64 bit

[mattbeat]

Hey,

i wonder are there many workstations supporting 64 bit processing already?
i mean not just 64 bit os support , but really 64 bit audio proccessing.
mastering with a 64 bit resolution would be great.
i think the new wavelab also works with 64 bit floating point.
does anybody use 64 bit applications for his projects?

[Cellotron]

Originally Posted by mattbeat

Hey,

i wonder are there many workstations supporting 64 bit processing already?
i mean not just 64 bit os support , but really 64 bit audio proccessing.
mastering with a 64 bit resolution would be great.
i think the new wavelab also works with 64 bit floating point.
does anybody use 64 bit applications for his projects?

SAWStudio (the DAW app I use for mastering) has used 64bit double precision fixed point math for its internal processing since its introduction 9 years ago. Welcome to RML Labs - The Makers of SAWStudio

Best regards,
Steve Berson

[Jerry Tubb]

Logic Pro and MainStage 64-bit mode FAQ

just experimenting with it at this point.

JT

[kjg]

check out Reaper as well.

[Table Of Tone]

AudioCube here!
Modified WaveLab platform with all processing at 64 bit floating point.

[Adam Dempsey]

Sonic Solutions' NoNoise was utilizing 80bit dsp 16 yrs ago.

I'm curious as to sonic improvements of the use of 64 bit in digital mixes that come in, if I could compare the same made via 32 bit float and 48 bit fixed.

[Jerry Tubb]

Originally Posted by Adam Dempsey

I'm curious as to sonic improvements of the use of 64 bit in digital mixes that come in,

I don't really think so, unless it has to do with improved performance of using 64-bit plug-ins, and having more memory for the application.

Perhaps more so for composing & mixing than mastering.

Here's a couple of blurbs from the Logic 9.1 support site:

What advantages are there to using 64-bit mode?

The main advantage is that you can address vast amounts of memory. With 64-bit mode, the application memory is not limited to 4GB as with 32-bit applications, so there is essentially no practical limit by today's standards. As a result, all the installed memory that is not needed by the OS is available for use by Logic Pro or MainStage and all the included plug-ins, which can be meaningful if your Mac has more than 4GB of memory installed. This larger amount of memory allows you to run far more instances of memory intensive plug-ins, such as sample-based software instruments.

Do Logic Pro or MainStage perform better in 64-bit mode?

Logic's processing has always been highly optimized, so in most cases there are only nominal performance improvements when running in 64-bit mode, other than the ability to run more instances of third-party memory intensive plug-ins. However, since Mac OS X v10.6 Snow Leopard is highly optimized for 64-bit operation, there may be some performance gains in areas where Logic interacts with it.

Do Logic Pro or MainStage sound different in 64-bit mode?

There is no difference in sound quality between running in 32-bit or 64-bit mode. Prior versions of both applications have already used 64-bit processing resolution for plug-ins where it was felt there could be an audible benefit. In either mode, both Logic Pro and MainStage offer huge dynamic range and extremely high audio quality.

I'm just in the initial test stages with it : - )

Cheers - JT

[Adam Dempsey]

Originally Posted by Jerry Tubb

I don't really think so, unless it has to do with improved performance of using 64-bit plug-ins, and having more memory for the application.

Perhaps more so for composing & mixing than mastering.

Do Logic Pro or MainStage sound different in 64-bit mode?

There is no difference in sound quality between running in 32-bit or 64-bit mode. Prior versions of both applications have already used 64-bit processing resolution for plug-ins where it was felt there could be an audible benefit. In either mode, both Logic Pro and MainStage offer huge dynamic range and extremely high audio quality.

Exactly...

Originally Posted by mattbeat

i mean not just 64 bit os support , but really 64 bit audio processing.
mastering with a 64 bit resolution would be great.

(though I see the benefit of high bit dsp for intensive restoration processors)

[Cellotron]

As always with bit depth's it's important differentiate what parameter they are in fact describing in what context.

