EXS 24 Logic Sampler Users Group group photo

Yahoo Groups archive

EXS 24 Logic Sampler Users Group

Index last updated: 2026-04-28 23:25 UTC

Thread

CPU overloads

CPU overloads

2004-07-25 by Julie Larson

Hi all,
I'm using a 1ghz TiBook, 1gig ram with a Digigram Vx pocket.  I have 
fast external drives that contain my audio files and samples.  The Vx 
pocket has been a little trogan and I've used it for years.  But I'm 
having frequent audio overruns as well as Disk I/O failures recently.  
The music is orchestral....the instruments are large....  GOS, Sam 
Brass, Dan Dean Winds, Post BOS290 etc.   I'm not a techie.... I'm not 
even sure I spelled it right but could my audio card be causing these 
problems....or I guess the better question is....What is the most 
probable cause of these problems?  When I look at my system 
performance....the "audio" side is maxed out...it chokes.  So...I 
freeze some files.  Almost imediately the Disk I/O is maxed.  When I 
Solo the "frozen" tracks... no disk labor.  When I mute the frozen 
tracks....the Audio side is at manageable  levels.  But when I un-mute 
the "frozen" tracks I get Disk I/O failures.  Then I unfreeze the 
tracks....Audio overruns.   I would really appreciate some insight into 
this.

Thanks in advance,
julie

Re: [EXS] CPU overloads

2004-07-25 by Brio Taliaferro

It sounds a bit like these problems have crept up on you just recently, so
my first guess is that disk fragmentation is to blame. Try using norton's
speeddisk or similar to defrag all your harddisks. Run Diskdoctor while
you're at it to make sure all the disk catalogues etc are in ship shape.

Secondly, try to spread the load over many disks (i.e. have the streamed
samples, the audio, and the freeze files on up to 3 different disks for very
heavy projects). You could also set the freeze files to a lower bit depth
(not sure where this is done).

Thirdly, you can try tweaking the buffer values in the audio driver section
of the Preferences. I am mentioning this last, because it will have adverse
effects on latency. However it might be worth fiddling with to get that
extra bit performance out for a very heavy session.

Hope this helps,
Brio
Show quoted textHide quoted text
On 25/7/04 4:05 am, "Julie Larson" <julielarson@...> wrote:

> Hi all,
> I'm using a 1ghz TiBook, 1gig ram with a Digigram Vx pocket.  I have
> fast external drives that contain my audio files and samples.  The Vx
> pocket has been a little trogan and I've used it for years.  But I'm
> having frequent audio overruns as well as Disk I/O failures recently.
> The music is orchestral....the instruments are large....  GOS, Sam
> Brass, Dan Dean Winds, Post BOS290 etc.   I'm not a techie.... I'm not
> even sure I spelled it right but could my audio card be causing these
> problems....or I guess the better question is....What is the most
> probable cause of these problems?  When I look at my system
> performance....the "audio" side is maxed out...it chokes.  So...I
> freeze some files.  Almost imediately the Disk I/O is maxed.  When I
> Solo the "frozen" tracks... no disk labor.  When I mute the frozen
> tracks....the Audio side is at manageable  levels.  But when I un-mute
> the "frozen" tracks I get Disk I/O failures.  Then I unfreeze the
> tracks....Audio overruns.   I would really appreciate some insight into
> this.
> 
> Thanks in advance,
> julie

Re: [EXS] CPU overloads

2004-07-25 by Mac Duff

But are not the buffer values really effect input only?? I know they impact
playback, but once you're playing back, doesn't Logic simply get itself
together of its own accord? It's the real-time interaction with the
sequencer as it tries to handle a heavy load at the same time where latency
come into play? I'm probably wrong on that.

The following may not effect disk performance much (definitely try
defragging the disks, as suggested), but these settings helped to increase
Logic's overall efficiency when handling virtual instruments, perhaps
reducing your dependency on freezing tracks:

Audio driver:
12 tracks
64 Busses off
Process Buffer Range Large
Larger Disk Buffer on

EXS24 prefs (open an EXS24 edit window -- where you can see the samples laid
out across the keyboard -- and from the Edit menu of that window select
preferences):

SR Conversion: "Normal". "Best" is better at stretching samples across wider
keyboard ranges, but demands more processor. If you're already using
converted Giga sets, the sample ranges are most surely quite small. So,
"Best" is probably unnecessary.

