XXHighEnd Model 0.9y-6 (Full XP support with Kernel Streaming)

[PeterSt]

(download is at the very bottom)

For those who are new here : do notice that Engine#3 (WASAPI, which only runs on Vista/W2008/W7) is a totally different beast for the character of the sound than Engine#4 (Kernel Streaming, which runs on XP too). Both are in the bit perfect realm, but both sound totally different and also have different capabilities to vary the SQ.


Another little project is finished : Kernel Streaming (Engine#4). All known features from Engine#3 are supported now, including "Q1 manipulation" though by very different means.

Below enhancements/changes are not necessarily for Engine#4 only.

• Full Kernel Streaming support
  
  So, now XP users too can benefit from all what is done within XXHighEnd to create the spectacular sound, known from Vista (and Windows 7) by means of Engine#3. Actually, Engine#4 is expected to superceed Engine#3 by miles for sound quality or music listening through loudspeakers in general. On that matter it contains techniques not possible within Engine#3, and it goes more ultimately in the direction where Engine#3 had to stop. If all is right this is easily audible, while at the same time Engine#4 pertains its own character to the sound as known from the 0.9y-5 versions.
  
  
• Q1 support and Special Mode
  
  Edit / Important : One of the key elements to let the below work as fluently (and as low latency) as possible, is the Core Appointment Scheme. Now, it was only realized afterewards that anyone in demo mode can't apply that. Please keep this in mind at your "low latency" attempts !
  
  The early-attempts from the last 0.9y-5 version have been exploited further, and now are working at a level which goes way, way beyond what "Special Mode" in 0.9y-5 could do;
  
  In the Settings Area there's a new "Special Mode" checkbox, and most probably it can be used by everyone now.
  When ticked you will switch to an ultra low latency mode, unknown to the Pro world of (computer aided) musicians. This is so low, that a sneeze may loose some samples, so to speak. The core2duo - 2.4GHz of your humble developer goes as low as .... 5 samples, with the data rate as implied by 32bits and 176.4KHz (@ 2 channels). Now, 5 samples may be ridiculous, and it is better to stick with 22 or so (which plays glitch free here, for albums and albums in a row). Notice that the effective latency is always determined by your Sound Device's latency, which -if settable- normally will go as low as 48 samples. In a windows environment (any environment in fact) it would normally be fair to say that 48 samples can not be reached for effective latency, while now this is achieved easily (!). Ok, is this important to music playback ? ... just have a listen.
  
  There's a little fun in this too, because at this latency you can be about sure you will be loosing samples indeed, and at perceiving the unbelieveable detail this brings, well, you may think you do indeed. After listening to some random music, and at being sure not to perceive any ticks, cracks or glitches (this is a prerequisite, because you obviously can go too low on this !), you may start some piano music. It doesn't need to be good quality, just piano, and just part of a combo or something. *Now* to will be subjective to the 100% suggestion you are loosing samples. Or wait, repeats maybe ? Right. When you are as far as this, now try to proove it with orther music.
  Can you ?
  
  
• A new combobox in the Settings Area "Device Buffer Size" (for Engine#4 only)
  
  Although originally created for something which has not been exploited yet, currently it does this :
  When in Special Mode, and when set to the latency in samples of your sound device (assumed it is known), it implies a best Q1 range for that device latency;
  The choice you make there does nothing to the sound as such, but it may help you "being" in a Q1 range that is allowed by the device without ticks or anything. This does not imply you won't have ticks, because generally at the low end (towards -4) you will almost sure perceive them. But, there will be a range in the middle somewhere, where all is right. This in other words :
  
  No matter what choice you made for this combobox (which indirectly includes the situation that you don't know the latency of your device), you will be offered a range which is different per choice with an overlap from the one subsequent choice to the other. Now, the lowest latency choice will offer the lowest latency working range for Engine#4. And, what your latency choices are can be seen on the tooltip of the Q1 slider button, and it will show in samples.
  All summarized : whatever choice you made for the Device Buffer Latency, the tooltip on Q1 shows the active latency in samples.
  
