XXHighEnd

OK, so at the time of writing, this version had been downloaded 43 times... enough to start a topic, I think.

On my system, Engine#4 comes across in a very 'down to earth', 'organic' sort of way. Engine#3 is great (better than before?) but is 'in your face' compared to Engine#4. Maybe there's a time and a place for either... but for me, it's Engine#4 all the way.

Mani.

Mani, I downloaded it and would agree with your comments.  I'm using a Lynx AES16 card on Win7 and I have found that I can't use the lowest buffer size of 32 in Special Mode - it sounds like slow motion.  I hear a few cracks here and there....still experimenting....but overall a nice step forward in SQ.

Question - if the SQ improvements of this latest version are due to reduced latency, then would it not stand to reason that a properly designed async interface would be even better?  My experimentation with my AES interface has made it clear that final SQ out of my DAC is influenced by a bunch of variables that I'd rather not be at the mercy of: e.g. proper cable impedance/type, latency, electrical noise isolation, jitter of the source clock, etc.   Asked another way, if one is using an async USB interface would this latest XXHE version produce audible improvement?

 

I had the same problem with the music playing in slow motion when buffer size was set to 32, but when I set player priority to low and thread priority to real time, the music played correctly.  All the pops and clicks also went away when I set the priorities.   

Peter,

I cant get special mode working. From 1 to 8186 there are cracks. I like to say though that the higher the better it gets. But even at 8186 the sound is not good. Full off cracks. Changing priority does not help either.

Can you help?

 :)

Hi Gerard,

If you tried with Q2/3/4/5 = 0 I guess you have to give up on special mode. But the cause will be the soundcard with a too large buffer, I think.

Peter

No special mode here either. Also, one would think that 9y-6, with AP turned on (2x in my case) and peek extension checked, would sound identical to the same set up with 9y-5-06 (peek extension there by standard) but I think they are a little different. Might just be my imagination on this (more testing needed, the Olympics one upped my listening session last night) but the integration of the mids and highs seems more smooth in 5-06. Anyone else notice this?

Also, with Q 2-5 at 0 the lowest I can go with Q1 is 8 (normal mode).

The same for me as for Gerard!

A little disappointing because in the previous version special mode worked.

It works fine with buffer size set to 256, and the Q1 set to -4, (Q-parameter 7). The sound is more detailed and here are a lot of "room information". The piano is great :)

Hi Peter GerardA,

i have connected the dac trough usb and now it does work...

 ;)

Mani,

I'm new to the forum but tried the different versions of the last months. I never used the Vista or Windows 7 path but could hear always a significant difference of the XP versions. This version sounds very musical for me. My coax louspeakers are very accurate and the actual soundstage is fantastic. Although the frequency-response characteristics of the speakers are catastrophic the sound is very sweet and musical.

Georg

hey Peter ...

in the release notes for 0.9y-6 you make mention that with Special Mode and a latency of 2 samples that
I would have to disagree on that point .. haha

With the Device Buffer Size set at 48, I have my system running at 1 sample ( Q1 slider at -3 , and the tool tip indicates a parameter of -4 !?!? ) in  attended mode and everything sounds fantastic.

In fact I've been running with the Q1 slider at -1 in Special Mode using Unattended most of the afternoon. The only change to previous procedures has been to untick "Start Engine#3 during conversion", otherwise I get a few clicks etc until things settle down.

I'm really enjoying what i'm hearing with KS and it's ultra low latency.

Bravo !!

Now then , any tips as to what the numbers in my X3PB log are referring to ??


Cheers,

Russ

oh BTW ... might be an idea to put a trap in the code to stop someone from trying to run with Q1 slider at -4 in Special Mode. It might save wear and tear on the reset button :grin:

Peter,

Thanx! By far the most beautiful sound i have ever heard on my system!! :ok:

 :grin:

You guys keep on saying that each time ! You don't want me to keep on believing that, right ?
:swoon:

But I think similar. If you only know how enthusiast I was when I heard this for the first time ... I even created a post about it, but since it started with :teasing::teasing::teasing: and I wasn't ready for an upgrade back at the time, I didn't post it.