Sample Storage: "Original". "Yes, 32 Bit Float" does load more of the
samples into RAM, effectively not using Logic's virtual memory. If your
sample banks exceed your Mac's RAM and OS X starts to use its own virtual
memory, however, things can get bad really fast.

And finally, really scope out the polyphony assignments for your EXS24
patches. Logic is fixated at importing patches to a polyphony setting of 64
NOTES!! This is in most cases RIDICULOUSLY HUGE! Polyphony kills your audio
driver -- even if you're not using all the notes assigned.

Hope this helps!

MacDuff


PS: If you can get more RAM in your Powerbook (some 1GHz books can take
2GB), I'd suggest maxing it. I have 2GB of RAM, and I am well over 1GB with
large banks (current one is 1.6GB in size!). These days, I have all 64
instruments active, all of them EXS24 except for one EVP88 and one EVB3,
using one instance of Space Designer. While I do have dual 1.3GHz G4
processors, I'm not sure Logic is as dual processor savvy as OS X itself is.
Show quoted textHide quoted text
On 7/25/04 8:07 AM, "Brio Taliaferro" <logic@...> wrote:

> It sounds a bit like these problems have crept up on you just recently, so
> my first guess is that disk fragmentation is to blame. Try using norton's
> speeddisk or similar to defrag all your harddisks. Run Diskdoctor while
> you're at it to make sure all the disk catalogues etc are in ship shape.
> 
> Secondly, try to spread the load over many disks (i.e. have the streamed
> samples, the audio, and the freeze files on up to 3 different disks for very
> heavy projects). You could also set the freeze files to a lower bit depth
> (not sure where this is done).
> 
> Thirdly, you can try tweaking the buffer values in the audio driver section
> of the Preferences. I am mentioning this last, because it will have adverse
> effects on latency. However it might be worth fiddling with to get that
> extra bit performance out for a very heavy session.
> 
> Hope this helps,
> Brio
> 
> On 25/7/04 4:05 am, "Julie Larson" <julielarson@...> wrote:
> 
>> Hi all,
>> I'm using a 1ghz TiBook, 1gig ram with a Digigram Vx pocket.  I have
>> fast external drives that contain my audio files and samples.  The Vx
>> pocket has been a little trogan and I've used it for years.  But I'm
>> having frequent audio overruns as well as Disk I/O failures recently.
>> The music is orchestral....the instruments are large....  GOS, Sam
>> Brass, Dan Dean Winds, Post BOS290 etc.   I'm not a techie.... I'm not
>> even sure I spelled it right but could my audio card be causing these
>> problems....or I guess the better question is....What is the most
>> probable cause of these problems?  When I look at my system
>> performance....the "audio" side is maxed out...it chokes.  So...I
>> freeze some files.  Almost imediately the Disk I/O is maxed.  When I
>> Solo the "frozen" tracks... no disk labor.  When I mute the frozen
>> tracks....the Audio side is at manageable  levels.  But when I un-mute
>> the "frozen" tracks I get Disk I/O failures.  Then I unfreeze the
>> tracks....Audio overruns.   I would really appreciate some insight into
>> this.
>> 
>> Thanks in advance,
>> julie

Re: [EXS] CPU overloads

2004-07-25 by Julie Larson

Thanks everyone for all the help,
For starters....I'm using OSX...sorry not to have mentioned that.  I'm 
up against the wall on memory for a laptop.  1gig is all I can have.  I 
have de-fragmented the hard disk and it was pretty badly fragmented, 
but I didn't get any noticeable improvement.  What is odd about this 
case is that freezing tracks seems to exacerbate the problem.  This has 
not been my experience in the past.  Freezing does alleviate strain on 
the "Audio" side of things....but I immediately get Disk i/o failures.  
1 frozen track will do this.  I am freezing at a lower bit level.  That 
eased the problem on a previous composition.   By the way, you set bit 
depth for freeze tracks in the Audio preferences window.  I've set the 
buffer values on my audio driver to maximum.  I thought that would give 
the best playback.   I reduced polyphony on several of the EXS 
instances...It helped in a couple of spots.

