H9000 / RNBO public beta

[tbskoglund]

Hi all,

The Eventide H9000 package featuring RNBO export is now in public beta.

Setup instructions can be found here: https://cycling74.com/forums/eventide-h9000-package-for-rnbo-public-beta

Further instructions and tips can be found here: https://cdn.eventideaudio.com/manuals/h9000/2.1.12/content/development/rnbo.html#content-development-rnbo–page-root

Let us know what you think! You can also post questions/issues you may have on the RNBO forum: https://cycling74.com/forums?category=RNBO

[Chuck Zwicky]

I think RNBO could be a tremendous development tool for the H9000, but it it seems that it requires MAX to operate…

MAX is $399 RNBO is $299  is there a development version that doesn’t require a $700 outlay?

 

[moschops]

VERY cool !

 

Going to try it now, so does this mean we can do FFT & granular on the H9000?

Please can you upload some H9000  patches made with this RNBO so we can get an instant fix?

Thank you.

[Puppeteer]

Chuck Zwicky wrote:

I think RNBO could be a tremendous development tool for the H9000, but it it seems that it requires MAX to operate… MAX is $399 RNBO is $299 is there a development version that doesn’t require a $700 outlay?

The RNBO guitar fx pedal package is a good package to start with.

[tomhall]

Chuck Zwicky wrote:

I think RNBO could be a tremendous development tool for the H9000, but it it seems that it requires MAX to operate… MAX is $399 RNBO is $299 is there a development version that doesn’t require a $700 outlay?

FWIW you can demo both of them (no restrictions) for 30-days, even export to the h9k

There’s also a subscription which gets you both for $19.99 a month

[tomhall]

Max/rnbo~ patches and rnbo~ made H9K Algorithms Here

[peff]

I’m running into an issue with the RNBO export.  The process builds and receives the binary, but doesn’t quite make it to the H9000 transfer.  Is there a network or firewall setting that needs to be altered?  I have tried both wi-fi and ethernet connections as well as both DHCP and manual IP assignments. Vsig and Emote both connect fine.

macOS 15.5

max 8.6.5 & max 9.0.7
rnbo v1.2.7-eventide.0
eventide H9000 1.0.4-beta3

H9000 FW: 2.2.11

This is the report from the Export Log:

Starting Code Generation
Generating Objects
Compiling Patcher
Printing CPP Code
Fetching platform description
Requesting cloud compile
Uploaded generated code and started cloud build (id: 59860)
Waiting for cloud build
Retrieving cloud build
Downloading compiled binary
Sending binary to target
connect EHOSTUNREACH 10.0.1.191:9000 – Local (10.0.1.127:62583)

[tbskoglund]

Sorry to hear you are having issues.

Did you ever have the export working or is this your first time trying the export?

In your Mac settings, go Privacy & Security > Local Network and make sure Max has this option enabled.

Let me know what happens.

[peff]

tbskoglund wrote:

Privacy & Security > Local Network and make sure Max has this option enabled. Let me know what happens.

This is the first time I’m trying out the export to H9000, and the privacy & security setting was the problem.  It’s working now – Many Thanks!

 

Max 9 does not seem to let me downgrade to RNBO v1.2.7-eventide.0 – after the downgrade is performed, package manager says that  RNBO 1.3.4 is installed.  However, the export of Tom’s mid-side patcher works fine on the H9000.  I’ll experiment a bit more with max 9/RNBO 1.3.4 and see what happens.

Max 8.6.5 does allow the downgrade to RNBO v1.2.7-eventide.0 and the export was also successful.

 

 

[peff]

There has to be a physical limit to a RNBO DATA buffer size, but what is the practical limit on the H9000 (assuming that only one algorithm is running on a processor/fx-chain)?  8 million samples? 16 million samples? Or is there a limit per algorithm instance?

i have a patch that’s stable with a large data buffer running at 48k. I’m wondering how far I can push this at 96k

 

 

[Puppeteer]

I’m wondering the same thing as peff.  How big can we make the buffer?  How much delay memory is available for a rnbo algorithm?

How high have you pushed it @peff?

[Puppeteer]

Just looking at VSIG, I note the dlysmp2n has a maximum sample size of 720 seconds at 48kHz.  The length gets divided by the number of channels and halved if you go to 96kHz, so that is perhaps a guide for how much we can use.  By my calculation thats 34,560,000 samples available, each of 32 bits.