Btw, I am going to have a small problem with it ... I need more output dBs ... haha

Peter, Russ,

Do like what I hear now, it needs more investigation, but first impressions are very good.
Q = -1 in special mode works best for now if I go lower I will get clicks and pops. (at attended)

For the first time I like KS for all music......but again these are just todays findings.
Have to find the absolute boundaries of all settings.

Glad you didn't gave up on KS, Peter.
I know It took you lots of work, but I think It did pay of in the end.

Good Job :clapping: :good:

People may also watch the OSD Time and whether that "clicks". In an earlier version I had to shut it off myself, although now it's allright again. But *if* this bothers to anyone, please let me know, because I can do that one somewhat smarter (less eating).

Russ, about those numbers :
They indicate that you are running with a latency of 0.048ms at 4 samples which actually is 1 and one byte spare. :yes:


Does it ? the tooltip on the button itself ? I think you misplaced a couple of things *or* the tooltip wasn't initialized for some reason. So, thinking about the latter (of course :)), if you'd slide the thing a bit back and forth, it will show the samples allright. 1 in your case.
Btw, if it doesn't show that, how would you know it is 1 ? hehe

Peter

And Russ, did you check for the lot being bit perfect ? I mean, don't be disappointed if it isn't.

I am *not* implying that what you have running, can't be. I just didn't go lower than 5 samples myself because I thought it would be rediculous. So, I stay on 22, and know you might have better sound. Why not, haha.

Take all compliments you can get
You won't get them about your good looks, hehe
Then we would be lying for sure

In both (un)attended modes i can have the Q1 slider at -2 ( 1 sample ) and play back my 48kHz DTS encoded albums fine :)


Cheers,

Russ ( with the pedal to the metal ! )

Hi Peter,

The Audiotrak Prodigy HD2 is working in KS mode (engine#4) in Special mode Q1=4 (parameter=22). The sound is great!!! Detaills,detaills, stereo effects are great,....I think ....the best sounding version so far. Congratulations! Did you hear the last Yello cd "Touch Yello"? Very spectacular sound! 

Sometimes i hear a small hickup/crack in the sound! What is the best strategy to solve this? I suppose to set the Latency of the soundcard to the lowest as possible value. The Split file size back to 100. Treat Prio= Realtime. Changing these values back didn't solves it anymore.

Johan

Hi Johan - Thank you very much.

No, it is not necessarily true to use the lowest buffer setting on your sound card. E.g. for me I have to set it to 128 for the 22 sample Q1 setting. 64 works too by I don't like the CPU usage coming from that, and 48 just doesn't work without regular ticks (or small skips). So, it depends ...

In a next version I will try to make something (out) of this.
Peter

:heat::heat:

You may not believe it, but I really didn't dare to check it myself. I just didn't want to know it if what I perceive as extraordinary good, would be specifically wrong. This is why I gave the piano example, because it really is special (I won't tell what so others can discover it) and leads to the suggestion of something being wrong. Well, as I told in the release notes.

Thanks,
Peter

Hi Georg, welcome. It's a great time to join the XXHE 'club'...

Mani.

No, not here.
I think I'll go back to the previous version, there I played LOUD, now I don't feel like doing that anymore.
Maybe the disappointment-placebo-effect?
Peter, can I just put the old version over the new one? (Backups why?)

Peter,

What's a good target CPU usage?  I'm not getting any ticks but my CPU is at 24%.  My buffer size in XXHE and on my Lynx card is 32, and my Q1-Parameter is 16. 

Thanks,
Dan

Should be just virtually zero Dan. A constant 24 % ... hmm ... but that can be because of a just "under powered" processor. But I have serious spikes at illogical moments, coming from the driver. Like once in the 5 seonds (and there's nothing "I" (XX) do once in the 5 seconds ... this, while it influences SQ).

