[Solved] Exported Mix drastically different than what I hear - Sonar audio engine bug?

Hello friends,
during my more than 10-year old various audio works, this is the first time I ran into such incredible problem, and I'm pretty much beyond being pissed off from it, let me write as accurately as possible what's the problem:
 
What I hear is drastically different than what is exported, most prominent is completely missing kick drum (about 25% of the song overall) which is not just audible, but even clearly visible on waveform (comparing Sonar exported "entire mix" and "live recorded, what I hear" output via RME Totalmix zero-latency Loopback into SoundForge),
 
"background" info: decent project of about 26 tracks (each of few takes) recorded on one machine at "humble 44100/16), being mixed on another machine (both rigs have same Sonar version, just the first one doesn't have the plugins since it's for pure recording), 9 of them are drums, kick drum recorded with 2 mics snare with 2 mics, saxophone with 3 mics, guitar with 2 mics etc., every "instrument" - all the "instruments" have own mixdown busses,
drums are manually slightly sorted with AudioSnap (phase coherence between all the drum tracks is maintained), practically all clips in the project are cut/adjusted,
some tracks have volume/pan automation applied, many tracks have slight effects, but majority of FX stuff is done on the busses (some Waves, iZotope, , AIR Music and DMG Audio plugins, but not more than few per bus),
 
at export, all things are ticked, 44100/16 (so absolutely no sample rate or bit depth recalculation)
interesting thing is that exported result seems to be about equally bad regardless if it's fast or slow bounce, which makes me think that plugins are not the cause of the problem;
what irritates me most - even the bounce of the individual tracks visually results in difference, which leads me to (probably wrong but who knows) conclusion that Sonar (exporting) audio engine is far less accurate (to process Audio Snap especially) than my "realtime" RME Hammerfall DSP - which appears really strange to me, since the PC specs are more-than sufficient (i7-5820K 12 cpu threads, 32GB ram in quad-channel, 500GB 850EVO SSD for system, 2TB Seagate HDDs for storage and up to 20GB (12GB/s I/O) RAMdisk for caching+temps, RME HDSPe AIO PCIex audio interface)
 
thanks in advance, any help appreciated
Tom

it's my custom preset, which has ticked same things as "what you hear", plus with ticked "64-bit Engine" and format is "FLAC"

I use the same audio interface but never experienced this issue.
 
If the difference would be only the reading of TotalMix meters and SONAR meters, that could be explained by a setting in TotalMix preferences:
RMS +3dB
 
This setting assures that full scale level is identical for Peak and RMS at 0 dBFS which is different from SONAR's RMS peak level (RMS -3dB).
 
But you said your exported mix is drastically different, so it must be a different (export or routing) setting which causes what you hear.
 
 

To rule out plugins more definitively have you tried bouncing after bypassing all plugins or removing them?
Check if all the audio renders properly then. This isn't related to your PC or audio interface since bounce doesn't care about those specs. 
If you can share a version of your project that exhibits this problem we can certainly take a look at it. Have you observed this on other projects or is it specific to only this one?

This sounds like it could be related to AudioSnap. There are separate playback and rendering options. With one of the tracks selected click on the Track button along the top of the Inspector (Usually docked left). There is a online (Playback) and Offline (Bounced) algorithm. Typically the Online is Groove and the Offline is Radius.
 
You can test this by setting the offline to "Same as Online". Afterwards Freeze that track or create a new bounce of it.
Depending on the results the offline algorithm may be creating unwanted artifacts. Different types of sources require different settings here so it's good to get use to trying different ones out as you get a feel for when & where to use each one. If it sounds good while mixing you may be able to keep this setting of "Same as Online" your exported mix should then sound the same.

Any update on this?

But the offline render should have sounded better - unless you like lofi stretching :)