Thank you very much Dave. It is the most convenient to have at least two voices in the same direction (even if one of them is less trustworthy me ).
So you were already used to it, but I think last night I received the best playback in my life. Imagine, in the end I was listening to russian choires.

Gerard, assuming that I have some authority left (which is not likely ), this is what is happening in the base :

- The OS being on the HDD will keep it spinning forever (as long as we can't get rid of those few I/O's it does once per 6-12 minutes);
- The spinning of a disk influences sound (via the PSU);
- At least when you just start playback of a track, the disk providing the music data (which will not be the OS disk) spins just the same. You could say "so, that can't be avoided, why bother". True, but with the OS disk that's two disks spinning now.
- Personally I think there is more to it, meaning that the OS bening on the HDD incurs for more "bad SQ" apart from disks spinning. I mean, the difference in SQ when the OS is on the SSD is just too large to blame "a" (!!) spinning disk.

Then :
- When you play a FLAC (or any file which must be converted) it *is* played from the OS disk (better : from the disk where XX resides).
- IMO this can hardly be related because the only thing what happens is that the converted file is read - done. Of course, with the HDD this incurs for spinning the disk ...

Lastly :
I have already in mind to provide an explicit "copy to the OS disk" of the files to play (as implied by the Playlist Selection), just to allow for what you just asked (copy temporarily). Why ? this will spinup the music containing disks only at the start of playback (mind you Unattended, otherwise the lot won't work), and the explicit copy to the OS disk (conversion needed or not) will from then on let stand still all the disks in the system. Only with the SSD (and the OS being on that) of course.
The explicit copy will hardly be noticeable, just as the conversion including such a copy is not (0.9w and up only !).

Peter

Oh sure , but he already received my love !! Have to share that with my wife now. But here's another :
hehe

But it is so true how important it is that at least ONE person pays attention here !
Just look how things went overhere :

1. In order to split the large files (a longer (very legitimate) pending request from JohanZ), half of XXEngine3 had to be turned upside down.
2. Because I had to test that, from then on I started playing with 12MB file parts. I heard clicks between the parts though (never knowing when the parts exactly split *and* I hear clicks since my preamp was removed, the unity gain buffer being there instead (so, without the buffer I never heard clicks, and the clicks should be there for everybody in between tracks. Not that I ever received such a message (since that was solved many months ago)).
3. I kept on listening to the small file parts, and it was my idea that *that* caused the more "lean" sound. Lean for me : better fluent, and not exactly tinny (as how I think Joerg meant it).
4. Although I didn't like the sound much, I kept on listening to the small parts and never went back, and again kept on thinking it would be the small parts (with the explanation on memory boundaries as I expressed elsewhere).
5. Then the SSD came in; As I explain it now, that change emphasizes "accuracy". The sound now was the most bad. Indeed the mids (the base sound of cymbals) got over-expressed, and things started hissing all over, with the most bad signature of all cymbals starting to sound the same. The harmonics (singing of cymbals) disappeared (or got overwhelmed by the mid freuqencies), standing waves in the mid area (voices) started to buzz, and lastly the bass output just went wrong (too much of the wrong -> not tight and standing waves in that area again).
6. I went back to the spinning disk, and all seemed to be okay again. But mind you, all is relative and I still didn't like the sound much (the small file parts still being active).
7. Then LydMekk posted.

The rest is known.
It is exactly as Calibrator said : When the one stupid LydMekk had not been around, maybe nobody had noticed, and by the time I would have switched off the small file parts I would have been sad for not the best sound still, BUT THEN I WOULD NOT HAVE KNOWN WHERE TO START LOOKING. Remember, there are no switches "good - better - best sound" in there. I wish it were that easy ...

Peter


PS: You read well : In the earlier versions (like 0.9v-6a) there are very tiny clicks (one "vinyl tick") between tracks. I wonder whether anybody can hear them.

Interesting ...

And besides that, Johan, it is remarkable that you gave it the time to find this out. I too used the Mem checkbox but merely to find out whether it could improve SQ on 0.9w (u/i -3) and I imagined it did a bit. But it didn't occur to me that it removed the (btw very small for me) clicks.

And I must say, although I know what I changed, it has been a work of far more than 100 hours to get where I wanted to get for the end result, which doesn't mean I have to spend another 100 hours to do it again but right. However, at this moment I still don't have a clue on what is actually wrong (sound wise that is), but now knowing the end result I will be able to let things work similarly to 0.9v7 ... I hope.
The knowledge of this Mem box thing is valuable to this, and although you may think this is related to gapless only, to me it says more.