It could describe:

the OS's throughput- which has no effect on the audio quality - but can allow for faster processing speed or larger/denser sessions to be handled IF the application running on it has been optimized to take advantage of this

the application's throughput - which again has no effect on the audio quality - but would allow it to handle more/faster processing and/or larger/denser sessions if it's running on a similarly compatible OS.

the DAW's internal processing math - which is the amount of places that are allowed when something like a multiple or divide occurs - and can be further differentiated as fixed point (aka "integer"), floating point, and single or double precision.

a plugin's internal processing math - which can in fact be different and even larger (as long as the host app allows for this) than what the DAW does during its own native processing.

the bit depth of returned figures after an internal processing calculation is done - after a calculation is made generally the returned figure is not at the larger bit depth size that are done for the internal processing calculations but at some easier to handle size before it is passed on to the next signal process. For many applications this is a 32-bit floating point IEEE file (which is in fact a 24bit + 8bit mantissa) or a 24bit fixed point file. For the app I use it's a full 32bit fixed point file.

the bit depth of a source or rendered file

Best regards,
Steve Berson

[Key]

You guys are talking about 2 different things.

64-bit Native Applications - which will run under an x64 OS like Vista 64,Windows 7, or the unsupported XP 64. The main advantage for audio would be using hardware with more than 4 gigs of ram.

64-bit float dsp - which can be run on both x86 and x64 native apps.

There is no measurable fidelity gain from x64 currently because the apps are still limited to 64-bit float processing. If the app can go above 64-bit float processing (64-bit fixed point and higher) then maybe there will be some measurable benefit.

[mattbeat]

thx a lot to all ya. plug ins which work with double precision floating point format already work under 32 bit cpus? they just dont work as fast as they would work with a 64 bit os with a 64 bit daw is that right?

[Cellotron]

Originally Posted by mattbeat

thx a lot to all ya. plug ins which work with double precision floating point format already work under 32 bit cpus? they just dont work as fast as they would work with a 64 bit os with a 64 bit daw is that right?

Correct - developers can code plugins or applications to use 64bit internal processing math even if they are running on 32-bit OS's / cpu's.

Best regards,
Steve Berson

[Bob Olhsson]

My understanding is that modern floating point calculations use 80 bit math unless the programmer has disabled it.

Memory addressing is a totally separate issue. For audio, the only thing I'm aware of that 64 bit addressing might buy you is the ability to hold more samples in RAM. Even then a RAM disk might work just as well.

The next generation of solid state storage is going to need to be addressed directly which is why computer manufacturers are promoting a transition to 64 bit memory addressing.

[Cellotron]

Originally Posted by Bob Olhsson

My understanding is that modern floating point calculations use 80 bit math unless the programmer has disabled it.

afaik, based on conversations with a number of developers, this is not correct. Internal processing math depends on the algorithms and routines coded by the application developer, not through some default of the cpu or the OS - you can still run fixed point math routines using floating point cpu's after all - although the architecture of the OS and cpu can definitely influence what type of routines will run more efficiently and stably.

Best regards,
Steve Berson

[Bob Olhsson]

I've only conversed with one who told me the whole 32 vs. 64 processing argument is marketing B.S.

[synthoid]

The original floating point unit of the Intel x86 family operated in 80 bits. But even at that, it's a bit dicey to say something sweeping like "all math done in 80 bits", because in practice the machine had very few 80-bit registers and they were organized as a stack, so it was frequently necessary to truncate an intermediate result to 64 bits even when using this 80-bit unit.

Later, this unit was made largely obsolete by the MMX and later SSE, SSE2, SSE3 etc instruction set extensions to the x86 family. These are much (much) faster than the stack-based floating point unit, as they are capable of vector operations. On top of that they conform better with programming language standards. All modern compilers by default generate floating point code for these vector units. So it is unusual to find recently developed code that operates in 80 bits.

The SSE units also provide an extended precision type that has more than 80 bits, but this is seldom used and it is massive overkill for audio -- it might arise in some physics calculations where really extreme orders of magnitude are encountered for a single quantity, but it's quite unusual. Virtually all scientific computation and signal processing is done in single or double precision (if it's done in floating point).

-synthoid

[wado1942]

N'Track has the option of 32-bit or 64-bit float. Of course the majority of plugins out there are just 32-bit and that's where most of the processing happens so I'm not sure how much of a difference it makes.

[Sunbreak Music]

You guys need to check your math. 64 bit is twice as good as 32 bit.

[wado1942]

Sunbreak, you need to check YOUR math. The equasion is 2^64, not 1x64. 64-bit is mathematically 4.294967e+9 X the precision of 32-bit.

[Table Of Tone]

I don't actually think the new operating systems are any better sounding or even as good as some of the older, sorted OS's.

They gotta be selling ya something, right?

I've been playing around with Win7 X64 and X86 (no sonic difference between the two) and I still wouldn't use either of em for mastering right now!
They're just not ready yet!

The same goes for Snow Leopard.
The normal leopard still has spotlight and permissions issues.
The permissions actually get messed up by apple's own software update process!
It's only just solid enough to use for playback.

