To Bob

[Carlos Mills]

Hi Bob,

I am writing an article about SS for an Audio Magazine here in Brazil. I would like to write about the summing bus abilities of SS (and also mention SAC). Could you elaborate something about it? ;)

Thanks in advance!

[Iain Westland]

bloody good ~ will that do:D

[Bob L]

Carl,

Lots of this info spread thru-out the forum... a search for "summing bus" brings them up.

Here are a few I found.

The mix bus is held internally at 32 bits wide creating 8 bits of headroom above the highest audio bit depth available in the typical hardware (24)... while maintaining a full 64 bit width register to handle multiplies and divides during the internal calculation process... there's no where to go much beyond that at the moment... and certainly not that makes any sense to try to improve the sonic quality.

The current SAWStudio internal summing engine can mix 128 tracks of audio at just below full clipping without clipping internally.
SAWStudio uses integer math, instead of floating point math for its internal processing in everything but the eq algorithms (which makes minimal use of hand coded assembly floating point)... the summing bus is always a full 32 bit DWORD wide, with all multiplies handled using the full 64 bit internal register width of the EDX/EAX pairs.

The internal data is stored at 24 bits always... even 16 bit data is shifted up to 24 bits with zeros in the bottom bits to start out with.
From that point on, as things are processed and reverb is added and data is mixed together, the lower bits start to develop into real data, no longer remaining zeros.

All internal math is processed using 64 bit integer multiplies and divides... this actually takes things beyond many dsp chip processing capabilities. The eq uses certain algorithims processed in floating point, all of which are still written and handled in assembly language.

The internal summing bus stores data in 32 bit DWORDS, which allows for an internal summing of 128 tracks at full digital clipping without overload.

Bob L

[Bill Park]

Carlos,

You could hear the results of suming in both audio programs and in analog setups via Lynn Fustons DAW SUM project, at 3daudioinc.com. I did the SAW part of the test, and I did it without speakers... I followed the instructions 'blind' so that there could be no thought of cheating.

Bill

[Carlos Mills]

Thanks Bob,

This is the info I was looking for.

Cheers!


Carl,

Lots of this info spread thru-out the forum... a search for "summing bus" brings them up.

Here are a few I found.

The mix bus is held internally at 32 bits wide creating 8 bits of headroom above the highest audio bit depth available in the typical hardware (24)... while maintaining a full 64 bit width register to handle multiplies and divides during the internal calculation process... there's no where to go much beyond that at the moment... and certainly not that makes any sense to try to improve the sonic quality.

The current SAWStudio internal summing engine can mix 128 tracks of audio at just below full clipping without clipping internally.
SAWStudio uses integer math, instead of floating point math for its internal processing in everything but the eq algorithms (which makes minimal use of hand coded assembly floating point)... the summing bus is always a full 32 bit DWORD wide, with all multiplies handled using the full 64 bit internal register width of the EDX/EAX pairs.

The internal data is stored at 24 bits always... even 16 bit data is shifted up to 24 bits with zeros in the bottom bits to start out with.
From that point on, as things are processed and reverb is added and data is mixed together, the lower bits start to develop into real data, no longer remaining zeros.

All internal math is processed using 64 bit integer multiplies and divides... this actually takes things beyond many dsp chip processing capabilities. The eq uses certain algorithims processed in floating point, all of which are still written and handled in assembly language.

The internal summing bus stores data in 32 bit DWORDS, which allows for an internal summing of 128 tracks at full digital clipping without overload.

Bob L

[Carlos Mills]

Thanks Bill,

I will check this out.

Cheers!


Carlos,

You could hear the results of suming in both audio programs and in analog setups via Lynn Fustons DAW SUM project, at 3daudioinc.com. I did the SAW part of the test, and I did it without speakers... I followed the instructions 'blind' so that there could be no thought of cheating.

Bill

[Grekim]

Carlos,

You could hear the results of suming in both audio programs and in analog setups via Lynn Fustons DAW SUM project, at 3daudioinc.com. I did the SAW part of the test, and I did it without speakers... I followed the instructions 'blind' so that there could be no thought of cheating.

Bill

Hey Bill, it might be note worthy to say that the sum test was done with an older version of the engine. I think we can assume it was improved still with SAWStudio 4?
I think it bears repeating that to test one variable at a time like summing is the scientific way to go about things. But in reality, when you factor in all the variables of a real mix (plug-ins, pans, gain stages, bussing, dithering, etc), the sonic differences between DAW's can become quite obvious. So my point is I doubt anyone will go "wow" when comparing summing alone. But, I had an eye opening experience when my first SAW mix was finished and I tried in vain to come close with the other DAW's at my disposal.

[Carlos Mills]

I am curious,

Does anybody know how other DAWs do regarding this?

"The internal summing bus stores data in 32 bit DWORDS, which allows for an internal summing of 128 tracks at full digital clipping without overload."

TIA for your information.

[Grekim]

If you had exactly duplicated the settings from your SS mix into another DAW, you would probably have ended up with identical results.

The result of the infamous DAW SUM test (amongst the participants), was that the sonic differences in DAW mix buses were negligible, and that it was the feature set(s) and work flow that determined the effectiveness of one DAW over another, i.e., personal taste....and personal taste makes you enjoy what you are doing better....and therefore you get better results.

Mark

Yes, there is definitely something about the results you can get when working "in your element." I think I'll always prefer playing with analog mixers and gear, but there's no denying that SAW can do as good a job or better. I'm just don't like staring at a computer all day. Getting up out of your seat once in awhile and plugging stuff in behind your rack is good for the old back.

[Carlos Mills]

Hi Mark,

Thanks for the link, but the white paper you refer to shows a 48 bit Mixer (Interesting paper by the way).
Does anybody have something with 32 bits?



http://akmedia.digidesign.com/support/docs/48_Bit_Mixer_26688.pdf

[DominicPerry]

Does that mean that PT is 48bit summing but that information passed to plug-ins (other than the mixer itself) is always 24bit?

Dominic

[Grekim]

Hi Mark,

Thanks for the link, but the white paper you refer to shows a 48 bit Mixer (Interesting paper by the way).
Does anybody have something with 32 bits?

I believe it's actually a 56 bit accumulator when you consider the headroom. The bottleneck is at the end of each plug-in where you only retain 24 bits. The common DAW's (that run without proprietary hardware) are passing 32 bit floating point data between plug-ins. Because it's floating point, your precision is only about the same as a 24 bit fixed point system. So right there is one obvious strength of SAW, with 8 bits more precision because of being fixed point. Again with the everyday DAW's, your summing can be adversely affected by how numbers get shifted and truncated when you add floating point data.