PAB sample set with release truncation

[SMann]

Hello Csaba,

On December 1st I made a forum post asking about using the PAB sample set with release tails truncated:

http://www.crumhorn-labs.com/forum/viewtopic.php?p=27146&highlight=#27146

Your reaction to the idea of doing this was (understandably) negative as can be seen from your response. Yet we now have the Bach Passacaglia demo with release tails truncated to 250 ms and the results with some artificial reverberation added are quite good. I am in complete agreement that release tail truncation is not desirable and as a general rule should be avoided.

As I mentioned in my previous post my Hauptwerk computer is a Quad core based PC with 8 GB of RAM. I could not upgrade it without significant expense and have no plans to do so at this time.

Apparently at 250 ms truncation it was possible to load 49 stops into 3.4 GB of RAM. So my question is, how many stops can be loaded into 8 GB of RAM using the same approach? If I were to purchase this sample set I would want the maximum number of stops possible to be playable at any given time. Suggestions have been made that multiple configurations of the set could be created and loaded for playing music of different styles. I have not had to pursue this approach with other sets because of their smaller size. However, and someone please correct me if I am mistaken, taking this approach is not a simple matter. It would be necessary to use the "load organ, adjusting rank audio output" screen to manually select the stops each time a change was required. At that point the entire sample cache for the instrument would have to be rebuilt. As we all know, this is very time consuming and I for one would not want to go through all of that each time I wanted to play a different "version" of the set.

So I repeat the question from my earlier post. How many stops can be loaded into 8 GB of RAM with release tails truncated? Csaba, I would very much appreciate it if you or someone else would run some tests and post memory requirements under different loading scenarios (multiple loops, etc.) when truncation is used. Perhaps you could update the RAM requirements page on your website with this information.

Eventually as hardware prices continue to fall (and my finances improve!) I will no doubt upgrade my sytem to at least 16 GB or more and this will not be an issue.

As I have heard the demos and read Rob Stefanussen's impressions of it I would seriously consider purchasing the set but want to know what I will be able to achieve given my current situation before doing so.

On another topic, I would like to ask if there is a user manual for the set that you would be willing to make available for download. I am intrigued by the innovations and special devices you have introduced and would like to learn more about them.

Best regards,
Steve Mann

[mdyde]

Hello Steve,

As suggested in this post, you can currently have two indepent sets of rank options (without regenerating the caches) by using the stand-alone and MIDI sequencer configurations:

http://forum.hauptwerk.com/viewtopic.php?t=4129

[Csaba Huszty]

Hello Steve,

it takes some time to do all this, but as a start I checked loading:

- all stops with 1st loops and all release samples truncated to 250 ms (option 4): takes up 4.54 GB of RAM.

- all stops with all loops and all release samples truncated to 250 ms (option 4): takes up 9.18 GB of RAM.

All the best,
Csaba

(I will edit this post once I get through the other scenarios.)

[SMann]

Thanks Csaba for taking the time to do this. It is much appreciated!


Regards,
Steve

[stephenphillips]

Hello Steve,

Perhaps the following will be of use (hot off the press!):

Have success today loading the full organ: 8 GB RAM Windows Vista, default samples, first loop, all releases, all truncated @350ms. Sounds good, audio routed through Logic's Space Designer, Roman Cathedral setting (3.4s IR).

I am not sure how to confirm RAM usage under these conditions. Anyone have that computer know-how? Until I can do this, I am a bit loathe to strike all my good work and start again. There is always the @500ms option to consider!!

Indeed it is a far-from-trivial exercise to experiment with different loading solutions. Hauptwerk (users) would benefit greatly from parallel-ODF loading options for a given sampleset.

All the best,

Stephen.

[SMann]

Thanks Stephen for taking the time to do this. I am running Windows XP Professional 64 bit which (and others please correct me if I'm wrong) requires a somewhat smaller memory footprint than Vista, so I should be able to duplicate the configuration you described. The approach I've been using to determine how much memory a sample set occupies is to run Hauptwerk with no set loaded, bring up the processes tab of task manager (Ctrl+Alt+Del) and note the memory being used by Hauptwerk.exe. In my case that number is 121,112K. Once a sample set is loaded the "before and after" difference in memory use by Hauptwerk.exe should be the amount being utilized by the sample set. No doubt this is an oversimplification of the manner in which Hauptwerk utilizes memory but it gives at least a rough idea.

I would be very interested to know what those numbers are on your system with the PAB set loaded in the configuration you described. Regarding the 500 ms truncations, you may find (assuming you have enough RAM to load the set that way) that the sound quality isn't necessarily better than 350 ms. In fact it may not sound as good. I've done a lot of experimentation with truncation of releases over the years and it seems that trial and error with a given set is the best approach.

I would like to stress to everyone reading this that truncation should be avoided if at all possible. This is especially true in the case of very wet sets. I am considering using it here only because I am intrigued by the PAB set (which is fairly dry apparently) and would consider purchasing it only if I could load the entire instrument.

Thanks,
Steve Mann

[Csaba Huszty]

There is a task manager replacement free software called Process Explorer, which can show (and do) various things Task Manager cannot. Once you installed it, you can customize what it should show you, and you can accurately see how much of RAM a process takes - and tons more. You can also suspend an application or restart it with a mouse-click which comes very handy when working with many applications at a time. It is an advanced tool, and was developed for the really serious people (so probably most of you already use it). For those who does not yet, you can make accurate RAM measurements with it:

http://technet.microsoft.com/en-us/sysinternals/bb896653.aspx

[Stefanussen]

Right now I'm using PAB w/ 250ms truncation. The truncation does make a noticeable difference in RAM consumption. Sonar / Pristine Space (PS @ 250ms latency). The IR is the FrB40, which works very well. I believe I have the reverb set to -8.5 dB.

The above combination yields very good results.

[SMann]

Hi Rob,

What is your configuration of the PAB set with the 250 ms truncated samples? I'm referring to multiple loops, bit resolution, number of stops loaded etc. Also, what is the total memory useage?

Thanks,
Steve