Tape was never accurate. Thats why a second gen tape did not sound as good as the first. Digital can make unlimited copies without degradation where tape can't.
All the little iron dust particles that made up tape was as variable as water.  It got fluffed off with each pass plus you didn't even have to use for it go bad.  Tape would become brittle and separate the substrate from the binder. There was no real archiving because each time you made a copy you lost data. Than if the tape recorder was just slightly out of alignment all hell could result. Now we have people touting the sound of tape and they forget that it was darn hard to get a really good recording. 
 
Have you ever looked at a video tape and compared it to a digital HD Blue ray disk? If you think about it that difference is what audio tape offers in comparison to digital audio on a hard disk.  
 
Tape was and is an awful medium for audio or video. But it was all we had. 
 
I should stop now because it really upsets me when a reference to tape is made without a curse also applied. 
 
And don't get me started on digital tape either. LOL

Hi, Beep.  I find this subject to be very interesting.  Since the Console Emulator was introduced in X2, I've been using it on every project, and I have fallen in love with it because I perceive it to add depth to my mixes and has an analog vibe that was lacking.  I'm grateful that CW quickly found that turning off the 64 bit engine allowed users to use X3c with their CE's.  However, I noticed since then that the moderators have been largely silent on this issue, especially since it conflicts with their advertising philosophy over the last 5 years.  If they come on here and say that it doesn't make any difference in the overall sound of a project, then they negate their advertising.
 
I've heard nothing but OPINIONS since then, but nothing that I would really call concrete evidence.  I've been asked to do blindfold listening tests to determine whether or not the 64 bit engine is really necessary, but that's not persuasive to me because I don't have "golden ears" nor the world's best monitor system.  So, my judgement is based on how stable and solid the processing is overall.  Right now, I'm rock solid with X3b using the 64 bit engine, the Tape Emulator, and Console Emulator, but not with X3c.  So, I'm hoping that CW issues a patch before they come out with X3d, and soon!  They've done this in the past, and I'm hoping they can do it again.

From Cake's website:

SONAR Core Technology and 64 bit Double Precision Engine

There’s a reason SONAR just sounds better. SONAR's industry-first, end-to-end, 64-bit double precision floating point mix engine allows you to mix with sonic clarity using a suite of versatile effects, powerful mixing tools, and endless routing possibilities. The 64-bit audio mix engine means you are mixing with the best audio quality in the industry today. The 64-bit engine is featured in SONAR X1 Studio and Producer only.

Best of all, you can take advantage of this audio engine on both 32-bit and 64-bit PCs giving you unmatched audio quality with spacious amounts of headroom and footroom through extended dynamic range. SONAR X1 also features native support for 64-bit floating point audio files, allowing you to import, stream, and render tracks and mixes at the highest quality available in the industry.

The 64-bit Double Precision Engine provides greater resolution, meaning more accurate audio reproduction and more headroom. You'll especially notice the benefits of the 64-bit Double Precision Engine when working in large projects containing many audio tracks and plugs-ins. Your chances of clipping will be significantly reduced. 

The 64-bit signal path includes plug-ins and buses. SONAR sends and receives 64-bit data to and from all plug-ins that accept 64-bit data. Combine this 64-bit data chain with SONAR’s line of Linear Phase accurate audio effects and you have true a mastering environment. If a plug-in requires 32-bit data, SONAR will automatically handle the conversation. 

Partial list of SONAR’s 64-bit effects and instruments:

• ProChannel
• LP-64 EQ Linear Phase Mastering EQ
• LP-64 Multiband Linear Phase Mastering Compressor
• TS-64 Transient Shaper
• TL-64 Tube Leveler
• VC-64 Vintage Channel
• Roland V-Vocal 1.5
• Dimension Pro
• Rapture LE
• Session Drummer 3

I still think having a performance improvement by playing Grand Theft Auto 5 at the same time as recording is a bit of a scoop....

