HI everybody,

I just got back from a discussion with our chief Audio Programmer at the video-game company I work for...(He is considered one of the very best in our business)


I explained the differences to him regarding the old EXS24 16bit VSL VERSUS the new VI instrument 24bit Lossless Compression system...

Here are his comments:

When playing back an uncompressed WAV file that is loaded into RAM the CPU takes no hit whatsoever.

When playing back a 24bit Lossless compression Audio file (it is 2to1 compression....meaning the 24 bit file only takes up 1/2 the space it would uncompressed) the computer takes quite a hit....He could not say how much, but it's there....Now factor in crossfades, stacking, each file needs to uncompress on the fly to play (you see what I'm saying), release samples...polyphony builds up quick (each file hitting the cpu)

So ........It stands to reason that if we were able to turn the compression OFF......We would need twice the RAM to load all this stuff BUT the CPU would be able to play WAY more VI instances.

SvK

I would rather have compression turned off.....and strap 2 to 4 G5s together.......and not have my CPUs crap out, when composing.....

SvK

@Another User said:

HERB,

Processor: G5, 2 x 2.5 GHz,
Ram: 4 GB
OS 10.4.2
Data storage: Firewire 800
Host application: Logic 7.1.1
Soundcard: RME HDSP MADI with 648 MADI Interface

Latency at 512 Samples (12ms) / 2.5 GB Ram usage
Maximum Vienna Instrument instances: 24 (CPU 90%)
Maximum Samples loaded: 45000
Maximum polyphony: 300 stereo voices / (CPU 70%)

Herb I can load 24 instances of VI easy.....BUT I can't play any of it multitimbrally without massive cutting out and stalling.....I'm not making this up and I really do know what I am doing.

The loading up is not the issue.


Any chance of you guys running these tests with current OSX?

10.4.2 is ancient 10.4.6 please

And current Logic 7.2 (7.1.1 is old now)


maybe something happened since Logic went over to universal binary....

thanx Wink

SvK

Hi,

I don´t have the G5 dual here at the moment but can grab it later today and run the same tests in 10.4.6 and Logic 7.2. However I don´t expect it to be that different.

Maya

MAYA,

please do, that would be great......

Is it possible that you tell us what sort of test this is?


thanx so much...

SvK

MAYA,

Here are some insights in to other MAC CPU related problems...

problem 1:

While in Logic 7.2, select a sustained VI string instrument. Now play the keyboard.....every once in a while the sound just "cuts off" (this is not a polyphony thing, as we are only sustaining 1 note....it happens in standalone as well......so not a L7 issue)


problem 2:

While in Logic 7.2, enable the record-switches of any 3 Leg String instruments...(so you are playing 3 patches of VI simultaneously).....now play an appregiated triad in Legato (just drum your fingers on the keys)......Everything starts to stutter and break-up.....Even with release samples turned off......Now intersestingly, when you record this appreggio and have the sequencer play it back, the stuttering is reduced (probably, because the sequencer can look ahead).

I am still certain that "Lossless compression " is really impacting the CPU performance....THIS IS A BIG ISSUE.........

Since when opening old-sessions created with VSL/EXS 24, and replacing all those instances with VI's (newlibrary) ....those sessions will NOT play...



Like I said, RAM is not the issue, everything loads...It's CPU performance that gives in.


LAST THING:

I understand that PC's have a 2 Gig RAM cap, as opposed to the possible 4 to 8 gig RAM on a MAC..........Something tells me that the lossless compression, was a sheme to get more sounds into the PC's at the xpense of CPU power, consequently MAC users now have to suffer the shortcomings of this compromise due to the PC's RAM limitations.....

I still belive that a "16bit-uncompressed" version of this Library would be a GOOD thing...for Layout purposes and composition....All these CPU issues would go away.....At this stage things are at a standstill.

Setup a bunch of servers, launch the batch processors, and dither down to 16bit. So we can enjoy this awesome achievement, instead of having to wait for technology to catch up with your product.

SvK

I think streaming 300 stereo voices with 70% CPU usage proofes that there should not be any issue with compression.
I'm no expert, but I'm quite sure that uncompressed 24bit data wouldn't allow this amout of voices, because of physical bottlenecks.

best
Herb

HERB,

All I am asking is that you perform the same test I layed out....

specifiacally appregiating 3Leg patches at the same time.......The sounds stutter.