Sorry that it doesn't work out for you Gerard.
If you know wat you are doing you can put anything over anything ... but you can also just download the base version (0.9y-5-00) again, and paste over that the 5 version you like.

To reduce CPU use, am I better off increasing Q1 or changing buffer to a higher setting?

Increasing Q1 helps, but relatively little. What the influence of the soundcard's buffer is you can only see yourself. Notice that the cpu useage a soundcard's (etc.) driver may incur for (the amount of it) allows for playing 100 tracks at the time ... that bad. And to keep in mind : it is the spikes (like once in the x seconds) which are bad. A constand useage is okay, whether it is near zero or 24%.
All IMO.

Peter

I have the Lynx AES16 card on Win7 like Dan and I'm getting a CPU usage of a fairly stable 5-7% with my Lynx card and XXHE Special Mode buffers at 32 and Q1=16.  If I move Q1 to zero it increases the CPU usage to 34-40%.  I'm still experimenting.

Earflappin,

Which Q1 value of 16 do you mean?  The tool tips value or the old one?  When I set the old value of Q1 to 16 (tool tip value of Q1 = 35), my CPU-utilization is 7-9%.  When I set the tool tips value of Q1 to 16 (old Q1 value = 5), my CPU utilization is 19-21%.   



Thanks,
Dan

Dan, I guess I'm confused about what you mean.  Can you clarify?  I was referring to my Q1 setting in this latest version of XXHE - 0.9y6.  i.e. My Q1 setting on the main panel is 16.  In the previous software version I had this set at Q1=0.  Man, this stuff just keeps getting more and more geeky.

I couldn't agree more about the geekiness! 

What I notice is that when I have my mouse on the Q1 slider, it give me a new and different value than when I finish making the adjustment and move my mouse away.  It behaves like a tool tip in that it appears only when the mouse is hovering over it and disappears when the mouse is moved away.  What appears is "Q1-Parameter NN" when NN is the current value, and it is different from what you see when you look at the panel after the fact. 

You've answered my question though! 

OK..I think I broke the code now....I went back and read the release notes for the 0.9y6 version.  I am running in engine 4 special mode with 32 buffer and with my Lynx card buffer at 32.  Q1=16 and Q1 tool tip = 8.  This is giving me a CPU usage which is pulsing between 1 and 12 and seems to vary based on the sample rate of the music I play.  I need to experiment more to see if I can get a more stable CPU usage.  Sound quality is very good however.

OK...think I'm getting a handle on this.  As a next step I changed the special mode buffer from 32 to 64 and it immediately reduced and smoothed the CPU usage to an almost constant 2-3%.  My tool tip Q1 value changed from 8 to 18. I kept the Lynx card buffer at 32.  Changing it doesn't seem to have much impact on the CPU usage.

How does one decide the optimal Q1 setting for a given special mode buffer setting?  Peter, I love ya man and appreciate the progress on sound quality, but is there any way to de-geek the GUI? 

OK people, here is how it goes on laptop with DAC via USB in 16/44 mode without any upsamplings:

I can run good and detailed sound with amazing stereo having the following settings:
All Qs=0, Special mode, buffer=32, attended or unattended. Core appointments - Nothing and High. Scheme-2.
Not sure still about bass, have listened just this morning, but for sure sound since last version has more details inside, individual stuff is more in sync with each other, spatial presentation. Engine#3 for example also has a high detail rate, but these individual things do not make such a concerto like it is done by engine#4. So, these are my first impressions.

Problem: I tried running 24/96 with Anti-Image with the same settings as above (my DAC does not support 88 over USB) and after each track has played around 75%, sound just disappears. After 15 more seconds Engine#3 fails with error message that it is closed.

