Korg Forums

SpIdErWeB · Posted: Wed Mar 16, 2011 4:14 pm Post subject:

EvilDragon: I know some bugs have been solved (I worked with NI, so I'm fully aware of that LOL), but it doesn't mean it will work seamlessly. Doing a bank of few presets, even few hundreds is way different than handling several plugins, with several samples, for thousands of presets...

Doing some test at home for personal use is way different that doing test for production release...

If you don't believe me, Ok then.. show me your setup and give me all specs of your configuration, with 12+ Gb of multi-samples, 9 different engines (not only Kontakt) using and sharing these samples, some using them with streaming, with effects and thousands of presets made with direct access. Gimme your setup, I will crash it in less than 2mn... because it won't work. I know it, because it has been my job to know it...

Now you can claim everything you want on forums, I will just tell you: Show me, prove it! If you can't prove it, you can't say it...

Phil

EvilDragon · Posted: Wed Mar 16, 2011 4:37 pm Post subject:

Naturally hardware workstations still have better integration (but I wonder for how long yet), I'm not denying that.

But I just wanted to say that indeed it IS possible to get seamless patch switching with Kontakt, and I countered the statement that sample sharing isn't working correctly (food for thought - it's always been working correctly for me, but I'm using Reaper, whereas users of some other hosts experienced memory sharing issues Smile

)

That's all I did say, and I know for a fact that what I said is correct, nothing more, nothing less.

SpIdErWeB · Posted: Wed Mar 16, 2011 5:55 pm Post subject:

As I said, if it works for you at home for few presets for a song or project, it doesn't mean it will be ready for production release with thousands of presets for heavy duty use by people around the world...

The only way it could work, would be to replicate what Korg did with the Oasys/Kronos environment. Build a VST Host application, with integrated sample player (such the HD-1), that will go through samples allocation and voicing management. Then, you'll need to development your own memory management for your application (if your OS allows you to do so) and your own drivers to guarantee stability of the entire signal path... so you'll end up designing your own audio interface...

And then you'll realize that most of VST plugins aren't all well coded, so you need to development your own Synthesis engines... such Korg did with AL-1, MS-20, CX3, etc...

Then, you'll run thousands of test, de-buging, sound design and voicing before to be sure the system is stable enough to start mass production; of course you'll get surprises...

So, in the end, you'll end up with a development pretty close of the Oasys and you definitely can't do it by yourself and you'll need a full R&D team and freaking budget...

I'm not saying it's not possible thought...

Phil

EvilDragon · Posted: Wed Mar 16, 2011 6:39 pm Post subject:

But I wasn't talking about mass production at all... I just mentioned that some things Kronos touts do work in software realm, as well. Not as integrated, sure, but it's happening alright.

SpIdErWeB · Posted: Wed Mar 16, 2011 7:25 pm Post subject:

I won't waste my time, so as said: Prove it, do a youtube video... I can't wait to see it... LOL!!

JimiJones · Joined: 12 Apr 2011 Posts: 18

Hello Dan;

Let me get this straight, you quoted that you can unload specific sounds in an Exi expansion library, something that could not be done on my previous OASYS, on the OASYS you had to unload the entire Exi expansion, lets take the "Brass and Woodwinds" for example, you are stating now that on the KRONOS we will be able to unload a specific sound such as the "Soprano Sax" or as the person stated the "Trumpet" without unloading anything else, I ask you again Dan Phillips is this true...?

Jimi Jones
Http://www.myspace.com/jimijones21

EvilDragon · Posted: Fri Apr 15, 2011 5:26 pm Post subject:

I'm not Dan, but that's true. Laughing

Gargamel314 · Posted: Sat Apr 16, 2011 8:41 pm Post subject:

zahush76 · Junior Member Joined: 09 Mar 2011 Posts: 98

RC-IA · Posted: Wed Apr 20, 2011 6:38 pm Post subject:

Dan said 3/4 (75%) is FULL ! With korg EXs which you can unload to get 1 gig free

EXer · Posted: Sun Jun 19, 2011 1:37 pm Post subject:

panrixx · Posted: Sun Jun 19, 2011 2:03 pm Post subject:

For those of us, like myself, that do not speak/read french here is a VERY ROUGH translation. You'll need to work out the very non-english phrases the best you can. Hope it helps.

Translation:

The processes can work only with infos figuring in the card of the memories. The memory Flash of the Motive XF appears in the card of the memories. The data of the SSD do not appear in the card of the memories, and have therefore to be copied in a memory (the RAM) to appear in this card, in order to be able to be used by the processes. These copies are ordered by the BONE, and realized in a manner material by the inspector of the SSD. These copies are realized by pads (multiples of 4KO) and ask every time an interruption of the cycle multitâche. All that must be taken into account in the measure of the access time.