SvK

HERB...

it would allow MORE


......i'm sending you an email

SvK

Our stresstests are exactly testing this issues.

One polyphony test where the amount of streaming voices (sustains) is tested.
And a second one where the amount of instances, each performing very fast simultaneous movement with performance legato instruments are tested.
So when we say we can perform 24 instances without glitches, than 24 performance legati instruments are running at maximum speed.

best
Herb

HERB,

svonkampen computer:

Dual 2GHZ with 4 Gig RAM
OS 10.4.6, Logic 7.2 (i/o buffer size 512....Process Buffer Range LARGE)
RAM Usage is 2.95 GIG
Metric Halo 2882 audio interface


I open 3 VI instances of Legato Strings.
I drum my fingers (legato style) on the keys

The sound stutters and the CPU meter in Logic goes RED.



SvK

My guess is that you've loaded too much into your "only" 4GB of RAM, SvK, and the plug-in went to virtual memory for the sample start buffers.

Nick...RAM usage 2.9 Gig

Exactly, 2.9GB and you only have 4GB installed. Please check the virtual memory setting as you load up Vienna Instruments - that will tell you whether my theory is correct.

Where is this virtual memory setting?

We are talking about the new VIs not EXS24 Right? Virtual memory settings don't exist with VIs



SvK

SVONKAMPEN,

1) Concerning the test with OS 10.4.6 and Logic 7.2, here are my results:

Processor: G5 Dual 2.5 GHz
Ram: 4 GB
OS 10.4.6
Data storage: Firewire 800
Host application: Logic 7.2
Soundcard: RME HDSP

Latency at 512 Samples (12ms) / (2.46 GB Ram usage)
Maximum Vienna Instrument instances: 23 (CPU load activity monitor 95%)
Maximum Samples loaded: 24.095
Maximum polyphony: 384 stereo voices


2) The RAM management in VI works differently from the EXS. Nevertheless the VI uses disk streaming technology like the EXS does.
I explained this in detail in an email that I send to you before.
However I would suggest to get more RAM in order to run 8 or more instances of the VI properly.

3) I can confirm that when playing 3 legato instruments (= 3 instances of VI) simultaneously I also encountered a problem (a few clicks), but the tracks played back perfectly after I recorded them. You will find my Logic file on my iDisk.
I will look into this and keep you updated.

Best,

Maya

Maya,

thanx for confirming the bug......yes, they do play back properely after they are recorded, but they need to play correctly while performing the part,

Thanx again [;)]

SvK

SvK, I mean virtual memory as shown in Activity Monitor, i.e. the system is using the hard disk because it's run out of RAM. Not EXS virtual memory.

Nick....

gotcha


Yeah....maya and I have been emailing back and forth....she's real nice and helpful. it is the way it is. Guess I'll hit-up IT for a couple more G5s.

Ummm....but as you can see from this post the "Live" playing on the MAC at least does stutter when selecting multiple instances to play at the same time...however it does play back ok.........

funny thing is.....I've been forced to write "minimalistic"....since the CPU craps out so soon now.........and I'm liking it....my cues are starting to breath [;)]

hahahaha

Instead of covering-up hastily programmed string-lines with Horn pads, I'm forced
to expose the string-lines and do ultra-realistic programming.

I guess the glass is half-full after all.

SvK

Thanx for takin this time to post your issues and tests, svonkampen.
This is really appreciated.
For the moment I'm using a Quad and haven't tested VI's limits.
Also, I use Logic but I load my Vienna VI's in Rax (separate host than Logic), so as not to to have to reload all the samples every time I swap songs etc...

I read all this thread but rather quickly...

I would just like to point out that in some cases it may be your drives that might be touching their limits and giving you audio quircks. Upon 'live' playing, the system clearly has less flexibility to organise resources than upon 'sequenced playback'...

Also maybe Logic is the cause of your 'real time" quircks.
In Logic, there is clearly a difference between real time playing and sequence playback !!!
It would be interesting if you tested the VI in an external host and play the notes on a midi instrument in Logic connected vie iacbus to the VI host.. If you got the same problem on 'live' playing it would indicate that it is not Logic.