Question: If I feel crackless and popless sound with all Qs=0 and the lowest buffer setting in special mode, can I consider that I have almost ideal setup for what KS can bring me??? Have not yet tried going with Q1 < 0.

Ivo

Hi Ivo - Good !

Well ... from your description about the problem with Anti-Image, I now suddenly realize that this uses something which Engine#4 can't bear. Or at least I need to pay attention to this, and I never did ...

Q: Does AI work fine with Engine#3 for you ?


You are the best to answer that Ivo. Keep in mind though, the lowest Q1 just is -4, and there is really no logic in zero vs. below that. So, e.g. -1 is just lower than 0. That's all.
But generally ? yes I think you already have it at a level that is difficult to outperform. But then keep in mind : I never really listened to anything under 22 samples, and I wouldn't know (at all) what it can bring more for SQ at going lower.

Peter

Well, to all of you who can use 'special mode' at really low Q1s... congratulations! I myself have totally glitchless QAP sound with a Q1 of 569 in 'special mode'. Perhaps a little high, but there you go.

I loved the previous Engine#4, with my own Q configuration of 3, 9, 15, 5, 10. But note that this remained in 'normal mode' - I could never get 'special mode' to work without distortion. Even in 'normal mode', the tonal balance just seemed right. But there did seem to be a downside, and that was a slight loss of dynamics. Now with the latest Engine#4 in 'special mode', I hear the same beautiful and natural tonal balance but with all the dynamics intact. And there really does appear to be more detail. Where on earth this could have come from, God knows.

If you can't get 'special mode' to work, my strong recommendation would be to change whatever you need to in order to get it working.

Mani.

Thanks Mani, green with envy here..... Hopefully, soon, I'll be hearing it too!

Just like to say (again) that i am so pleased with this new sound!! Vocals have a new dimension and the extra details i am hearing is unbelievable!!
If this is the end of what is possible i will be very very happy with it!

 :xx:
 
 

Additionally, because it hasn't been said all that explicitly so far : keep in mind that it is an interaction between what the driver does and what XXHighEnd does. So, suppose your driver can be set to 48, 96, 128, 256 (etc.), then while you can't find a working mode for (e.g.) 128 and for 48, 96 may well do it !

Maybe I should apologize for this indeed geeky way of going about with it, but in the end it *is* just so that what happens is far from normal. So in that context we have to operate this, but then it is for the good cause.
In a next version I hope to "automate" this process a bit, but I am not sure (at all) it can be done.

Peter

I want to add my voice on the SQ of y6 with KS. It seem to add the best of wasapi and KS in the previous version.

Moreover, the occasional noise when browsing explorer or the library is gone. The sound is solid.
The downside is an AWFULLY SLOW interface. No matter what priority i used :/

-The gallery is unable to show more than 8 albums at once, then it reloads them and enters an endless loop.
-Playback of the first song starts after 5 seconds

It seems like going back to the very first version with an underpowered machine. Looking at the task manager there are cpu spikes when using the interface, but nothing when playing.

Settings used: q1=4 (my old time favourite is back!), q2-5=0, special mode, latency 32.

Yes, I will do something about it. This topic clearly shows how confusing it is. Maybe not for working with it, but indeed for talking about it.

Hi Telstar,


Is this with the thumbnail cache in use ?


Although this depends on many things, I don't see logic in this being more slow than before. So, there's only logic in this if you "caused" it yourself (like using FLAC now, while before it was just WAV).
I don't want to put tasks upon you, but if you can proove that with the exact same settings and tracks an earlier version started playing quicker, please let me know. In any case, don't forget what I told myself about W7 and its inconsistencies, which can be observed at using the Normalized Volume which for one album may take 5-7 seconds (which would be normal) and for another 1 minute.

Peter

earflappin,

I'm not sure that you ever got an answer to your question. I certainly don't have one.

What I have done though is to have a look at everyone's setup and the latency (Q1 value) they're getting in Special Mode. As you can see, my isochronous firewire setup (perhaps the closest to asynchronous USB here?) doesn't fare too well. It seems that PCI is the way to go...