My samples and freeze files are on one external 80gig 7200 rpm hard 
drive that is partitioned into 2 sections. Maybe it's time for another? 
  I have actually been suspecting my audio card.  I recently tried to 
switch from the Digigram V2 to the 440 to get 2 additional outputs.  I 
couldn't use the 440 because servicing the additional outs caused a 
marked decrease in performance.  Sooo....I thought maybe another card 
might increase performance?  	Relieve strain on the CPU?   When I 
bought the digigram V2 it was just about the only "professional" option 
for laptop users.  That has changed.  Any ideas on that front....or am 
I on the wrong track there?

Thanks again
julie

Re: [EXS] CPU overloads

2004-07-25 by Andy Hardwake

Hi Julie,

On Jul 25, 2004, at 12:08 PM, Julie Larson wrote:

> I'm using OSX...sorry not to have mentioned that.  I'm
> up against the wall on memory for a laptop.  1gig is all I can have.

Do you have virtual memory enabled in EXS options? Have you tried 
experimenting with settings there? This has saved my day many times. 
Just an idea.

Best,

Andy

Re: [EXS] CPU overloads

2004-07-25 by Julie Larson

Hi Andy,
Yes virtual memory is on.  I have it set to Slow for hard disk 
speed....and Extensive recording activity.  Even though both of those 
are false...I figured that would give me the biggest buffer for play 
back?  right?  wrong?

Thanks,
julie

Re: [EXS] CPU overloads

2004-07-25 by Mac Duff

On 7/25/04 4:00 PM, "Andy Hardwake" <digitalmechanics@...> wrote:

> Do you have virtual memory enabled in EXS options? Have you tried
> experimenting with settings there? This has saved my day many times.
> Just an idea.

Yeah. Sample Storage to "original" helps (as I suggested in my previous
post)

Re: [EXS] CPU overloads

2004-07-25 by Eli Krantzberg

On Jul 25, 2004, at 3:08 PM, Julie Larson wrote:

>  I am freezing at a lower bit level.  That
> eased the problem on a previous composition.   By the way, you set bit
> depth for freeze tracks in the Audio preferences window.

I know this defies conventional wisdom; but I seem to remember reading, 
on some past thread on the LUG, that reducing the bit depth of the 
freeze files from 32 bits actually increases the strain on the CPU in 
some way.

I think it was that because logic works internally at a 32 bit 
resolution, decreasing the freeze files gives the CPU extra conversion 
work to do. And so that while it might reduce the disc reading load, 
reducing the bit depth might not actually be a net gain to overall 
performance. Does anyone remember the specifics of this??

Freezing the files at 32 bit instead of 24 might be worth a shot, if 
for nothing else, than to rule it out as a possibility.



--------
Eli Krantzberg
http://www.nightshiftorchestra.com
Almat Productions



[Non-text portions of this message have been removed]

Re: [EXS] CPU overloads

2004-07-26 by Andy Hardwake

On Jul 25, 2004, at 1:17 PM, Julie Larson wrote:

> Yes virtual memory is on.  I have it set to Slow for hard disk
> speed....and Extensive recording activity.  Even though both of those
> are false...I figured that would give me the biggest buffer for play
> back?  right?  wrong?

Right AFAIU, but I would still try setting them to medium just to see 
if it makes any difference at all in a situation like yours.

Best,

Andy

Re: [EXS] CPU overloads

2004-07-26 by Don Newmeyer

on 7/25/04 12:08 PM, Julie Larson at julielarson@... wrote:

> Thanks everyone for all the help,
> For starters....I'm using OSX...sorry not to have mentioned that.  I'm
> up against the wall on memory for a laptop.  1gig is all I can have.  I
> have de-fragmented the hard disk and it was pretty badly fragmented,
> but I didn't get any noticeable improvement.  What is odd about this
> case is that freezing tracks seems to exacerbate the problem.  This has
> not been my experience in the past.  Freezing does alleviate strain on
> the "Audio" side of things....but I immediately get Disk i/o failures.
> 1 frozen track will do this.  I am freezing at a lower bit level.  That
> eased the problem on a previous composition.   By the way, you set bit
> depth for freeze tracks in the Audio preferences window.  I've set the
> buffer values on my audio driver to maximum.  I thought that would give
> the best playback.   I reduced polyphony on several of the EXS
> instances...It helped in a couple of spots.
> 

Hi Julie,

I have almost the same setup you do (in fact until just a couple weeks ago I
was using a TiBook 867 MHz with 1 Gig RAM and the VxPocket 2, and a Kanguru
external 180GB drive) and use symphonic/piano libraries like you do.