I'd rather go with Tiger on the macs and Win 2000 or XP (X86) on the Pee Cee's.

[Jerry Tubb]

Originally Posted by Table Of Tone

I don't actually think the new operating systems are any better sounding or even as good as some of the older, sorted OS's.

They gotta be selling ya something, right?
<snip>
I'd rather go with Tiger on the macs and Win 2000 or XP (X86) on the Pee Cee's.

I pretty much agree with you here Tone.

Although I like the kewl functionality of the new Mac OS(es) and DAWs...

They don't sound any better than Tiger!

Same goes for Windows XP Pro.

Staying on the Bleeding Edge is an addiction of sorts, when in reality it's basic audio quality and DAW stability that counts.

Look at the venerable Mr. DC, he's still running a vintage Sonic HD rig iirc.

Reminds me of a session about 12 years ago, I was running an early PPC 601 Mac, the young computer enthusiast client was amazed that I only had 128MB of RAM.

My reply was "it's more about audio quality than computer power".

Cheers - JT

[UnderTow]

Originally Posted by Cellotron

SAWStudio (the DAW app I use for mastering) has used 64bit double precision fixed point math for its internal processing since its introduction 9 years ago. Welcome to RML Labs - The Makers of SAWStudio

Best regards,
Steve Berson

I am still not convinced this is true. I still believe this has more to do with the accumulators in the processors in the same way that 32 bit float operations go to 80 bit accumulators. Saying a 32 bit float DAW is 80 bit isn't really true.

I read a lot of the comments on the the SAW forum and I found the developer deliberately vague about all this so I don't trust it. I could be entirely wrong of course.

Alistair

[Dubai]

Originally Posted by mattbeat

Hey,

i wonder are there many workstations supporting 64 bit processing already?
i mean not just 64 bit os support , but really 64 bit audio proccessing.
mastering with a 64 bit resolution would be great.
i think the new wavelab also works with 64 bit floating point.
does anybody use 64 bit applications for his projects?

Sonar 32-Bit Engine... vs. Sonar 64-Bit engine...
Mackie Tracktion 32Bit vs. 64-Bit...
dont know about Logic..
Cubase / Nuiendo are 32-Bit...
Protools 48-Bit.

in Sonar... the 32-Bit engine its designed to be an analog summing bus emulation

Protools & Cubase/Nuendo have the same analog summing bus emulation in MAC.
"verry suspicious", dont know if MAC sounds different than PC engine, but i guess it could happen.

Sonar 32-Bit its different, but also analog summing bus emulation, the most colored one.
Mackie Tracktion at 32-Bit also has an analog summing bus emulation, but more transparent than Sonar 32-bit engine.

the 64-Bit engine in Sonar its designed to be 100% transparent.
sounds like the Logic Engine.

the Mackie tracktion at 64-bit also becomes a bit more transparent than the 32-Bit engine but not as transparent as the Sonar engine.

Reaper dont know if its 32 or 64 but the summing bus its on the transparent side.
Magix Samplitude its on the analog summing bus emulation side also, but different than the other analog summing bus emulations.

Propellerheads record, also has analog summin bus.

dont know about SAW...

Harrison consoles software also has analog mix bus engine, but dont know if its 32 or 64.

Yet another DAW comparison Test

[Colin Leonard]

[QUOTE=Jerry Tubb;5074807]I don't really think so, unless it has to do with improved performance of using 64-bit plug-ins, and having more memory for the application.

agreed. we are running audiocube as well, which is a really great sounding workstation, but the 64bit audio engine does not come into play until 2 or more of the cube-tec plugs are used.

[alyricalmind]

Originally Posted by Table Of Tone

I don't actually think the new operating systems are any better sounding or even as good as some of the older, sorted OS's.

I'd rather go with Tiger on the macs and Win 2000 or XP (X86) on the Pee Cee's.

Word to that. I'm still using XP 32bit cuz I simply never felt the need to have more than the 3 and what odd GBs of RAM and it's stable as can be. I wouldn't downplay Vista or 7 though. I use them on my laptops and I have had no problem using Pro Tools 8. Most the software I'm using still runs on 32bit so I think I'll give it a couple more years for 64bit software to get more popular before I switch my OS.

[Table Of Tone]

[QUOTE=Colin Leonard;5083695]

Originally Posted by Jerry Tubb

I don't really think so, unless it has to do with improved performance of using 64-bit plug-ins, and having more memory for the application.

agreed. we are running audiocube as well, which is a really great sounding workstation, but the 64bit audio engine does not come into play until 2 or more of the cube-tec plugs are used.