Mani.

For the record:
I can't get any better than Q1=22 (587) with Buffer Size =768 playing QAP Double.
My interface is Firewire to Fireface400 followed by SP/DIF to TwinDAC+ (battery powered).
The sound however is absolutely marvelous, using Double only because my DAC doesn't accept Quad.

Cheers,
Eric.

Also for the record,
with USB I can get very low (2 samples), but only one channel is working, guess a DAC problem.
With firewire/SPDIF I can listen at highest Q-rate and only 88 samples in the Terratec, but with constant rattle.

I'm now running 32 buffer size on both XXHE and my Lynx AES card with Q1 sample size of 3 with no QAP and its the best sound I've ever gotten from my system. 

I've also found that AES cables make a big difference in SQ.  I started off with a DH Labs 110 cable which was a step up from the stock Lynx cable and the Gotham cable.  Then I switched to a custom Cardas Neutral Reference DB25/XLR cable and the improvement was immediate and not subtle. 

Once Antelope starts shipping their Zodiac DAC's I'm going to get one to reclock and noise isolate the AES output from my Lynx card.  I'll also be able to A/B an async USB interface vs the AES interface.  Should be interesting.  Again, I don't see why async USB shouldn't be superior to AES.

Gerard - Interesting ... That is, at a higher setting the two channels start to work ?

I have the same DAC and i am playing at 5 samples and everything is working great!! Below this 5 there is no sound on both the channels.

 :)

Can confirm once again my current setup:

SETUP: Dr.DAC2 over USB, No upsampling, DAC is set to 24/96, all Qs=0, Special mode, buffer=32 (the smallest), playing 16/44 and 24/96 material. All this on Vista HP.

SOUND: Crackless, popless and very nice!!! Stereo amazing, space huge - very well. However, CPU constantly at 25- 35 % during playback. I have AMD Athlon 64 Dual core. Otherwise files are loaded and converted FLAC-->WAV very fast. I am satisfied with the latest XX.

FUTURE: Waiting on HiFace unit, so I can go into DAP and QAP world with Arc Pred. Maybe someone can share the lowest settings for engine#4 with HiFace?

Ivo

I don't know about the lowest setting, but I used it with 22 (samples).

No, I would wish so!
I guess they made a bad USB-to-SPDIF converter in China for my DAC.
I don't think Left and Right get seperated before the input/samplerateconverter but it looks that way.
Or it must be the USB-codec in the PC, since 'test' is also like that.

... meaning the sound going bananas at a volume change (or Absolute Phase change) at the ultra low latancy settings ... this is a bug, so to say. Notice that I'm talking about the situation that the OSD Time would turn white, which is at track (part) boundaries.

I only want to say : if you calibrated everything including being able to cope with this ... the next version won't make "fuzz" :) about this anymore.

Thanks,
Peter

It's working on USB!
Both channels. I think it's the shorter USB-wire that made the difference but not 100% sure.
Don't want to break it to find out...
It plays 16/192 max so no arc, don't know if it sounds better, but it sounds good.

Me having a real short USB cable so you may be right on that!

 :good:

(I never have enjoy music so much as i have done these past week) (Really!)
 
 

Haha, Everything is better now with your new little babyboy!!!   :drinks: :drinks:


Yeah, but that means the Dac has to take the place of the recordplayer, oei, oei...(ai, ai)

Amen!

Just wait a bit. I'm improving the sound here.
I think.
:blob8:

 :rofl: Why am i not suprised!  ;)

         Go Jerry Go Jerry.... :yahoo:

 

I have been on Special Mode for a couple of hours. It's frustrating to find the noise-free spot*, but when it works: Oh, my, oh my - the sound quality takes Red Book into new territory where mankind hasn't set his fot (ear) before!

It's like the final CD veil has been liftet. More bloom, more inner detail and more transient details.