Also, bare in mind that OSX has the ability to re-organize resources to better achieve a task that it may have only partially achieved at a previous trial. Evan Evans wrote some very interesting thoughts on this last year...
He told of how, at 3rd or fourth playback, some exs instruments were running fine while at first playback the audio was full of hiccups.
I'm not saying it's totally connected to your problem, but it's worth knowing about it...

One last side issue worth knowing about (I donno if it has been corrected since Logic 7.2) is that in some cases, when Logic has hit a CPU limit, and has given you the alert message... You may find that it will repetitively give you that same message at 'given' point in the arrange, even though you've changed the arrangement and playback should run fine.
Usually this happens at entry point of a sequence. Clearly this is a bug !!! Logic has no reason to give you the 'CPU maxed' message.
There is one simple way to bring back Logic to normal : Nudge 1 or a few sequences (with start point at problematic position) back & forth a bar.

You might run into this problem if you're having CPU issues, so I thought this might be a useful tip.

Just a few thoughts...

Charl

@charl said:

I would just like to point out that in some cases it may be your drives that might be touching their limits and giving you audio quircks. Upon 'live' playing, the system clearly has less flexibility to organise resources than upon 'sequenced playback'...

Also maybe Logic is the cause of your 'real time" quircks.
In Logic, there is clearly a difference between real time playing and sequence playback !!!
It would be interesting if you tested the VI in an external host and play the notes on a midi instrument in Logic connected vie iacbus to the VI host.. If you got the same problem on 'live' playing it would indicate that it is not Logic.

Also, bare in mind that OSX has the ability to re-organize resources to better achieve a task that it may have only partially achieved at a previous trial. Evan Evans wrote some very interesting thoughts on this last year...
He told of how, at 3rd or fourth playback, some exs instruments were running fine while at first playback the audio was full of hiccups.
I'm not saying it's totally connected to your problem, but it's worth knowing about it...

One last side issue worth knowing about (I donno if it has been corrected since Logic 7.2) is that in some cases, when Logic has hit a CPU limit, and has given you the alert message... You may find that it will repetitively give you that same message at 'given' point in the arrange, even though you've changed the arrangement and playback should run fine.
Usually this happens at entry point of a sequence. Clearly this is a bug !!! Logic has no reason to give you the 'CPU maxed' message.
There is one simple way to bring back Logic to normal : Nudge 1 or a few sequences (with start point at problematic position) back & forth a bar.

You might run into this problem if you're having CPU issues, so I thought this might be a useful tip.

Just a few thoughts...

Charl

Excellent Post.

This is a reality that, for me, goes on day after day.
After entering a few tracks in a full orchestral arrangement, it takes logic 3 or 4 playbacks to 'straighten out'. As i use only a laptop, and push the limits constantly, this is something i'm used to. Freezing tracks helps, but if i'm writing a robust passage, then i will get meesage of of CPU overload, and as you rightly point out Charl, shifting the playback point a bar or two frees up the system again, the messages dissipate.
It's interesting to note that Logic seems to 'learn' fairly quickly. Once a full passage is set, if i add a track, the activity monitor shows the previous audio load, plus an added percentage for the new track. When i playback the updated passage two or three times, the audio load reduces to just above the previous setting. (I should say this is with EXS. A clever tool, and i mourn the fact this sampler isn't sold in standalone format anymore. Apple would make a fortune if they did this, i'm sure.)
Even on my little laptop (G4,867mhz,512mg ram, 5400rpm HD), i can, with care and patience, load upwards of 80 tracks of instrument articulations, playing back each new 'total' until the audio comes clean, then adding another track. It's not perfect, but the work gets done. I still have to freeze tracks occasionally, but in general my little rig get's to the end of some fairly hefty arrangements unscathed, after the compromises mentioned.
I'm about to add an external HD (7200rpm), and i think this will reduce the instances of audio 'glitch' even further.

I think some of the discussion about the VI's and overloading symptoms stems from the assumption that live playback first time should be perfect, each and every time.
Logic, and other programs too, don't do this even with a smaller library (Just my example and experience). I'm not sure how many work as i do, playing back a passage over and over again, fine tuning and modifying, but it's a routine that i use, that i virtually ignore the first one or two playbacks, until Logic learns what to play, and when.

Good post Charl, and a timely reminder that more than one cause is possible for a symptom.


Alex.