Thanks,
Peter

Am I allowed to give this topic a twist ?

I know, it has only be one evening of listening, but I dare say : go for the SSD.
I, and maybe more, have wasting my time on the 0.9w (-1, -2, -3) versions because it was really that doing it (wrongly) ...

Now, at using 0.9v-6a, it turns out that using the SSD for the OS disk just works out for the better. It incurs for more accuracy, and whatever the base of playback is, it is more accurately played.
Note : I used 0.9v-6a because I didn't know there was a 0.9v-7 as well (how could I know ), and I have no reasons to believe 0.9v-7 doesn't work out the same (but has many useful functional enhancements opposed to 0.9v-6a).

0.9w (-1, -2, -3) has something wrong in it, and the SSD indeed emphasizes it. Ah, the SSD just emphasizes. And luckily now things are consistent again, because I seemed to hear the SSD being "better" right from the beginning (read back this topic) ... better at reproducing what it's fed with (hahaha don't think about this latter, or your mind will get into a knot).

I am sorry for all the confusement, and I can only say that sometimes things get really difficult.
One thing I like the very best, and that is that in here we are not just copying each other's words and feelings and hence won't get placebo'd by that.
Keep that up.

Peter

LydMekk, Joerg too : 


All,

I can't emphasize more on please COMPLAIN when you have the idea that something is wrong soundwise. You for sure won't embarrass me with it, and in the end it will help yourself.
I have only two ears and two speakers, and for sure won't claim they are the absolute best.

Usually I know when something can be changed for SQ and then I announce it. Like here, from 0.9w-1 :


be extra-alert in such a case. Even if you are not sure, please share.

Peter

Hahaha Dave, you just pointed out to me there's a 0.9v7 as well.
Have been using 0.9-v6a tonight and hated it for lacking functionality (wow, one gets used to things quickly !).

I can safely tell you that all the changes that can influence SQ are not in 0.9v7.

Dave, please don't bother. No matter how I hate it to be true, 0.9w is just no good.
Don't waste your time on it.

Right now I tend to believe playing with the SSD as OS disk has better SQ than I ever experienced. With 0.9v6a that is ...

Hi Johan,

Although I do not recognize anything of your "seconds" gaps, I hear clicks for weeks now on the boundaries of loaded pieces of tracks. For me this came totegther with removing the pre-amp (all being more audible because of that ) and most probably the new memory management which for me came earlier than "official" (as per 0.9w-1) but anyway is in my mind a being related and ever since wating for someone coming up with it. Ok, your "seconds" are not exactly what I meant (I really don't know how *that* happens) but it was to be kind of expected by me anyway.

Just throw 0.9w (u/i -3) in the trash bin because by now I'm sure it s*cks all over.

No, but that is not what I meant. It is merely the interrupts towards the cpu (otherwise Core Appointment would be without reason anyway -> nobody plays music with a "stressed" cpu because of other tasks). And it *is* audible (Appointment) so it does matter "somewhere".

I think there is not much experience with the quad core, and the only thing I can say is that it is more prone to problems (software wise), that's all. XXHighEnd only can benefit from it because of the faster conversions (which are rather important of course).

Peter

No, nothing special. Theoretically a process internally may not "want" what XX directs and that (again theoretically) may crash the OS. But since I never heard of it before (and I certainly would if it happened regularly, I'm sure) it is not likely this is the case.
How it could destory the mobo is certainly beyond my comprehension, and it will be the other way around (the mobo is not okay, and makes the OS crash -> note that when appointing more tasks to one core than normal (which is what would happen by means of using XX) certain parts will be stressed more than usual, although the only parts I can imagine is the cpu itself -> did you overclock ?!).

Ok, wanted to come back on this one.

Read the article, and all what is in there looks completely, say, logical to me (with some extended knowledge of "performance" issues on computing), and regarding to this, these are my observations :

Yes, there seems to be some "interaction" going on between some backend (the SSD) response, and general things you are doing on the PC (those general things for me would be the development of XX which happens in Vistual Studio). The "response" behaviour (take that literal now) is a little different. Small stalls are there which weren't there before.

During just using XX the comparison is somewhat more difficult. At using the spinning disk

- All is slower in general (note this is about accessing the Galleries only, and of course the OS itself needing the data on the disk).
- It rather frequently happens that I have to wait the standard 6-7 seconds before anything happens, because the disk has to spinup (it stops after 5 minutes of inactivity which latter is rather random (see the tweaked my Vista virtually to dead topic).
- Behaviour is consistent.