Although I see the same dilemma between "audio" choking and hard drive
limitations, I could freeze something like a dozen tracks before the I/O
maxes out. So I think there's something wrong with your drive, disk I/O
system, or possibly the settings in Logic. OTOH, I don't think the settings
in Logic could possibly account for your getting problems with only one
frozen track.

A couple weeks ago, I fried my screen on the old TiBook, forcing an upgrade
to a new AlBook. Performance is considerably better, but not as much as I
expected. The FW800 port does help things, and I'm starting to use a newer
faster Firewire drive, but the result is not twice as fast as the old FW400
port.

I'm also wondering if one of the newer interfaces, like the RME multiface,
would be more efficient than the VxPocket. I noticed that the Vx was less
efficient (lower track counts) even on OS9 than a USB interface I was using
before.

Don Newmeyer

Re: [EXS] CPU overloads

2004-07-29 by Julie Larson

First of all....thanks so much to everyone for all of the great 
suggestions.  I think I've located the at least the immediate problem.  
I ran the activity monitor yesterday and discovered that once my Song 
had played all the way through....I was using all...completely all of 
my ram.  I'm surmising that this is putting a crunch on performance.  
My only problem is that I can't add any more.  Soooo I'm afraid I'm 
coming to the grim realization that I'm in the market for a new 
computer.  I need portability so....Has anyone heard any rumors about a 
G5 powerbook?  I've heard Apple is having a hard time keeping them from 
melting.

julie

Re: [EXS] CPU overloads

2004-07-29 by Mac Duff

On 7/29/04 8:19 AM, "Julie Larson" <julielarson@...> wrote:

> I ran the activity monitor yesterday and discovered that once my Song
> had played all the way through....I was using all...completely all of
> my ram.  I'm surmising that this is putting a crunch on performance.

It will definitely cause playback problems -- particularly if Logic is
passing the virtual memory chores on to the OS itself! When I had my EXS
prefs set the wrong way, it was damned nasty. In my previous post, I
mentioned what I found to be the best settings to avoid this, keeping ass
much of the RAM handling chores within Logic:

"Sample Storage: "Original". "Yes, 32 Bit Float" does load more of the
samples into RAM, effectively not using Logic's virtual memory. If your
sample banks exceed your Mac's RAM and OS X starts to use its own virtual
memory, however, things can get bad really fast."

So, try running EXS's prefs with Sample Storage set to "original". That
imparts a heavier load on the processor, but probably might be better than
OS X's virtual memory.

(also, did you reduce each EXS instance's polyphony down to each part's
acceptable minimum?)

> My only problem is that I can't add any more.

Recent Powerbooks can take 2GB of RAM.

> Soooo I'm afraid I'm
> coming to the grim realization that I'm in the market for a new
> computer.  I need portability so....Has anyone heard any rumors about a
> G5 powerbook?  I've heard Apple is having a hard time keeping them from
> melting.

I wouldn't expect a G5 Powerbook until AT LEAST Macworld San Francisco in
January (G5 iMacs are due for release in September, BTW). When that finally
happens, I too will be looking into moving to a laptop. It would be nice to
do a TV series score from a beach, somewhere :-)

Re: [EXS] CPU overloads

2004-07-29 by Julie Larson

Hey Mac,

On Jul 29, 2004, at 10:06 AM, Mac Duff wrote:
>  So, try running EXS's prefs with Sample Storage set to "original". 
> That
>  imparts a heavier load on the processor, but probably might be better 
> than
>  OS X's virtual memory.

