AAX (Dsp) & AAX Native (Daw Cpu) ?

[zavid]

Originally Posted by JFreak

It all comes down to gain staging. TDM as a 48bit fixed-point architecture was very strict, because once you went over you immediately heard it and could fix it on the spot. AAX on the other hand is 64bit floating-point and going over is much more subtle -- it means you can smear your sound without noticing and it all adds up. Eventually your sound sucks if your gain staging is wrong.

Sonnox on the other hand -- pls correct me if I am wrong -- has implemented their plugins to be "TDM compatible". Which means they clip the signal internally if you go over. IOW using Sonnox plugins your gain staging is likely better than using other AAX plugins and when your gain staging is good your plugins sound like they should. Not better, but they sound worse if you go over and don't care.

Originally Posted by TrackerLe

Its not hard to unsterstand, what Zavid wrote.


A friend has a HD PCIe TDM system and got a HD native rig. He says, that the session, he used, sounds better with is HD TDM hardware, than with the HD Native rig. Even the tdm plugins are superior over their AAX counterparts.


For me, this is a myth. I have a very long history with Digidesign and later Avid...
For me there is no difference between TDM and AAX. I used the Sonnox Plugins TDM and later AAX DSP...(no difference). Or another example, the MDWEQ; No difference between the tdm version and the aax one...With the Avid stock plugins, its the same. For me HDX and AAX DSP has one big advantage. Low latency... You can track and record over HDX with plugins in low latency. If you are only mixing, HD Native is very nice, cause of the I/O.
At last, if someone says AAX, he means AAX Native. Today AAX DSP has no priority for the Plugin Manufacturers. We still have Sonnox, McDSP, Plugin Alliance, Cranesong, Softtube (old plugins, but the new ones are all native)... and some more like Flux Audio, but we see,that HDX is dying and Avid is not Universal Audio with their DSP plattform... UA pushes their DSP solution with their own plugins...Avid, hmm... does nothing. No new plugins, no special things for HDX. The promised upgraded Voice Count of HDX cards will never come... For me HDX is dead...still useful, but dead in terms of development... There won´t be anything new about HDX...And the prime time is over... TDM was great and HDX is the end...

much thanks for the posts !!!

if computer cpus continue to increase the bit resolution (64,96,128…) probably the dsp cards will no longer be necessary for low number tracks sessions-songs….
like electronic music , advertising spots for radio …..

but in case of hi number of tracks sessions-songs
like 64-96 of classical music , or post production ?

cheers

PS:

Originally Posted by musicman691

Your response is not understandable. I gather English is not your native language right?

as you do not understand a relatively simple technical post
unlike almost all others ……

should i assume you are a digital audio processing "newbie " ?

or simply the very Big "P.O.S." on duty ?

[musicman691]

Originally Posted by zavid

much thanks for the posts !!!

if computer cpus continue to increase the bit resolution (64,96,128…) probably the dsp cards will no longer be necessary for low number tracks sessions-songs….
like electronic music , advertising spots for radio …..

but in case of hi number of tracks sessions-songs
like 64-96 of classical music , or post production ?

cheers

The number of bits a cpu operates at has nothing to do with whether dsp chips are needed. It's how many transistors can be crammed on to a chip, keep the heat (and power) down at speed and deliver the performance. And how many cpu cores are on said die. I'm talking real physical cores and not hyper-threaded cores.

[zavid]

Originally Posted by musicman691

The number of bits a cpu operates at has nothing to do with whether dsp chips are needed. It's how many transistors can be crammed on to a chip, keep the heat (and power) down at speed and deliver the performance. And how many cpu cores are on said die. I'm talking real physical cores and not hyper-threaded cores.

Then if i have a 32 bit Operating System (Mac or PC) and a plug in or v. instrument that work in both 32-64 bit ,
it sounds identical in both bit resolutions ?

[musicman691]

Originally Posted by zavid

Then if i have a 32 bit Operating System (Mac or PC) and a plug in or v. instrument that work in both 32-64 bit ,
it sounds identical in both bit resolutions ?

Yes.
I'm beginning to think you don't understand much about computers - am I right? But you're slowly learning I hope. The number of bits in an operating system has nothing to do with audio quality. It's the number of bits in the audio that count towards audio quality/sound.

[zavid]

Originally Posted by musicman691

Yes.
I'm beginning to think you don't understand much about computers - am I right? But you're slowly learning I hope. The number of bits in an operating system has nothing to do with audio quality. It's the number of bits in the audio that count towards audio quality/sound.