  Be a bit careful with this, because when switching from Special Mode to not Special Mode (Normal Mode), Q1 behaves differently, and the Tooltip will just show the absolute settting as you are used from Engine#3 (+30 up to -4). So, with a setting of e.g. -2 for Normal Mode, and next switch to Special Mode without adjusting Q1, may give you a sudden latency of 2 samples, and nothing will be able to bear that. This is not window breaking, but a nasty sound anyway.
  Important : The sound coming from missing samples can vary very much from setting to setting. One may be much louder than the other, or the one may sound really nasty while the other just softly ticks without further sound.
  Besises this, a too high setting also will create a nasty sound, which merely counts for Special Mode ! It just means that you overdo things at the wrong side, and while the mechanism is made for low latency, you feed it with high. It can't take that. However, hopping over to Normal Mode and all will be fine. ?
  
  Be careful : When at first glance everything works tick-free, you always have to take into account other "activities" you may imply, and which just may cross that border of being tick-free. A main area of this is the initial startup of playback (with conversions going on etc.), the switch to a new track while the Wallpaper has to change, and the change of the Digital Volume (which also changes the wallpaper). The latter could be the worst because it won't recover. Thus, *if* you receive ticks/cracks/worse at changing the Digital Volume, the level of it will stay and things are definitely out of balance. One remedy : stop playback (which in that case is extra difficult because of the resources eaten).
  While this merely counts for Special Mode, in both Special Mode and Normal Mode (but only Engine#4) you may notice that you "switch" something at changing the Digital Volume. Mostly this will be inaudible, but sometimes it will be clear. Maybe this can change in a future version.
  
  
• Lost Buffer Detection for Normal Mode (Engine#4)
  
  Why ? well, because it just could be done. Remember, this is about internal buffers and not about the Sound Device buffers BUT a "wrong" (read non-matching) setting of the Sound Device Buffers (latency) may incur for internal buffer loss. And also to remember :only for Normal Mode and only for Engine#4.
  Lost buffers may not be audible when they come with a few only. Nevertheless you wouldn't want it, while at the same time you (most probably) will incur for it by means of too low Q1 settings.
  When the lost buffer count reaches 250 - and which is counted per track - a message will show you this, and playback will stop. Besides this, the X3PB log file will show the lost buffers (Buffer Errors : XX) when they are there, and they increment by 10. So, even when you never see the message, the errors may be there although in a quantity of less than 250 per track. It may be good to check this once in a while.
  
  
• Changed priority schemes
  
  The whole set of priority schemes has had a workover, and the main reason is to sustain the low latency mentioned before - when other tasks need to be employed. However, since this has been tuned on a Windows7 machine, chances are (very) fair that those who encountered glitches when they started to use W7, are now rid of those. Now keep in mind : most probably you already *were* rid of those, but by means of tweaks not meant for that, and they are rather meant for better SQ. For example, setting the Split File size lower helped many, but is really is the intention to set that as high as possible (keep the disks sleeping etc.). So, don't forget to review your settings !
  
  
• 24 bit "only" support for Engine#4
  
  For Engine#4 too now, devices which can take "24 bits only" (opposed to 32 bits which is normal), are supported.
  Additionally a message has been built in which tells you to be careful at choosing this ("DAC Need 24 bits" in the Settings Area), because to many (if not all) it seems the obvious choice. However, it is not, and devices needing this are rare (and it will always be USB devices, often of the asynchronous type).
  This message, which is annoying once it is known that you actually need the setting, can be avoided by removing a small file. Which file this is shows in the message itself.
  
  
• M2Tech HiFace support
  
  One could say this is a specialty, because ... it actually is. It behaves somewhat different from normal USB devices, and therefore needs special treatment, at least for Kernel Streaming (btw, with the latest 1.03 drivers for it this may go unnoticed for other players, but in here it needed an adjustment).
  
  To be clear and conclusive : all is supported for Engine#3 (WASAPI) as well as Engine#4 (KS), and the latter for Vista/W7 (W2008) as well as XP.
  
  Keep in mind that the 1.03 drivers are needed, and that the drivers for XP are different than for Vista et al. Also, it may occur to you that the drivers for XP won't install by means of the (with the driver) withgoing Setup program (it may hang forever), and that installing the driver by means of pointing at it via "Update driver" from Device Managament, just works.
  
  Note : The HiFace allows "DAC Needs 32 bits" as well as "DAC Needs 24 bits". Both may sound different, and the 24 version theoretically should sound better.
  