I have sample storage set to "original" per your first post.
>
>  (also, did you reduce each EXS instance's polyphony down to each 
> part's
>  acceptable minimum?)
>

I systematically went through and reduced polyphony on each track.  
This helped with something that sounded a bit like track stealing but 
hadn't caused the computer to completely grind to a halt.  I'm still 
out of RAM.   I have a question here though.  Is it standard for an 
instrument that has multiple layers for....say keyswitching, to use 
polyphony for all layers even though only one layer is playing?  I 
understand that if you were using xfade to smooth transitions between 
layers that more polyphony would be used.  But this isn't the case.  I 
play one note and the polyphony use is 4.  Is there something I'm 
missing?

>
>  Recent Powerbooks can take 2GB of RAM.
>

I just didn't want to buy a new G4 only to find out that the G5s were 
coming out next week.

Yeah....The park bench idea sounds pretty dreamy.

Thanks again,
julie





[Non-text portions of this message have been removed]

Re: [EXS] CPU overloads

2004-07-29 by Murray McDowall

dropzone@... wrote: 

>
> "Sample Storage: "Original". "Yes, 32 Bit Float" does load more of the
> samples into RAM, effectively not using Logic's virtual memory. 


Actually, this parameter does not affect virtual memory - it determines whether
or not samples are converted to 32 bit float when loading them into RAM. This
is the format the audio engine uses so it saves the CPU load of converting the
formats at the time the sample is triggered but at the cost of increased RAM
footprint. 

If you are using virtual memory and you have this setting on my guess is that
it would only affect the preloaded buffer section of the sample.

Regards,
Murray

Re: [EXS] CPU overloads

2004-07-29 by Andy Hardwake

Hi Julie,

On Jul 29, 2004, at 5:19 AM, Julie Larson wrote:

> I ran the activity monitor yesterday and discovered that once my Song
> had played all the way through....I was using all...completely all of
> my ram.

Is your machine OS 9 bootable? If so, have you tried it under 9? Have 
you experimented with Logic memory settings there? I know it's a crazy 
idea, but if it might help, why not :-) ?

Best,

Andy

Re: [EXS] CPU overloads

2004-07-29 by Mac Duff

On 7/29/04 1:45 PM, "Murray McDowall" <murraymc@...> wrote:

>> "Sample Storage: "Original". "Yes, 32 Bit Float" does load more of the
>> samples into RAM, effectively not using Logic's virtual memory.
> 
> 
> Actually, this parameter does not affect virtual memory - it determines
> whether
> or not samples are converted to 32 bit float when loading them into RAM. This
> is the format the audio engine uses so it saves the CPU load of converting the
> formats at the time the sample is triggered but at the cost of increased RAM
> footprint. 

But it does effect OS X's virtual memory. As you say there's "a cost of AN
INCREASED RAM FOOTPRINT". And, if you have ten tons of samples loaded and
that footprint gets too big, you run out of physical RAM.. And then, OS X's
virtual memory kicks in -- and that's when I experienced significant
crackling and Logic crashing. We're talking well over a gig of samples on a
2GB system across about sixty instruments, including one EVB3, one EVP88 and
one Space Designer instance with dual 1.3GHz G4 (an upgraded Sawtooth).

I tried loading this large Logic project with this EXS setting to both
normal and 32 bit. With OS X's Activity Monitor showing me my RAM usage, 32
bit float mode scarfed up my physical RAM like crazy and paged to disk. It
seems that Logic would load up samples for instruments being used, but then
if you jump to a different section of your song file where other samples are
being used, or select the track with the EXB3 for instance and started
tickling the ivories, boy would crap start a'crackling! Activity Monitor
showed fervent VM pageins... and CRASH! Logic would go down. Repeatedly.

Then, set the EXS prefs to normal and the physical RAM usage is much better.
 
> If you are using virtual memory and you have this setting on my guess is that
> it would only affect the preloaded buffer section of the sample.

Actually, does Logic in OS X even have its own virtual memory?? In OS 9,
there was that System extension you could use (forget its name -- probably
something obvious). But, in OS X is there anything like that going on with
Logic??

Anyhoo, I stand by my point: OS X's virtual memory is poison to a stable
Logic system. "Original" sample storage and "Normal" sample rate conversion
helps avoid it.

Re: [EXS] CPU overloads

2004-07-29 by Mac Duff

On 7/29/04 12:18 PM, "Julie Larson" <julielarson@...> wrote:

> Hey Mac,

> I have a question here though.  Is it standard for an
> instrument that has multiple layers for....say keyswitching, to use
> polyphony for all layers even though only one layer is playing?

