What can it be ?

[guytou1]

I am trying to use a newly arrived M2Tech hiFace on my XP desktop.
Sadly, I have the following messages (see pictures)
I have no problem to play this 96 kHz 24 bit file with 09y-4 and engine #2.

Best regards.

[PeterSt]

Just for starters ... do you have a .wav file laying around ? try that with the same settings and tell me what happens ...

Strange though, assumed this only happens with the HiFace ... And if indeed so, what worked ?

Thanks,
Peter

[guytou1]

Peter,

It was a .wav file.
I tried later a 44 kHz 16 bit .wav file.    The first time I had a few seconds of sound, and then the message "Too many buffer errors".
I tried then to play it again. This time I had the message in picture 1.  Since that, it doesn't work any more with foobar (picture 2). OK I think if I restart my computer.

Thanks

[boleary]

The first day I tried the Hiface I repeatedly got the "Device allocated but cannot Play" error with everything I tried to do. Initially, I had to restart the computer every time to get xx to work again. The next day I switched from scheme 3 to scheme 1 and everything worked. Scheme three works for Peter and Mani (See:  http://www.phasure.com/index.php?topic=1161.0) so I would definitely try changing the core appointments to see if you can find one that works for you.

[guytou1]

Thanks Boleary.

I found some thing else on my one core computer.
I put usualy XX 09-y4 on a RamDisc drive. Do I have a problem of memory space with 09-y8?  Only 50 MB are free.
When I put the XX 09-y8 folder on my HDD it is OK.  I tried XX 09-y8 on RamDisc with "Data Folder" on HDD. It is not better.

The next question is ...  What can I do if I like to use XX 09-8 on RamDisc ?

Regards

[PeterSt]

The answer to that is simple, but keep in mind it is my point of view :

Using the program (which would be XXHighEnd.exe / XXEngine3.exe) from a RAM disk does not make sense at all. Not to me.
Besides, the amount of RAMdisk needed it out of your control, because XX will use temporary files which consume your ram disk.
Next, your RAM disk consumes memory, and *that* memory is deeply needed (keep in mind, XXHighEnd is a memory player).

Using the RAM disk for the music data, not only doesn't make sense, but also makes it WORSE. Why ? again, XX is a memory player, so the only thing you would be doing is creating another step (or layer if you want) at copying the data from disk to your RAMdisk, while from RAMDisk it will go to normal memory.

Notice though there would be a kind of (implicit) trick at using the "Copy to XX drive" parameter, because the music data will be copied to your XX drive which will be the RAM disk when XX is put into that RAMDisk (see first alinea above). Thus, this allows you not to put it there yourself (did you find that out already ? if not, you could try it).

Of course it is up to you to work like this, but you must know that I (XX) myself do everything and all to let these things go as efficient (or "less consuming") as possible for a memory player. Of course, part of this comes from it just *being* a memory player, but another part just brings additional problems (like the copying which takes time).

All 'n all, the "Copy to XX drive" parameter was created to be used with XX on an SSD drive, this allowing for the data drives to spin down, BUT when you don't have an SSD, this indeed can just as well be your memory itself. Sadly, you will never have enough memory to hold everything which is needed (at some stages) and which, as said, is out of your control. It could be several GBs for one album, or relatievely infinite for a playlist of, say, 100 tracks.

Not advisable !
Peter


PS: I am not quite sure whether you have overcome the original problem of this topic !
Let me know ...

[guytou1]

Yes, the original problem is linked to RamDisc.
Now, evething is nice when XX runs alone from HDD. I have more problems when Firefox is running at the same time. I will explain it later.

[PeterSt]

Firefox is a bomb of overhead anyway ...

[Telstar]

Yes, the original problem is linked to RamDisc.
Now, evething is nice when XX runs alone from HDD. I have more problems when Firefox is running at the same time. I will explain it later.

This is a new one. But could be.
On my studio I have firefox running all the time, and seldom i experience xxhe crashes (MUCH less with the newer version i have to say). I blamed creative drivers.
Not a single crash (finally) on the dedicated computer where nothing unneeded runs, since y6 i think to remember.

[guytou1]

It seems to be much better now after setting thread priority as RealTime.

[guytou1]

Having strongly increased the size of memory allocated to Ram Disc I can now play XX on RamDisc. As you told it, Peter, it is not easy to share the limited RAM memory between windows (and XX through windows) and the RamDisc.  But I really prefer the sound of XX on RamDisc, even if it is difficult to find a rational explanation.

An other problem (also with XX on HDD) :
Especially when I use Anti Image sample rate conversion from 44,1 to 96 kHz.   When going from one track to the next one XX creates a UnicodeTrack0002a.wav file, begins to read it and creates at the same time a UnicodeTrack0002.wav file.   But the cursor begins only to move when UnicodeTrack0002.wav is finished. It causes a gap at the end of the track, can lead to a crash of engine #4, and sometimes to a crash of Hiface's driver.

A very positive point to finish : I measured the file resulting from sample rate conversion with XX and compared it with a file resulting from a conversion with foobar and SRC. The residual components above 21 kHz are lower with XX, and it uses less computing power.   
(as I tried several sample rate converters, there seems to be a correlation between high frequency components and sound quality)

[PeterSt]

I must say, you did some thourough investigation here.

But the cursor begins only to move when UnicodeTrack0002.wav is finished.

When it has finished creating, yes. I know this, and this is why I told (somewhere) better not to use this in Attended mode; With Unattended this can't bother you, and besides that the a version will be created for the first track only in that (Unattended) case.

Of course this can be solved, but it is a kind of complicated, hence takes time I didn't want to spend yet. But merely :
If you are keen on perceiving a better sound at playing XX from RAM (which sure will be true, never mind a lacking explanation from at least my side), you 100% sure should use XX in Unattended mode. The difference in SQ is "huge", and should at least be constant from version to version. This *will* not be the case for Attended, where SQ really is not under my control.

Please try it ...

About the sample rate conversion ... well, it should be better, knowing that I created it by means of measuring for just those high frequency components to be the least. I must honestly say though, that I didn't pay as much attention as could be possible to all the different from-to conversions, but at least 44.1 to 176.4 should be optimal. Notice though this is not ony about what's left for nasty distortions, but that it is also about ringing (always subjective to Anti Imaging filtering). Generally we can say : the further the rolloff starts (towards 20K) the steeper it must be (transition band), the more it will ring.
This is why I created Arc Prediction, which doesn't ring at all. But, this can not be used for "uneven" conversions (like from 44.1 to 96) ... sorry.

Thanks,
Peter

[pedal]

...This is why I created Arc Prediction, which doesn't ring at all...

Bingo!

Next step is to have ArcP available also with 24/88.2 and 24/96 files.