  One thing to be careful of : The HiFace (at least for XP) shows "Kernel Streaming" for the normal WDM driver. For XP this can *not* be chosen, because it is not meant for Kernel Streaming at all (this is a bug in the driver). Instead - and as always, for a Kernel Streaming device choose one preceeded by "KS:".
  
  Anyone who appreciates the given possibility to make this work, may send an email to boleary, who was so kind to send a HiFace to here, but which got lost on the trip.
  
  
• New checkbox "96 KHz or 192KHz only" in the Settings Area
  
  When ticked, this allows for upsampling to an uneven rate, like from 44.1KHz to 96KHz or 192KHz. This can only be done via the upsampling method "Anti Image", and is meant for DACs not supporting 88.2KHz and/or 192KHz.
  
  
• Auto switching of Sample Rates / bit depths within playlists for Engine#4
  
  Similarly to how this works for Engine#3 (except for one person out there ), tracks with mixed formats are allowed to be in the Playlist Area now, and during subsequent playback of those tracks, the format accordingly will be chosen. Of course "Allow Format Change" in the Settings Area must be ticked for this.
  
  
• Auto switching between Double-Quad and native hires material for both Engine#3/#4
  
  New for indeed both Engines is the feature of Double or Quad being ticked, while a native hires track passes by, and next Double/Quad is ignored. Thus, before when Double/Quad was used and a hires track was in the Playlist Area together with redbook tracks, a message would tell you that Double/Quad is not allowed on hires material when the hires track was to play. Now the Double/Quad function is shut off internally for such a track.
  
  
• Engine Selection now goes via Device Selection
  
  Instead of explicitly choosing the Sound Engine, now this is done implicitly - and only like that - by means of choosing the Sound Device;
  A device preceeded with "KS:" will imply Engine#4 and a device not having that implies Engine#3.
  
  
• A new checkbox "Peakxt" in the Main Area
  
  When ticked - and Arc Precition is the upsampling method, it will imply Peak Extension as known from HDCD. This was active by standard in the before versions. Now, when unticking this checkbox, Peak Extension will not be applied and the main (and actually only) reason for it is to gain back on the -3dB applied otherwise. So, this is for those having just not enough juice for a soft sounding album (where wonder oh wonder Peak Extension won't do a thing because it won't clip !).
  
  
• With Volume Normalization active, it could happen that the volume couldn't be notched up by means of the small "notch" buttons. This has been solved now.
  
  
• When playback initiated, most of the time briefly the Wallpaper was shown from the before playback session. This is solved now.
  
  
• Because of a bug it could happen that a Kernel Streaming device didn't show, or let XXEngine3 error at the startup of XXHighEnd (when XXEngine3 is called briefly). This could happen right afther the new install of a device, while at a next install the error disappeared. Solved now.
  
  
• Because of a nasty bug, the versions containing Arc Prediction did not apply Anti Image upsampling when asked for. It just didn't do a thing.
  Nice placebo, but unintentional of course. Solved now.
  
  
• The versions with Kernel Streaming did not allow Alt-n for Next Track anymore. This is now solved.
  Notice that it still can happen that Alt-n restarts the same track (not meant like that of course).
  
  
• The Normalized Volume only works for redbook material; When a hires track was played with the Normalized Volume active, a message "Header no good" appeared. This message has been removed now (for that situation) and playback continues without Normalized Volume for that track.
  
  

The usual repeat from earlier Release Notes :

Very important : Only when AutoHotkey has been installed, and is at least working to the extend XXHighEnd can be startup with a hotkey (default is Alt-X), it is allowed to tick the checkbox "Stop Services" in the Settings Area, and use Unattended Playback mode (checkbox at the bottom of the main screen). For AutoHotkey See AutoHotkey Install. When this is not done, hence XXHighEnd can not be brought up with a hotkey, at Unattended Payback - no desktop icons will be visible, no Taskbar will be there, no Startbutton will be around ... nothing. You might, however, may be able to use the "Windows" key which will bring up the Start Menu and find something in there to find your ways again. So, be careful, or maybe you will not be able to do anything, except for listening out the playlist concerned, and wait until that finishes and XXHighEnd will popup automatically, which by itself will bring back everything.
A reboot is a solution as well.

This all sounds more dangerously than it really is, and with the hotkeys available there is really nothing to fear.
And remember the objective : better Sound Quality.