>  I play one note and the polyphony use is 4.  Is there something I'm
> missing?

According to the "Used" indicator on my EXS24's interface, no. I have a
big-ass bass instrument with finger, slap, acoustic and distorted samples,
all key switched. When I play one note, however, that event only consumes
one note of polyphony. So, I don't know what's happening there... Unless
maybe the stereo sample is actually two mono samples panned and therefore
two voices??!

>>  Recent Powerbooks can take 2GB of RAM.
> 
> I just didn't want to buy a new G4 only to find out that the G5s were
> coming out next week.

Most conjecture on the web points to January.

Re: [EXS] CPU overloads

2004-07-30 by Murray McDowall

At 02:05 PM 7/29/04 -0400, you wrote: 
>
> On 7/29/04 1:45 PM, "Murray McDowall" <murraymc@...> wrote:
>
> >> "Sample Storage: "Original". "Yes, 32 Bit Float" does load more of the
> >> samples into RAM, effectively not using Logic's virtual memory.
> > 
> > 
> > Actually, this parameter does not affect virtual memory - it determines
> > whether
> > or not samples are converted to 32 bit float when loading them into RAM.
> This
> > is the format the audio engine uses so it saves the CPU load of converting
> the
> > formats at the time the sample is triggered but at the cost of increased
> RAM
> > footprint. 
>
> But it does effect OS X's virtual memory. As you say there's "a cost of AN
> INCREASED RAM FOOTPRINT". And, if you have ten tons of samples loaded and
> that footprint gets too big, you run out of physical RAM.. And then, OS X's
> virtual memory kicks in -- and that's when I experienced significant
> crackling and Logic crashing. We're talking well over a gig of samples on a
> 2GB system across about sixty instruments, including one EVB3, one EVP88 and
> one Space Designer instance with dual 1.3GHz G4 (an upgraded Sawtooth).


(puts on mirrored shades and suth'n accent)

What we have here is a communication problem.

In an effort to avoid patent infringement problems Emagic's sample streaming
was termed "Virtual Memory." This system of buffering sample starts in RAM and
spooling the rest up from disk is internal to Logic and is like the system in
other streaming samplers -- or DAWs which playback audio tracks for that
matter. 

In OSX (or WinXP ) Virtual Memory is the name given for memory paging to disk
and is another matter altogether. The use of the same name for OS paging and
sample streaming has led to our misunderstanding. 

You are correct that as soon as you run out of physical ram the buffering of
samples fails and you get crackles on playback of EXS24 streaming samples. It
is also correct that 32 bit Sample loading is nothing to do with Emagics
"Virtual Memory" - ie sample streaming.

Regards,
M

Re: [EXS] CPU overloads

2004-07-30 by Mac Duff

On 7/30/04 2:41 PM, "Murray McDowall" <murraymc@...> wrote:

> (puts on mirrored shades and suth'n accent)

:-)

> What we have here is a communication problem.
> 
> In an effort to avoid patent infringement problems Emagic's sample streaming
> was termed "Virtual Memory." This system of buffering sample starts in RAM and
> spooling the rest up from disk is internal to Logic and is like the system in
> other streaming samplers -- or DAWs which playback audio tracks for that
> matter. 

Okay. Fine. I know this sample streaming exists in the OS 9 version of Logic
(enabled via the extension). I remember trying it in OS 9 with my stock
450MHz G4 a while back and it just couldn't do it. Cymbal decays were a
mess, etc.

I have wondered what was going on with this streaming virtual memory in OS X
(as there is no mention of I that I can find in the prefs -- maybe I'm
missing it?) So, is there a pref to turn it on? Or, does Logic (in OS X)
invoke its "VM" when it thinks it is needed?

> In OSX (or WinXP ) Virtual Memory is the name given for memory paging to disk
> and is another matter altogether. The use of the same name for OS paging and
> sample streaming has led to our misunderstanding.
>
> You are correct that as soon as you run out of physical ram the buffering of
> samples fails and you get crackles on playback of EXS24 streaming samples. It
> is also correct that 32 bit Sample loading is nothing to do with Emagics
> "Virtual Memory" - ie sample streaming.