2 or more plugs to engage the 64 bit audio engine?
I never knew that.

I rarely even use one plug!

AC is a great sounding workstation though!
I love the workflow of WaveLab as well.

I don't know if it's that that's scoring me better sounding masters or the fact that I'm running soundBlade for playback these days, instead of PT or Logic?

[Cellotron]

Originally Posted by UnderTow

I am still not convinced this is true. I still believe this has more to do with the accumulators in the processors in the same way that 32 bit float operations go to 80 bit accumulators. Saying a 32 bit float DAW is 80 bit isn't really true.

I read a lot of the comments on the the SAW forum and I found the developer deliberately vague about all this so I don't trust it. I could be entirely wrong of course.

Alistair

Allistair -
I've never seen Bob Lentini deliberately vague about this. In response to inquiries on his forum he's spelled out explicitly what it does in some posts, and in direct phone conversation with him he's also confirmed and clarified some points for me.

to wit:
* Multiples and divides are in fact done using 64bit double precision fixed point math.
This returns a full 32bit DWORD (truncated, not dithered, from the 64bit) before being handed off to the next signal process.
* The built in eq's use 64bit floating point math however. Originally fixed point math was used for these as well but then he found it was causing some DC offset in some operations with this method so he changed the algorithm for these.
* When a VST plugin is placed in line the 32bit integer figure has to be converted to 32bit float first, and then the plugin handles the math at whatever maximum allowable bit depth it has been coded at. The returned figure needs to be converted back again from floating point to integer.
* SAW native plugins from RML Labs all use fixed point math except for the Reverberator which uses 32bit floating point as it is legacy code created by another developer
* SAW native plugins from JMS Audioware afaik 64bit floating point math.
* SAW native plugins from Sonoris use 64bit double precision fixed point math (except for some filters which use 64bit floating point) and return a full 32bit DWORD with TPDF dither used when requantizing this from the 64bit figure.
* Unless the native dither is activated or a dither plugin is placed just prior to output then the 32bit fixed point figure will be truncated to 24bit prior to hand off to the DAC.

So I actually think Bob Lentini has been more forth coming about what's actually going "under the hood" with his DAW app than nearly any other developer with the exception of Digidesign, who went to the length of issuing a white paper.

Anyway - SAWStudio is definitely not for everyone as it commands are a little different than other DAW's, and it's price is in the higher range for DAW apps - but I've been using some version of it since 1994 so that I'm super quick on it, it's incredibly stable, the support is excellent, and do find that the results after processing is done with it remain very high quality.

An older article of definite interest to this thread discussing advantages of fixed point and floating point math is at ->
Fixed-Point vs. Floating-Point DSP for Superior Audio

Best regards,
Steve Berson

[alyricalmind]

Woh, I just went on their website and saw the price for sawstudio... from the screenshots it looks like a dated GUI. Why does it cost so much ($2500)? With that money I could buy a DIGI 003 Mixing Console with automated faders, good preamps, decent conversion and a whole lot more features... PLUS Pro Tools 8. Is there something I'm missing about this DAW?

[UnderTow]

Originally Posted by Cellotron

Allistair -
I've never seen Bob Lentini deliberately vague about this. In response to inquiries on his forum he's spelled out explicitly what it does in some posts, and in direct phone conversation with him he's also confirmed and clarified some points for me.

to wit:
* Multiples and divides are in fact done using 64bit double precision fixed point math.
This returns a full 32bit DWORD (truncated, not dithered, from the 64bit) before being handed off to the next signal process.

This is why I say he is being vague. If a multiply or divide operation returns a 32 bit word it is effectively a 32 bit operation. It is even truncated which means you lose any additional precision that might occur. (Not that there really is any). Anything in between is irrelevant. That is the resolution available and nothing more. This is exactly what I was alluding to before. To me this is similar to saying a 32 bit float calculation has 80 bit resolution when using the CPU float units. It is stretching the truth too far for my taste.