------------

*The combination of Latency and Q1 seems to be interactive. The Q1 setting working with fex. 512 latency, doesnt necessarily work with 1024 latency. I have to scan the Q1 between -4 and 30, each time I change (reduce) the latency.

Also a setting works fine for one song, but next song in the playlist is heavy distorted.... In fact, it's hard to find a combination which gives noise-free sound from a full album. Right now I have the following combination (latency 1024 + Q1=13) = Q1 parameter = 1022

The gain in SQ is well worth the work, though!!!   :yahoo:

One more note regarding searching for Special Mode optimum settings:
Re-reading the Download and release notes (http://www.phasure.com/index.php?topic=1113.msg9623#msg9623) I realize all factors are mentioned by Peter.

If you havn't found your noise-free Special Mode settings, I urge you to re-read "the manual" written by Peter. It seems also that both Processor Core Apointment and Priority schemes must/can be changes to get noisefree Special Mode.

Hopefully Peter will make Special Mode more automatic in use.

What do you think I have just been doing, eh ?
:whistle:
It is not 100% equal to what you talk about, but the 90% will be sufficient for most.
:secret:

Watch for the next version, and Adaptive Mode.:swoon:

Several times in the past, we felt the present version to be the "best ever", and we asked Peter to go on with the user interface and get out of "Beta-mode". Then he went on and made the next version even better sounding. Again and again. I am glad you didn't respond to our crying in the past! LOL

I also love the SQ improvements that special mode provides in my set-up where I'm using a synchronous AES interface to my DAC.  But as I said in another post, IMHO the AES interface is the weak link in my system.  Why?  Because sound quality is compromised by: (1) PC latency, (2) clock jitter, (3) noise and (4) cable geometry.

Peter's special mode show that reducing latency improves SQ.  I have proven that reclocking and electrically isolating the AES output from my Lynx card improves SQ.  And I have also proven that different AES cables sound different and better. 

IMHO, a DAC with a properly designed async interface should offer superior SQ assuming the rest of the DAC system has been designed optimally (i.e. power supplies, output stage, etc.) and alleviate the audiophile from having to screw around with these other variables. 

BTW, the new Antelope Zodiac Plus DAC will ship on April 20th.  Should be interesting.

I've replaced my Weiss AFI1 firewire interface with an RME AES32 PCI card. I forgot just how much I love the RME gear - the hardware and the TotalMix/Digicheck software. I appreciate the simple things, like having a properly terminated BNC wordclock input, which is so important when you're slaving the interface to the DAC, like I am.

Also, with Digicheck, I can verify that QAP adds extra HF information, smoothing the drop-off above 22KHz. (Interestingly, I'm measuring QAP at 23 bits, not 24!)

Another benefit is that my Special Mode Q1 value using my humble Atom-based PC has come down - I can now play with QAP at around Q1=160. But I agree with pedal in that finding a 'stable' value is difficult.


If you could make Special Mode less hardware-intensive, I'd be a very happy chappie...

Mani.

Hey Mani - What if Peak Extension is unticked ? does it make a difference ?
(it's diificult to reason myself how things work out, which is because of the integration with the Normalized Volume, but if anything it shouldn't make a difference :scratching:)

Thank you,
Peter

Hi all,

Firstly a big word of thanks - Peter you have made yet another fantastic player upgrade  :clapping:

I wonder if anyone can help, I am trying to get my M-Audio Transit to run in Special Mode. I am just getting "slow motion" music for all of the Transit latency settings and XX Q1 settings I have tried so far. Has anyone managed to work out settings for the Transit or Empirical Off-Ramp that allow Special Mode to work ?

Given how much better Engine 4 sounds I am really keen to get engine 4 special mode going as well.

Nick.

With Peak Extension ticked = 24 bits
With Peak Extension NOT ticked = 23 bits

Is this what you expected?

Mani.

PS. It's good to have Digicheck back!