@stevec... Yeah, I don't see any spikes in my perf module so I assume it's okay but who knows. For stuff like that I just try to go with defaults or highest quality until it causes a problem. Does it help? I have no friggen' clue but that's what makes me wonder and I guess as I get all the other more rudimentary questions about audio out of the way my mind wanders to things like this. It truly is a touch of insanity but at least it's putting my obsessive nature to something useful, yanno? ;-)
 
@John... Fair enough. The only real first hand experience I have with producing with tape was years ago (I was like 13 or some crap) when me and my much older bandmates would rent one of those Fostex 4 track dealies that recorded to cassette. Tons of ping ponging, minimal (and crappy) effects, exporting to tape decks, endlessly scouring the manual to figure out what the heck we were doing, etc... and even then the drummer was dealing with most of it and the results weren't all that great. Otherwise it was all in fancy studios getting recorded to 2 inch or ADAT with real producers who were doing all the work. I never really understood it, still don't and at this point don't really care because it seems to be a dead art. I know guys who are still obsessed with tape and I figured there was a reason for it but now I realize that the digital stuff likely surpasses it all which was kind of my point. Forget any hardware benefits of tape consoles (which seemingly can be replaced by digital plugins anyway for FAR less cost and hassle)... at what point did digital exceed analog? 16 bit? 24 bit? 32 bit? And how far can we go with it? The human ear is a limited physical entity but how limited? I doubt we've even come close to exceeded what the human mind can process but the human ear seems to be the choke point... or is it? We are an evolving species. Maybe with the more accurate recordings of the modern era and the fact that everyone seems to have access to them maybe in a couple thousand years our hearing will be far more accurate through the process of evolution. Maybe then the difference between 32 and 64 bit will be very noticeable. Not that anyone will likely be listening to my crummy metal tunes a thousand years from now but music being produced today that ends up being considered culturally and historically significant will be and the higher bit depths will matter. Just like the Robert Johnson comparison I made earlier (I'm on the RJ kick because I've been listening to a bunch of that stuff the past few days). Perhaps if the engineers of those recordings realized the impact they would have they would have opted for the most advanced tech of the time.
 
Again really it's nuthin' to do with nuthin'. Just mumbling to myself as I let a bunch of other concepts I've been studying settle into the long term memory of my brain and plotting my next moves for the project I'm currently working (which I should probably get my arse in gear on today... lol). Rambling about inanities helps me think. This thread just happened to be the victim of today's chaotic Beepisms.
 
;-p

stevec
FWIW, normally there's no harm in leaving it enabled, other than perhaps a slight bit of resource overhead.  The big question is, and always has been, does it make any kind of audible difference whatsoever?    Good luck getting a definitive answer on that one.      Which is why I think John is essentially saying "don't worry about it" - that extra precision isn't going to be as audible as a 60Hz hum!
 
One thing to keep in mind with this preference is that it applies to real-time playback.   When you export audio you also have the option to enable or disable 64bit double precision, regardless of how the preference is set.   I personally believe that if it has any relevance at all it'll be here and not during playback, since exports are what the rest of the world will hear.  So if the project I'm exporting is something that I think might benefit from that extra precision in any fashion, right or wrong, then I'll enable it during export.   Why not?
 
 

Excellent point Steve about the export. Good post overall too. 
 
As to what will be used for audio in the future. When quantum computing is in every home this whole discussion will be moot. 50 quantum bits is greater in storage capacity than that of the entire universe.      

Lynn
I've heard nothing but OPINIONS since then, but nothing that I would really call concrete evidence.  I've been asked to do blindfold listening tests to determine whether or not the 64 bit engine is really necessary, but that's not persuasive to me because I don't have "golden ears" nor the world's best monitor system.

 
Oh I see. When I did an export both ways in your other thread and posted the measurements showing most of the 32bit errors often don't accumulate to the point that they will even make it into 24bit output (much less be audible), you consider that an "opinion"?
 
It's very simple:
 
1. If we go through the math we can see that the errors won't accumulate to the point of being audible.
 
2. If we take quantitative measurements, we will find that the errors are not at a level that would be audible.
 
3. If we do double blind testing in a controlled test, we will find that the errors are not audible.
 
What would you consider "concrete evidence"? If you insist that everyone including the "true believers" agree that it isn't audible, then I'm afraid you're going to be out of luck. 
 
