SQ AP,AI and Buffersize

[Flecko]

I tried the new version and accidentally I found I was using the wrong buffersize... I had all Q still set to 0 and suddenly after some minutes of listening there was a buffer error...so I guess 1024 is not the right size for Hiface? Ok, after that I set buffersize to 2048 and guess what? Better sound yeahh! So far so good. Now I wanted to try AI and AP again. Now I would say AP is definitly better than AI, it was the oposite before that buffersize change. Now AI sounds like there is some grain in the high frequencies. AP Is clearer and closer to the not upsampled sound, also better than the non upsampled sound as I can say so far. Another thing is, that I have now the best sound with all Q set to 0. I don't know if this makes any sense at all, it shouldn't as far as I know but it seems to be clearer set to 0. So that is for now and my impression might change but as always I cannot wait to settle the impression.
Greetings Adrian

[Marcin_gps]

AFAIK Q1=1 + 1024 samples = Q1=0 + 2048 samples (half of buffer size value), so you're back home

[Flecko]

Hi Marcin,

AFAIK Q1=1 + 1024 samples = Q1=0 + 2048 samples (half of buffer size value), so you're back home

Great! Always these autosuggestions. I suffer from it probably the most. But why I got the buffererror at 1024? That should not happen if I use the right size or not? So following the instruction I must set it to 2048?

[Flecko]

The instruction to "how to determine the right buffer size" says if I don't have buffer errors in the "first few seconds" it is ok. So what are a few seconds? 3, 7 or 29? With q1=0 I get errors after 29 seconds with a buffer size of 256kb but with q1=1 I just can listen 3-4 minutes until I get errors. 29 seconds are not a few I would say. It seems no good way to determine the buffer size of the device anyway. And as far as I have read in the hiface thread nobody found the true size yet. But that also means that adaptive mode should not work right for the most people because they don't know the right buffer size of the hiface (or is this a system dependend variable. it shouldn't if I understand the instruction right)? The sound is always good of course. There are small differences but I wouldn't say that they can show if this mode is giving it's full potential or not.

[Marcin_gps]

Hi Marcin,

AFAIK Q1=1 + 1024 samples = Q1=0 + 2048 samples (half of buffer size value), so you're back home

Great! Always these autosuggestions. I suffer from it probably the most. But why I got the buffererror at 1024? That should not happen if I use the right size or not? So following the instruction I must set it to 2048?

Yes. Check http://www.phasure.com/index.php?topic=1299.0 and my 'solution'.

[PeterSt]

The first few seconds is merely 1 or 2 *and* it is about "too many" (which is 250 x 10 I think).
Buffer errors along the way is always about a few here and there, and caused by external influences, with changing the wallpaper as an example. Notice that the time is mentioned in the X3PB log file at "changing wall paper" entries, so you can check.

No errors should be there of course, but "a few" go unnoticed, plus they are unavoidable some times. Only when it is a constant stream of errors, *then* it is really wrong and you will hear that. But long before that playback stopped already (because of "too many" etc.).

HTH
Peter

[Flecko]

Yes. Check http://www.phasure.com/index.php?topic=1299.0  and my 'solution'.

thx for the link. But I have no problems with distortions and I can only use DAP because of my DAC is a 96kHz unit.

I like to determine thie right buffer size of my hiface to use adaptive mode optimal. Most people were using 1024 so I thought I don't have to test it. With DAP 1024kb BS I get the to many buffer errors massege with Q1=0. This does not happen with 2048, Q1=0. But I can go down to 512kb when I don't use DAP and Q1=0. If I go lower than 512 I get the buffer error massege also with Q1=1. So 512 is the right size or 2048?
Is the buffer size set by the driver or is it given by the hardware (hiface) itself?  In both cases I can try to ask M2Tech for the right value.

[boleary]

Hi Flecko, using DAP, Q1=1, I've never been able to go lower than 2048 with the Hiface, core appointment scheme2. With scheme 3 I can sometimes use 2048 but more often 4096.

[PeterSt]

Is the buffer size set by the driver or is it given by the hardware (hiface) itself?  In both cases I can try to ask M2Tech for the right value.

That might be the best idea, and then hope you can an answer.
On the high side -and seen from within Windows- the HiFace's buffer is infinite (because there is OS stuff in between it).
On the low side I don't know it.

Notice that the answer on the low side may be rather important, because it probably is a strange number (which immediately will complain the mixed results). In that case I could anticipate on that strange number ...

Peter

[Telstar]

On my studio pc i'm using now 2048 samples with Q1=1.
1024 worked fine but rarely i got the "tto many buffers error", so i raised the buffers and all is good now. VERY good.

[Flecko]

1024 is the smalest size I "never" get too many buffer error. 2048 sounds softer than 1024, I like it. I now use doubling without AP or AI. I don't have a filterless DAC and so it makes sense not using a additional filter. After long time experience it is also best sounding in my system. AP is somtimes nice with rythmical music. It gives the sound a focus on single instruments. With classic music it sounds still a little bit "unnatural". AI is realy not as good AP. I listened to AI with classic music first and there it seems to make something nice but it turns out it is very bad with rythmical music. It adds some strange fuzzynes to the sound wich sounded first nice with the violin but then...