What i know (and for sure !! )
is that a vst , au. rtas….plug in version do not sound well like the same plug in but in TDM version !

you can compare it simply with any bundle of Waves plug ins
available in VST,AU,RTAS,TDM,AAX….

and the same session song sound so much better
with the TDM mixer (dsp cards based)
than any Vst , or Rtas , AU… (D.a.w. cpu based) mixer ,

have you read a Digidesign pdf doc called :
"48 bit mixer" ?
(... "word" bit truncation , data reduction / digital data "compression"… )

"ok" , it is true that the quality of the algorithm is the "key"
for a quality sound processing and then the final results ,
(otherwise Lexicon wouldn't be that it is about reverberation… , apart the fact that they were "pioneers" about dsp…)

but if subjected to :
"word" bit truncation , data reduction / digital data "compression"
it cannot sound well as with a "linear" processing .

cheers

ps:
reason why ( 20 years ago.. ) TDM (dedicated dsp cards based ) won the "pro target" over Vst (d.a.w. cpu based) and other similar systems,
based on "dynamic" word bit truncation , data reduction / digital data "compression .…

more tracks (and plug ins assigned ) less sound quality ….
for avoid cpu crash (after few tracks..)

[BScout]

Originally Posted by zavid

you can compare it simply with any bundle of Waves plug ins
available in VST,AU,RTAS,TDM,AAX….

and the same session song sound so much better
with the TDM mixer (dsp cards based)
than any Vst , or Rtas , AU… (D.a.w. cpu based) mixer ,

Originally Posted by zavid

but if subjected to :
"word" bit truncation , data reduction / digital data "compression"
it cannot sound well as with a "linear" processing .

Sorry to disappoint but the native engine (VST2, RTAS, AAX native) is the one without truncation (and data collectors) and the TDM engine was the "bit limited" one with truncation for a long time. You have to go back to the early days of Pro Tools when the TDM engine was superior.

It wasn't until AAX-DSP that they were on par with each other where the DSP engine had the same processing depth.

[musicman691]

Originally Posted by zavid

What i know (and for sure !! )
is that a vst , au. rtas….plug in version do not sound well like the same plug in but in TDM version !

you can compare it simply with any bundle of Waves plug ins
available in VST,AU,RTAS,TDM,AAX….

and the same session song sound so much better
with the TDM mixer (dsp cards based)
than any Vst , or Rtas , AU… (D.a.w. cpu based) mixer ,

have you read a Digidesign pdf doc called :
"48 bit mixer" ?
(... "word" bit truncation , data reduction / digital data "compression"… )

"ok" , it is true that the quality of the algorithm is the "key"
for a quality sound processing and then the final results ,
(otherwise Lexicon wouldn't be that it is about reverberation… , apart the fact that they were "pioneers" about dsp…)

but if subjected to :
"word" bit truncation , data reduction / digital data "compression"
it cannot sound well as with a "linear" processing .

cheers

ps:
reason why ( 20 years ago.. ) TDM (dedicated dsp cards based ) won the "pro target" over Vst (d.a.w. cpu based) and other similar systems,
based on "dynamic" word bit truncation , data reduction / digital data "compression .…

more tracks (and plug ins assigned ) less sound quality ….
for avoid cpu crash (after few tracks..)

48 bit mixer has NOTHING to do with the word length the computer is working on. There's no data reduction or anything like that. You really don't know much about what you are talking about and that's for sure. You're confusing data word length with audio bit depth.

[zavid]

Originally Posted by BScout

Sorry to disappoint but the native engine (VST2, RTAS, AAX native) is the one without truncation (and data collectors) and the TDM engine was the "bit limited" one with truncation for a long time. You have to go back to the early days of Pro Tools when the TDM engine was superior.

It wasn't until AAX-DSP that they were on par with each other where the DSP engine had the same processing depth.

just "as well" , never worked with an old 32 bit computer and a Vst d.a.w. ?
(when the number of plug ins usable depended by the cpu power , and in which more tracks were added and worse it sound )

if a session song require more dsp processing power than that you have on your d.a.w. cpu , you are in (serious) troubles…..

with a system based on dedicated dsp cards
you can add more cards…

"dynamic" digital data compression apart….
(…and related cut of sound frequencies and depth.. , including the obvious degradation of the same )

Originally Posted by musicman691

48 bit mixer has NOTHING to do with the word length the computer is working on. There's no data reduction or anything like that. You really don't know much about what you are talking about and that's for sure. You're confusing data word length with audio bit depth.

… and dynamic range ?

"linear vs data compression processing"…

bit is bit : 0 and 1

if a plug in native AAX sound better than the same plug in placed on dsp HDX card ,
is only thanks to 64 bit based cpu , O.S. , and obvious speed clock …
(more speed = more fast processing , then more quantity of bits /bytes per second , though the vst do not sound like TDM and neither Rtas …)

apart the quality of the plug in algorithm , as mentioned above ...

And :
Apart that there is something (that sound) very wrong about HDX cards .
(….or Avid became masochist…)