* The built in eq's use 64bit floating point math however. Originally fixed point math was used for these as well but then he found it was causing some DC offset in some operations with this method so he changed the algorithm for these.
* When a VST plugin is placed in line the 32bit integer figure has to be converted to 32bit float first, and then the plugin handles the math at whatever maximum allowable bit depth it has been coded at. The returned figure needs to be converted back again from floating point to integer.
* SAW native plugins from RML Labs all use fixed point math except for the Reverberator which uses 32bit floating point as it is legacy code created by another developer
* SAW native plugins from JMS Audioware afaik 64bit floating point math.
* SAW native plugins from Sonoris use 64bit double precision fixed point math (except for some filters which use 64bit floating point) and return a full 32bit DWORD with TPDF dither used when requantizing this from the 64bit figure.
* Unless the native dither is activated or a dither plugin is placed just prior to output then the 32bit fixed point figure will be truncated to 24bit prior to hand off to the DAC.

Thanks for clarifying all the details.

So I actually think Bob Lentini has been more forth coming about what's actually going "under the hood" with his DAW app than nearly any other developer with the exception of Digidesign, who went to the length of issuing a white paper.

Or Cakewalk. The devs have repeatedly discussed the details of the audio engine on the forum. Sonar has full 64 bit float throughout. There are a few legacy plugins but any of the new ones are 64 bit float. Any 3rd party VST 2.4 plugins will also communicate at 64 bit float. The limitation is the audio I/O and/or file bit depth if the user chooses less than 64 bit. (Which you usually will of course but if you want to use 64 bit, you can).

Anyway - SAWStudio is definitely not for everyone as it commands are a little different than other DAW's, and it's price is in the higher range for DAW apps - but I've been using some version of it since 1994 so that I'm super quick on it, it's incredibly stable, the support is excellent, and do find that the results after processing is done with it remain very high quality.

If it works for you and sounds good, what else can you ask for?

Personally I don't really think 64 bit makes that much difference in the end for most operations. For the more complex stuff any plugin can use a larger bit depth where needed. But yes I turn on the "double-precision" switch in Sonar if I sue it. If it is there...

Alistair

[Cellotron]

Originally Posted by alyricalmind

Woh, I just went on their website and saw the price for sawstudio... from the screenshots it looks like a dated GUI.

Except you can change the complete look of it with a single mouse click by loading a skin called "Shades" in SAWspeak - there's at least over 100 of these, most created by SAW users - all of them free to download, including a pretty darn close emulation of PT's look, as well a freeware app available that lets you create your own.

Why does it cost so much ($2500)?

Well there's also the Basic version that goes for $300, and the Lite version that goes for $1200.
As far as the full version, it costs that much because that's the price being asked for it, and it's still priced cheaper than other DAW app's such as Sequoia, SADIE and Pyramix.

With that money I could buy a DIGI 003 Mixing Console with automated faders, good preamps, decent conversion and a whole lot more features... PLUS Pro Tools 8. Is there something I'm missing about this DAW?

Well, to my ear the 003 pre's and conversion is ok, but certainly not anything to get excited about - but that's a separate issue.

Features that SAWStudio has that PT doesn't include:
much faster than real time "Bounce to Disc" functions, including numerous routines which allow for ultra-easy track consolidation with and without effects processing,
automatic plugin latency delay compensation,
the ability to use any audio interface that has Windows ASIO, MME, WDM or DWave drivers available for it - so your not tied into just one company's products for this,
support for VST, DX and SAW native formats (which includes a number of excellent sounding processors not available for any other format) - although I'll grant that this is definitely off-set by the extensive choice of RTAS plugins available - although again I'd say these come at greater cost than what is available for SAW,
an inexpensive add-on that allows PQ index and subcode creation directly in its Multitrack view allowing creation of Red Book spec audio masters from it,
a unique way of handling fades that for me allows for more control over the sound of the end result,
NO copy protection, NO dongle and NO Windows registry hooks, along with a generous license so that you can put it on numerous DAW's at your studio with no additional cost,
the ability to synchronize up to 8 remote and slave machines via TCP/IP so that you can have it run gigantic sessions all networked together,
the ability to hit a single key and have the cursor snap to the nearest zero energy crossing point
the ability to load soundfiles with different bit depths and sample rates and have them playback at any selected sample rate using a real time src routine without having to convert the files first,
an excellent history of support with numerous updates each year containing feature enhancements given for free, and when occasional paid upgrades happen none has been charged for more than $100 to date,
assembly language optimized engine that is extremely fast and stable,
a forum where you can get answers directly from the apps author himself, and where you can also make feature requests known to him as well.

There a number of things that PT can do that SAW doesn't do though as well. What is more important to you and what is worth what amount of money to you will vary on your own needs. For my own needs PT just doesn't cut it and SAWStudio works great. For you it will likely be the opposite. Whatever works for you is all good by me.

Best regards,
Steve Berson