how the technology VMT works?

afr · Post by **afr** » Sun Jan 16, 2011 12:16 am

would be interesting to understand how this new technology works considering that Kronos has 12 GB of samples and only 1 GB of RAM

practical example, how can I use in a combi or seq 16 EXs sounds?

thanks a lot

EvilDragon · Post by **EvilDragon** » Sun Jan 16, 2011 12:46 am

It works the same way as direct-from-disk streaming in Native Instruments Kontakt.

MartinHines · Post by **MartinHines** » Sun Jan 16, 2011 3:10 am

Tascam GigaSampler (later known as GigaStudio) was the first music product to use hard-disk sample streaming.

When a patch/sound is selected a very small portion of each of the used samples is loaded onto RAM. With GigaStudio for example about 64KB of each sample (equating to approximately 1/5th of a second) is loaded into RAM. This is enough data to support the intial attack of the sound, while the rest of the sample is streamed/pulled from the hard drive.

As far as I can tell, the pianos and audio are the only things being streamed from the SSD. The EX expansions are loaded and running from RAM.

** Update **
I saw the post on PlanetZ. The pianos and drums are streamed from SSD.

EvilDragon · Post by **EvilDragon** » Sun Jan 16, 2011 3:20 am

I'd say that even EXs are loaded partly in RAM and partly streamed, as well. Those drums sound much better than on any workstation before, they sure use some long unlooped samples! Seems like user samples can't be streamed at the moment. Let's hope for an update which would enable this!

Shakil · Post by **Shakil** » Sun Jul 29, 2012 6:02 pm

With OS 2.0 the user samples will be streamed.

Phabius · Post by **Phabius** » Sun Jul 29, 2012 7:51 pm

I believe the Mellotron sounds from EXs 4 are streamed as well. (?)

AntonySharmman · Post by **AntonySharmman** » Mon Jul 30, 2012 12:49 am

DFD (direct from disk streaming) method needs sample duration bigger than 0.5 sec and
all EXs samples with greater duration than this are using DFD method ,all other smaller
samples are loaded only to Ram , since as mentioned the attack portion (.5 sec) of
any factory Rom or EXs sample is always loaded to Ram and this will not change.
Therefore if user's samples DFD method will be supported in OPS v 2.0 then this will concern
only big duration samples and not tiny 5000 loops of mSec that will maintain the same
usage of Ram since all will be loaded as "sound attack portions" to Ram.

1jordyzzz · Post by **1jordyzzz** » Tue Jul 31, 2012 2:04 pm

if you want to know whether your sound is streamed or not, look at the disk light while you're playing it.. if it dims, it's streamed...
AFAIK the streamed sounds i've identified (using the above method) are pianos (sgx pianos) and ambience drums..

AntonySharmman · Post by **AntonySharmman** » Wed Aug 01, 2012 9:00 pm

If you want reliable way to confirm which Factory sample is partially streamed from
SSD then go to Global/Sample Manage and examine all contents and options.
Virtual Memory = DFD , RAM is exclusively to Ram .
Both these samples have been written to Ram since Booting up to their 0.5 sec
duration , the first one as bigger one continue it's rest portion streamed from
SSD when is recalled but second one stops there since is a tiny one ,like attacks
FXs ect , that's why it can't be changed to "Virtual Memory" method.

(500 ms value hasn't been accurately calculated yet but seem to be between 300-600 mS)