks okay. Thus, also that 24/96.

PPS: I sent you an email as well.

[PeterSt]

Question : What causes you Clock Resolution to be set at 1ms under the hood ? (you switched off the forthcoming of the message about this)