Other than it consuming more RAM, right? And when I ran out of RAM, I don't
know if LOGIC'S streaming kicked in or not. Activity Monitor just showed the
RAM get gobbled as Logic rejuggled the resources of my loaded song when I
went from "normal" to "32 bit float", and I subsequently experienced the
crackling and crashing. That's why -- at the expense of a greater CPU load
-- "normal" SEEMS to be working for me the best.

Just to request again, how is Logic's virtual memory streaming invoked?? By
the user? Or automatically by the app?

Cordially Yours ;)

MacDuff

Re: [EXS] CPU overloads

2004-07-30 by digitalmechanics

--- In exs-users@yahoogroups.com, Mac Duff <dropzone@r...> wrote:

> Just to request again, how is Logic's virtual memory streaming invoked?? 

Same way as it is in 9 - EXS pannel -> Options -> Virtual memory - make sure 
if the checkmark is ticked. Also there are some useful settings there, which I 
recommended to tweak in my first reply to Julie.

Best,

Andy

[EXS] Virtual Memory (was: CPU overloads)

2004-07-30 by Sean McCoy

At 11:51 AM 07/30/2004, you wrote:
>Just to request again, how is Logic's virtual memory streaming invoked?? By
>the user? Or automatically by the app?


Once you enable EXS's Virtual Memory, each and every sample will load a 
portion of its head into real memory for fastest triggering, while the rest 
of the sample will stream from disk. The size of the sample head is 
determined by how you set the Virtual Memory parameters. Faster systems can 
get away with smaller sample heads loaded, resulting in greater load capacity.

OSX's use Virtual Memory is much less predictable, and is something I'm 
trying to come to grips with. According to the following document,
http://docs.info.apple.com/article.html?artnum=107918 OSX's VM is "a system 
of putting information in RAM or caching it to your hard disk as needed." I 
haven't been able to find out exactly when and why OSX finds it "needed" to 
use the swap file, as the Activity Monitor on my system seems to be 
indicating the use of Virtual Memory before my physical memory is 
full---but based on the above definition, that doesn't necessarily mean 
it's not still in physical memory. Obviously, for us sample users, whether 
information is in RAM or cached to disk is of critical importance.

I'm right now in the process of testing my system, which has 5.5 gig of 
RAM, to see if I can figure out what's going on.

Sean McCoy
Oregon Sound Recording

Re: [EXS] CPU overloads

2004-07-31 by Murray McDowall

>
>
> Other than it consuming more RAM, right? And when I ran out of RAM, I don't
> know if LOGIC'S streaming kicked in or not. 


Streaming of samples is on only if you have Virtual Memory checked in the EXS24
preferences. If you have this checked you can have huge sample libraries
loaded. If it is off then you will be limited to some proportion of your
physical RAM - try to load a 2 GB piano sample set into your 1GB or RAM and you
will get crackling at best and a lot of disk activity when you try to load it
as most of it ends up in the system swap file. 

>
> Activity Monitor just showed the
> RAM get gobbled as Logic rejuggled the resources of my loaded song when I
> went from "normal" to "32 bit float", and I subsequently experienced the
> crackling and crashing. That's why -- at the expense of a greater CPU load
> -- "normal" SEEMS to be working for me the best.
>
> Just to request again, how is Logic's virtual memory streaming invoked?? By
> the user? Or automatically by the app?



See above - switch it on in EXS24 prefs - I have 1GB of RAM and I leave it on
all the time.

Regards,
Murray

Re: [EXS] CPU overloads

2004-07-31 by Mac Duff

Hey folks :)

Ya know, when Andy pointed out where the settings for EXS virtual memory was
located in the UI, I pulled one great big fat "DOH!!" I tried it out A LONG
time ago before my Sawtooth's G4 upgrade (still in Mac OS 9) and it just
didn't work for me. Since I got the upgrade and moved to OS X and Logic Pro,
I FORGOT about digging through that preference (I'm usually hitting the
"Edit" button in EXS and saving EXS patches via the Edit window's file menu.
So, I rarelygo into the UI's Options menu.