The only reason this is controversial is that people would rather argue than do controlled tests. And if I do the tests (as I have done), people who don't like the results just ignore them. And most of the time, the descriptions of what people claim to hear doesn't much match what the errors would sound like if they were audible anyway.

I believe you Drew. I'm not smart enough to test that stuff myself nor really understand it but it does make sense to me that perhaps we've moved past the point where such precise calculations really don't matter to our limited ears. I swear I can hear a difference between 16bit and 24 but beyond that.... I don't think so. Even at those lower depths I may be imagining it anyway. I was just being silly with my ramblings about us evolving Ferengi ears.
 
Perhaps if I ever get a dog he'll appreciate the difference in my bit depth selection. Then again subjecting animals to my noise might be considered cruel. ;-)

drewfx1

Lynn
I've heard nothing but OPINIONS since then, but nothing that I would really call concrete evidence.  I've been asked to do blindfold listening tests to determine whether or not the 64 bit engine is really necessary, but that's not persuasive to me because I don't have "golden ears" nor the world's best monitor system.

 
Oh I see. When I did an export both ways in your other thread and posted the measurements showing most of the 32bit errors often don't accumulate to the point that they will even make it into 24bit output (much less be audible), you consider that an "opinion"?
 
It's very simple:
 
1. If we go through the math we can see that the errors won't accumulate to the point of being audible.
 
2. If we take quantitative measurements, we will find that the errors are not at a level that would be audible.
 
3. If we do double blind testing in a controlled test, we will find that the errors are not audible.
 
What would you consider "concrete evidence"? If you insist that everyone including the "true believers" agree that it isn't audible, then I'm afraid you're going to be out of luck. 
 
The only reason this is controversial is that people would rather argue than do controlled tests. And if I do the tests (as I have done), people who don't like the results just ignore them. And most of the time, the descriptions of what people claim to hear doesn't much match what the errors would sound like if they were audible anyway.

Drew, I appreciated the tests that you did on my thread.  That's the only evidence that I've seen that seems credible.  Now, I, personally, don't want to spend time doing my own tests because time is limited, and I'd rather make music.  I'll take your word for it.  Your test seems to go against CW's own words (per amiller's post above), and that is where I get confused.  But, don't think that I don't value your time and input.  If CW is slow to release a patch or an update, then I'll likely convert to X3c just to see how it handles.  If it turns out that there is no audible difference nor hangups in processing, then you may very well have debunked CW's theory and advertising campaign for the last 5 years.  Again, I've not seen anybody from CW debate this.
 
P.S.- I think that I may be chagrined with myself for spending a lot of money to upgrade my equipment for 64 bits when it may not have been necessary.  Nobody likes to feel "had".

Lynn
If it turns out that there is no audible difference nor hangups in processing, then you may very well have debunked CW's theory and advertising campaign for the last 5 years.  Again, I've not seen anybody from CW debate this.

They tend to be fairly careful in their wording:
 

...unmatched audio quality with spacious amounts of headroom and footroom through extended dynamic range
 
The 64-bit Double Precision Engine provides greater resolution, meaning more accurate audio reproduction and more headroom.

 
This is all absolutely true, but note that there is no mention of audibility or the fact that 32 bit single precision already provide hundreds of dB's of excess headroom and dynamic range.
 
 

You'll especially notice the benefits of the 64-bit Double Precision Engine when working in large projects containing many audio tracks and plugs-ins.

 
Um, how large? How many tracks? Theoretically if you have enough tracks the errors will accumulate, but they accumulate very slowly. Oh, and I should mention that the errors accumulate on a per sample basis (i.e. not over time) and any tracks that are muted or silent for a given sample don't count. And if you had a LOT (and when I say "a LOT", I mean more than you are likely imagining) of tracks sounding all at once, how likely are you to be able to pick out tiny artifacts that are much quieter than your actual audio anyway?
 

Your chances of clipping will be significantly reduced. 

 
This is hysterical, because though technically true, you have no remotely realistic chance of clipping to begin with - the amount of headroom with 32bit float is already unfathomable from an audio perspective. I think the fact they they even make this claim speaks for itself.