The data structure used by ~gen in Tom’s circular buffer example (massive thankyou BTW) uses 64 bit floats.  According to some documentation I found the H9000 uses 32 bit floats as its maximum bit depth, though not sure if internal data structures are bigger than that or not.  I’m not sure if there is a conversion going on in creating the buffer on the H9000, or if a 64 bit array element is built from 2 x 32 bit bytes in the H9000, but conservatively we should have 17,280,000 samples available to ~gen via RNBO.

I have no idea what happens if we exceed the RAM cap. Probably nothing good.

[peff]

Puppeteer wrote:

Just looking at VSIG, I note the dlysmp2n has a maximum sample size of 720 seconds at 48kHz. The length gets divided by the number of channels and halved if you go to 96kHz, so that is perhaps a guide for how much we can use. By my calculation thats 34,560,000 samples available, each of 32 bits. The data structure used by ~gen in Tom’s circular buffer example (massive thankyou BTW) uses 64 bit floats. According to some documentation I found the H9000 uses 32 bit floats as its maximum bit depth, though not sure if internal data structures are bigger than that or not. I’m not sure if there is a conversion going on in creating the buffer on the H9000, or if a 64 bit array element is built from 2 x 32 bit bytes in the H9000, but conservatively we should have 17,280,000 samples available to ~gen via RNBO. I have no idea what happens if we exceed the RAM cap. Probably nothing good.

Thanks for that information about dlysmp2n.  I’ve managed to induce crashes with dlysmp2n VSig patches, but I do have one running stable that is 12 seconds x 4 channels– so max of 96 seconds at 48k.  To help with stability, I keep this algorithm isolated on its own fx chain.  I was running into issues when I had other delay algorithms running in series with the dlysmp2n based algorithms.

I’m encountering something similar with a RNBO patch that has data object set to 3,840,000 samples that constantly records incoming signals.  On its own, it’s pretty stable.  When I have other algorithms (Eventide native/Vsig or RNBO), in the FX Chain, the chain will occasionally crash.

 

 

Attachments:

You must be logged in to view attached files.

[peff]

I’ve been tinkering around a bit with the Max9/RNBO export.  Here’s an algorithm for a mono in/stereo out granular thing.  It’s fairly stable, but at certain settings you can hear the processor getting taxed.  It’s based on the max granulator~ object, and it’s difficult to determine how many simultaneous voices it’s using.  It’s a lot larger than Vsig .9ka files and loading seems to take a long time.  The first time I tried it, it didn’t properly load.  I’ve attached the binary below, but try at your own peril.  When it loads, the H9000 UI seems to freeze, but it’s not a crash.  If it freezes the H9000 you can check emote or setup screen to see if the DSP module is offline – this would require a reboot.

There are a couple of issues I’ve encountered with the RNBO param object interpreted by the H9000.  First is the fact that knob values do not translate the way they would in Vsig.  You can’t set a %2.2f message to get a higher resolution for the knob, so params are set with a @min 0 @max 100 then scaled to 0. & 1.  You can manually enter a fractional value via emote.  Also the RNBO [param @enum 0 1] enumeration attribute is interpreted as a knob with three positions 0 0.5 and 1 rather than just a 0 and 1.

The algorithm is only set to work at 48k sample rate.   It has a control for rec-gran.  When this is set to 0.5 or 1.0, the algorithm is recording to the data object of 192000 mono samples.  When rec-gran is turned to 0, it stops recording and enables the granular playback.  It’s preset with a stable usable setting, but when you increase the rate or alter the window size or decrease the pitch, you will hear processor taxing artifacts.  There’s a second enumerated control for position modulation type: modmode 0 is no modulation which enable the position control.  modmode 0.67 or 1.33 is a phazor ramp defined by the mod start and range knobs.  modmode 2.0 is a triangle wave that will apply a forward-backward position modulation.

 

Attachments:

You must be logged in to view attached files.

[peff]

Grnbo V.9ka

https://www.dropbox.com/scl/fi/rrjf75p931ug685mbbi3k/GrnboV.9ka.zip?rlkey=g15i09jhsi6rb13mwrqnfhb7k&dl=0

[Puppeteer]

The UI translation is the biggest issue at the moment IMHO with RNBO exports.

You can use @steps in max to make the H9000 knobs work like proper knobs. I’m not sure if they implemented switches yet for 2 choice switches/knobs.I vaguely remember a discussion about it, but can’t find it at the moment.

[Puppeteer]

Duplicate post. Please delete.

[Author]

Posts