As mudgel says, the round trip latency in samples, milliseconds, whatever is what you measure it as by using a cable to loop interface output to input, sending a signal out, recording the return and looking at the time line for the difference. Some manufacturer's drivers don't report latency figures correctly so Sonar's readings in preferences can be wrong.
 
The real question though isn't "what's the lowest latency I can get" but "does this amount of latency cause me a problem or not?" Figures of 2ms or 3ms are so close together the difference is really irrelevant. One thing's certain though, things will run much smoother and less chance of crackles or dropouts with a 64 sample buffer than a 32.

Perhaps in my original post I should have withheld the exposition and simply asked:
 
Does a positive manual offset figure ADD TO or SUBTRACT FROM your total latency?
 
 
That might have spared a bit of confusion here... (Unless that too is a muddled question.)

dimelives1
 
Thanks, John. But it was my understanding that that given figure from SONAR may not be entirely trustworthy, and/or that it may not account for the other factors that the ASIO reported latency appears to (A/D conversion latency, etc.).
 
And if it is accurate, is that telling you the round-trip latency, or just one way?
 

The figures are not entirely trustworthy regardless of the driver mode or there would be no need for a manual offset. Unlike other driver modes ASIO numbers do include more that just the driver buffers. The other driver modes also have to deal with the same h/w buffers but just do not report anything other than the driver settings.

dimelives1
Perhaps in my original post I should have withheld the exposition and simply asked:
 
Does a positive manual offset figure ADD TO or SUBTRACT FROM your total latency?
 
 
That might have spared a bit of confusion here... (Unless that too is a muddled question.)

The value of the offset is added to total latency. The value may be a positive number which would increase RTL or a negative number which would decrease RTL.

dimelives1
Perhaps in my original post I should have withheld the exposition and simply asked:
 
Does a positive manual offset figure ADD TO or SUBTRACT FROM your total latency?
 
 
That might have spared a bit of confusion here... (Unless that too is a muddled question.)

It doesn't add to or subtract from your latency.  Your latency is what it is.  The manual offset compensates for latency

BobF

dimelives1
Perhaps in my original post I should have withheld the exposition and simply asked:
 
Does a positive manual offset figure ADD TO or SUBTRACT FROM your total latency?
  
That might have spared a bit of confusion here... (Unless that too is a muddled question.)

It doesn't add to or subtract from your latency.  Your latency is what it is.  The manual offset compensates for latency

That's right. The need for a Manual Offset is an indication that the actual input/record latency is being mis-reported to SONAR by the driver. The reported value plus the empirically determined Manual Offset is the actual latency.
 
Thus, the actual values are:
(132+2)/44.1 = 3.04ms
(96+46)/48 = 2.96ms
 
Clearly the difference is not worth being concerned about, except that your CPU will be happier running the larger 132-sample buffer at the lower rate. It's pretty unusual for an interface to have such vastly different "hidden" latencies at different sample rates; I'm not sure what that's about unless possibly there was an error in your offset measurement procedure.
 
A much quicker way of finding the correct Manual Offset is to measure the actual round-trip latency using the free CEntrance ASIO Latency Test Utility (https://centrance.com/downloads/ltu/).
 
Then:
Manual Offset = CEntrance-Measured RTL - SONAR-Reported RTL
 
 

Thanks a lot, guys. The last few posts answered what I was wondering about.
 
 
@ mudgel, BobF, & brundlefly:  Just out of curiosity, why do you prefer working at 48k instead of 44.1?
 

The sample rate discussion is one that I suspect will never end.  After much reading back/forth on the subject, I found a couple of articles by folks that know this stuff.  I'm sure there are a bazillion yeah-buts out there, but I describe my decision in this thread:
 
http://forum.cakewalk.com/FindPost/3121040

dimelives1
Thanks a lot, guys. The last few posts answered what I was wondering about.
 
 
@ mudgel, BobF, & brundlefly:  Just out of curiosity, why do you prefer working at 48k instead of 44.1?
 

When I first started using 48khz it was because my efforts were heading to DVD. It kind of stuck, at one time thinking that the higher sampling rate would result in a better final sound. These day It's more of a habit. Despite all the reading I've done there are so many differing opinions in what I thought should be a scientific black and white subject.
Apparently that's not the case.
I'm not absolutely sure what my own listening experience tells me. Too many variables and too subjective. I don't think that 48 "sounds" better than 44.1 but I do "feel" that when working with acoustic/mic'd instruments including vocals that 96khz recordings end up as more "spacious" finished recordings even if they're mixed down to 44.1.
 
c'est la vie

mudgel

dimelives1
Thanks a lot, guys. The last few posts answered what I was wondering about.
 
 
@ mudgel, BobF, & brundlefly:  Just out of curiosity, why do you prefer working at 48k instead of 44.1?
 

When I first started using 48khz it was because my efforts were heading to DVD. It kind of stuck, at one time thinking that the higher sampling rate would result in a better final sound. These day It's more of a habit. Despite all the reading I've done there are so many differing opinions in what I thought should be a scientific black and white subject.
Apparently that's not the case.
I'm not absolutely sure what my own listening experience tells me. Too many variables and too subjective. I don't think that 48 "sounds" better than 44.1 but I do "feel" that when working with acoustic/mic'd instruments including vocals that 96khz recordings end up as more "spacious" finished recordings even if they're mixed down to 44.1.
 
c'est la vie

yeah, the discussions on sample rate usually span too many pages to follow through - 189 replies when Craig raised the question last time ...
 
anyway, for about a year now we have changed our sample rate up to 96 kHz because we have found audible evidence (for us!) that it sounds better (to us!) ... totally subjective I know ... it's only one particular guitarist with high-end gear who spends aeons on finding the right sound and records directly, so 96 kHz really does him justice ... other projects I've done in the same period it would not have made a difference but I also recorded at 96kHz because the system can easily handle it and changing samples rates between projects/sessions is too much of a hassle ;-) of course, it's a bit of a disadvantage when doing long distance collaborations (due to really large files), but that we handle by using mp3 in pre-production and accept longer downloads for production files ...
 
try it, see if your system can handle it for large projects, and use what makes you feel comfortable
 
BTW, some synths/samplers may sound quite a bit different when changing between 44.1 kHz and 96 kHz (depends on synth internal and samples) ... and when going after industrial style synth sounds you better stick with 44.1, 96 khz may make sound more open and transparent and reduce some of the industrial flair ...