The mask disc in RAM (that preserves the last attained pads) allows exactly to avoid all these interruptions to repetition, that are heavy to manage, and that do not have place in the Motive XF where Flash is used directly since the card of the memories.

The card of the memories is erected by the memory inspector at the time of the Boot. This card contains all the memories that are accessible 8-bit bytes by 8-bit byte. Between every memory, there is emptiness (that I drew in hâchures) that allows detecting the floods, therefore the addressing errors. In the card of the memories of the Motive XF, Yamaha must arrive to all to put in only 4GO (limit of the addresses on 32BITS) and is for that that the Motive XF accepts only 2GO maximum of Flash, and 128MO maximum of dedicated RAM to the sampling, while the Motive X'S did not have any Flash and could accept to 2GO of dedicated RAM to the sampling. Fairly, I think that in the Motive there is equally another RAM for the system, but the question is not there. The card of the memories of the Kronos is on 64BITS, and as these days no one has the usefulness of a such cuts, only the 36 indeed 40 bits of weak weight actually are wired in the memory inspector, and as more the addresses are long, and more the routage of the 8-bit bytes is slow, one has any interest to minimize the physical size of the addresses, therefore on the 64BITS of an address today, there are some at least 20 Who will always be to zero. Therefore even if one was sufficiently fortunate to have 1TO of fleas of memory Flash under the hand, it would be impossible to do them to hold in the card of the memories, less in his current physical implementation.

Let's come back to the SSD. The SSD is composed from sectors (512 8-bit bytes) of memory Flash + a plug of 32 or 64MO according to the models. Only the plug figures in the card of the memories, the memories Flash of the SSD do not there figure, and will never figure in the card of the memories. Besides, the stocked data in these memories Flash are not accessible 8-bit bytes by 8-bit byte, but sector by sector. If one wants to read a 8-bit byte in a SSD, one is obliged to ask the access to all a sector: the inspector of the SSD then will copy the entirety of the required sector towards his plug.

As for the BONE, it manages its potential memories thanks to the MMU, and the MMU of the processors Intel (and compatible: AMD, THROUGH. ..) cuts the memory in pads of 4KO called pages. Therefore the BONE always will work in pages, and will do the exchanges page by page between the memory and the disc, therefore the discs always will be attained by group of 8 sectors (for 8 sectors = 1 page) reason for which the manufacturers of SSD are campaigning to bring back the granularity of the addresses of storage to 4KO, instead of the 512 current 8-bit bytes. Why do it that? Because as a SSD works by 8-bit byte pads, his wiring is conceived of such sort that the 512 8-bit bytes are read (or written) in parallel, and non-the one after the other. If the granularity of the addresses of past storage to 4KO, they will be able to construct SSD 8x more parallel , therefore a lot quicker ones.

The Flash of the Motive XF is actually accessible 8-bit byte by 8-bit byte, in contrast to the Flash of a SSD. Even if the fleas Yamaha and the fleas of a SSD were the same, on a SSD they would be wired in parallel for more of performance at the time of the sector copies, which did not do Yamaha, the Flash of the Motive not being intended for archiving but to the direct usage, therefore 8-bit byte by 8-bit byte.

I visible that and there that accuse Yamaha that the load/safeguard of big files in Flash is long, and in conclude that Flash Yamaha is low-of-range. Nothing is more false: the Motive is a music instrument, and asks therefore an access 8-bit byte by 8-bit byte to the content of Flash. As it was constructed as that (to privilege the music) even the copies of samples do themselves 8-bit byte by 8-bit byte, and appear necessarily very long, while these same fleas in a SSD would have been wired in parallel and would have gone to the copy of big files of samples, but would have been unusable in usage "musical" of these samples.

EXer · Posted: Sun Jun 19, 2011 2:10 pm Post subject:

These automatic translators are horrible.

At least I would say 'memory map' and not 'card of memory'...

panrixx · Posted: Sun Jun 19, 2011 2:12 pm Post subject:

ScoobyDoo555 · Posted: Sun Jun 19, 2011 2:32 pm Post subject:

LOL - was going to say "well done" for doing THAT much translation. Cool

Would have made sense to post it all in English (as English is the predominant language on this forum) in the first place. rather than take the p1ss Wink

_________________
Yamaha SY77 & KX88, SSL Nucleus, Korg Kronos 61, Wavestation A/D, Access Virus B, Roland XP30, DeepMind12D, System 1m, V-Synth XT, Focusrite Red16Line, Unitor 8, Akai S3000 XL, Alesis Quadraverb+, Focal Shape Twins, Full fat iMac, Logic Pro X, ProTools 2021, loadsa plugins.