At using the SSD :

- No signature of waiting for the disk to "wake up" ever.
- No response at certain tasks, but, never bothering because you're not waiting for it.

The last one is what is described in the thread Telstar referred to, hence I'm sure that is happening. Thus :
At converting (mostly FLACs) at a certain stage a buffer seems to run full, but the way 0.9w works causes you won't notice by means of the sound already playing (this would be different for 0.9v for sure because there all the conversions need to be complete before playback starts).
The only way you notice this is at the XXHighEnd form staying there, all disk activity stops, but also cpu activity stops. So, it stalls. Sound just continues because of the high priority which is put to that, and again, it actually doesn't bother.

Net, I prefer the SSD a 100 times.

I have been thinking whether this could influence sound quality, but keep in mind that "this" would be the stalling "process", and I do not recognize at all that such a thing may happen, or is encouraged for during playback, when - or everything is already in memory - or only disk reads are in order (no large bulk writes). So from that view it can't influence.

Note that it can be proven that when a track of e.g. 10 minutes is loaded into memory, just zero I/O applies, excpect for that 3 times (very small) happening once per 6 - 12 minutes (tweaked to dead thread).

For the spinning disk those I/O's prove to be not important to anything, because when the spinning disk stopped spinning, nothing is bothered by it, and the 7 seconds it takes to spinup the disk doesn't stall anything really (all is just buffered, and no to the user visible process stalls or holds up anything for that user).

What happens when the SSD (controler) can't deal with -I assume- writes i hard to tell. Indeed it will be the controler starting to hold up things because the disk tells "wait a minute please" (the minute being literal almost as per the thread Telstar referred to ) and where the disk will be able to have dealt with writing in a few ms the controler thinks differently about this and just stops longer.
It is here where the dangerous part emerges, because from now on things are dealt with by means of interrupts. Interrupts are dangerous because they can emerge 1000ths of times per second, do not eat cpu,but stall the complete system. And, when something is not communicating properly, even the disk telling it is ready (which would be an interrupt by itself) won't come through because of the stream of interrupts the controller is generating. Note that this stream could be "questions" towards the disk which should happen per e.g. 1 ms, but instead it happens each us occupying the bus concerned, the answer (from the disk) not even coming through.

Remember, it was proven (Telstar thread) that something of this kind happens at burst writes.
These burst writes just don't happen during playback, nor does the OS incur for that (proven by myself), and all 'n all it is my conclusion that this is not related (to sound quality).
Note though that one of the first things I said is that it felt like each memory cell is talking to something (like the controller), which would be a very similar activity as the interrupts I just described. But I think this is a coincidence ...


I could keep on talking about what could be the matter, but right now it would be my idea indeed that the SSD incurs for more accuracy because of just more stable (or hardly noticeable) requirements from the PSU opposed to the spinning disk. Actually no different from how anything works out within the PC for PSU matters. This would come down to better emphasation of the base sound i.e. the stream as spit out by XX, and that stream is wrong. It is wrong (or not as nice) opposed to an earlier version, and it *is* true that the change in sound I noticed from the v to the w verions has a nature that seems to be emphasized with the SSD. So, stop talking and back to 0.9-v6 including going back to the SSD (which I will do in one go).

In fact I should try 0.9-v6 with the SSD. Uh-ooh, now I'm getting really confused :

As said, to me it feels the SSD is better, but it doesn't workout. Now what if it emphasizes the baddyness of 0.9w ? (emphasizes = more accurate). THE NATURE IS QUITE THE SAME ... I must admit that.

Oh boy.

Edit :

Did that ... (8ch).





(PS : Yes Dave, you suggested that it could be the XX version. Of course I didn't try. ).

Yeah, I recall you were a "double" fan ? so at last got rid of that ? hahaha


Btw, the fact that I know what changed, does not imply it will be easy to restore that in the current version. Quite impossible.
Challenge challenge challenge.

I think at 0.9w-1 I anounced that the sound has become more "lean" to the sense of it being more easy to flow out of the speakers.
So yes, something has changed. But it was unintentional (I know where/what it is though).

It is true that since then I am kind of struggeling myself whether it is better or not, because it seems to have aspects that are better for sure (to my ears).

You do have the "Split file at size" at the default of 2500, right ?

I will reinstall 0.9v6 and listen for a good while, but I think I already agree ...