Korg Forums

Let's hope that Aureo is not in any hot water with KORG. He is a valuable asset to the community.

I´m just impressed about the Krome sampling feature through SD card. It's direct streaming from card and the important is to see how able is the engine to manage polyphony in real time. Because a SD card is serial access of data the retrieve of multiple files in different clusters is a bit slower. So to avoid this problem is important on the Krome the implementation a good RAM I/O pre caching of data to avoid latencies on the retrieve/playback.

Congratulation again to Korg engineering !!!!

Given a slow transfer speed of SD cards, there's no way one can play 60-120 voice polyphony streaming off SD cards. On the removed video, the guy plays a single note.

Chriskk wrote:Given a slow transfer speed of SD cards, there's no way one can play 60-120 voice polyphony streaming off SD cards. On the removed video, the guy plays a single note.

If you pre cached the beginning of each sample using a fast RAM pre load cache in the same way as the serial ATA hard disks I/O latencies are avoided. In this approach you can play samples direct from a SD card at real time without problems. This is the way that use the new Krome, I think but are only assumptions.

From what I could see, triggering samples from the SD card is analogous to hitting the START/STOP button in Sequence Mode on the M50 in order to play back a sequence. Whatever method is used for one would be used for the other.

Only it looked like they were triggered from the keys of the krome much like RPPR. That would be very handy indeed.

Yes, you're right. Wonder if there will be similar functionality, e.g., the choice of looping samples continuously. It looked like hitting the key triggered the sample once. Repeated key presses started the sample from the beginining. Don't know whether there will be a checkbox to enable looping.

I don't know where did you guys get that the samples are streamed from the SD card. On the video, Korgman clearly load the samples from the SD card to the PCM user banks at boot time, then the sample is selected at oscillator level, but selecting the Option Bank M (Which I assume are MONO samples).

Now, the 3.8GB sampling memory makes sense. Lots of room for user samples.


loadbanks by MartinOcando, on Flickr


optbanks by MartinOcando, on Flickr

mocando wrote:I don't know where did you guys get that the samples are streamed from the SD card. On the video, Korgman clearly load the samples from the SD card to the PCM user banks at boot time, then the sample is selected at oscillator level, but selecting the Option Bank M (Which I assume are MONO samples).

Now, the 3.8GB sampling memory makes sense. Lots of room for user samples.

Korgman did not load the samples into RAM. The Krome doesn't have 3.8GB sampling memory. It has 3.8GB of PCM ROM

Chriskk wrote:

mocando wrote:I don't know where did you guys get that the samples are streamed from the SD card. On the video, Korgman clearly load the samples from the SD card to the PCM user banks at boot time, then the sample is selected at oscillator level, but selecting the Option Bank M (Which I assume are MONO samples).

Now, the 3.8GB sampling memory makes sense. Lots of room for user samples.

Korgman did not load the samples into RAM. The Krome doesn't have 3.8GB sampling memory. It has 3.8GB of PCM ROM

I tend to doubt that. Obviously the specifications are not complete, or final, and I won't be surprised if those 3.8GB of PCM ROM where in fact PCM RAM. Or maybe 800MB of PCM ROM and 3GB of PCM RAM. Whichever seems more likely.

By the way, I just checked the info page and specs sheet, and there is no mention to PCM ROM or RAM. They say only PCM Memory. That's it.

We need to wait for the circuit schematics to have an idea about the Krome architecture. Updating "PCM Banks" for me means the "pre loading or caching" of the beginning of the samples in a small RAM (maybe 64MB). After initial playback due the high speed of today serial devices there are no latency to stream from SD card. My point of view is this approach of "pre caching" the beginning of PCM waves before a "note on" event to avoid initial playback latency. This simplifies the data bus design allowing the use of a serial storage device (SD card and ROM banks) for samples. Is impossible that the Krome incorporates a RAM bank of 3GB or more for sampling purposes because this memory is in reality a very small RAM (no more as 64 MB) acting as "cache of PCM data".

Welcome to the new "EDS-X (Enhanced Definition Synthesis-eXpanded)" architecture, more cheaper... more powerful...

I suspect that because this is not mentioned in the spec, this was a developers tool demonstrated by mistake, and someone has gotten in big trouble over this.

Regards
Sharp.

Optional bank could be flash memory like the nords and yamaha's. No RAM, but still optional samples. Has nothing to do with the 3.8 GB ROM.
Still waiting for some real info from Korg though.

Sharp wrote:I suspect that because this is not mentioned in the spec, this was a developers tool demonstrated by mistake, and someone has gotten in big trouble over this.

Regards
Sharp.

Well, I only hope it will make it through to the final production firmware. We already saw it, and there are many (me included) deciding for the Krome over something else just because of this feature. I will be very disappointed if they clip it out at the final stages.