ANYWAY, I started messing around with the VM settings last night, in the
hopes of being able to use 32 bit float to ease off on the GPU. It crackled
a bit on my 1.3GB bank. For now, I'm returning to the setup that worked for
me (SR conversion normal, Sample Storage original, VM off, 512 I/O buffer,
medium processor buffer, no OS X pageins with 2GB physical RAM) and maybe
play with it more when the workload eases up.

Thanks to you guys for clearing some isues up for me -- mainly to remind me
of WHERE EXS' VM prefs reside! :-D


MacDuff
Show quoted textHide quoted text
On 7/31/04 9:21 AM, "Murray McDowall" <murraymc@...> wrote:
> 
>> 
>> 
>> Other than it consuming more RAM, right? And when I ran out of RAM, I don't
>> know if LOGIC'S streaming kicked in or not.
> 
> 
> Streaming of samples is on only if you have Virtual Memory checked in the
> EXS24
> preferences. If you have this checked you can have huge sample libraries
> loaded. If it is off then you will be limited to some proportion of your
> physical RAM - try to load a 2 GB piano sample set into your 1GB or RAM and
> you
> will get crackling at best and a lot of disk activity when you try to load it
> as most of it ends up in the system swap file.
> 
>> 
>> Activity Monitor just showed the
>> RAM get gobbled as Logic rejuggled the resources of my loaded song when I
>> went from "normal" to "32 bit float", and I subsequently experienced the
>> crackling and crashing. That's why -- at the expense of a greater CPU load
>> -- "normal" SEEMS to be working for me the best.
>> 
>> Just to request again, how is Logic's virtual memory streaming invoked?? By
>> the user? Or automatically by the app?
> 
> 
> 
> See above - switch it on in EXS24 prefs - I have 1GB of RAM and I leave it on
> all the time.
> 
> Regards,
> Murray\

Re: [EXS] CPU overloads

2004-08-01 by Hendrik Jan Veenstra

On a fine day, 29-07-2004, Mac Duff wrote:

>On 7/29/04 12:18 PM, "Julie Larson" <julielarson@...> wrote:
>
>>  Hey Mac,
>
>>  I have a question here though.  Is it standard for an
>>  instrument that has multiple layers for....say keyswitching, to use
>>  polyphony for all layers even though only one layer is playing?
>
>>   I play one note and the polyphony use is 4.  Is there something I'm
>>  missing?
>
>According to the "Used" indicator on my EXS24's interface, no. I have a
>big-ass bass instrument with finger, slap, acoustic and distorted samples,
>all key switched. When I play one note, however, that event only consumes
>one note of polyphony. So, I don't know what's happening there... Unless
>maybe the stereo sample is actually two mono samples panned and therefore
>two voices??!

The answer to Julie's question is: it depends.  If you have multiple 
layers and switch between them using velocity (the usual way of 
layer-switching), only the currently triggered layer is played (or 
more if you've applied x-fading).  This is all common sense of course.

However, if you've set e.g. Controller 1 to determine "Sample Select" 
instead of velocity, so that the mod-wheel cycles through the layers, 
_then_ all layers are played at all times, no matter how many 
actually sound.  If you think about it a bit, this makes sense as 
well: when using the mod-wheel, you can access a layer at all times 
and there's no way for the EXS to know when that will happen, so the 
best it can do is trigger all layers at note-on.
Besides, if the EXS only triggered a layer when it's really needed, 
you would always hear the attack-phase of all layers as soon as 
they're accessed.  In the mod-wheel-triggering situation this 
generally is not what you want -- you want to move smoothly between 
layers, and for that it's (again) necessary for all layers to be 
triggered at note-on time.

I'm not 100% sure about key-switched layers but it's easy enough to 
try out for yourself: make a 4-layer instrument and set up 
key-switching for all layers. Play one note and watch the polyphony 
indicator on the EXS: does it say 1 or 4 (probably the latter)?  I 
think I remember that using _anything but velocity_ for Sample Select 
causes the EXS to trigger all layers at all times, so there you go...

-- 
Hendrik Jan Veenstra   h @ k n o w a r e . n l
Omega Art: http://www.omega-art.com/

Move to quarantaine

This moves the raw source file on disk only. The archive index is not changed automatically, so you still need to run a manual refresh afterward.