Supercharged Organelle!


#1

I now have a supercharged Organelle :slight_smile:

Important Note

Before reading past this point - what I have done here, voids my warrant on the Organelle, could have caused serious damage to it, furthermore, there is still a risk of damage in long term running.
so I do not advise anyone else to do this, and certainly take no responsibility for them killing their own Organelle

Hope that is strong enough, as I dont want C&G to start getting users crying their Organelle has melted/stopped working.

so caveats done…

‘Factory’ Organelle

before

Supercharged Organelle

Changes:

  • 4 cores :slight_smile:
  • 2GB ram vs 512k
  • Built in Wifi

Im really pleased everything worked ‘out of the box’ including the built-in wifi.
(the WIFI also a long enough internal antenna to place next to the USB ports, so a good signal)

So now I’m using internal wifi and internal sd card (for patches), so a connected Organelle with 2 free USB ports :slight_smile:

I found the cheapest way to do this was to buy an Cubox- i4Pro ~ 150 euro, and swap its microSOM to the Organelle.
this has an added bonus, not only did I get a cheap upgrade, but the Organelle microSOM goes back into the CuBox, so another box to use.
Note: this cubox i4pro came with a v1.3 microSOM , rather than the newer v1.5, which would have meant the wifi would not have worked. I don’t know if this is always the case, or depends when it was manufactured.

Usage

as discussed elsewhere, most PD patches wont make use of the 4 cores - I would need to tailor them, though they will make use of the extra memory… so there may be a small performance gain.
However, Im also running other processes (MEC in particular) and the cores will help with this, as again will the memory.
also as a developer, compiling etc will be improved.

Testing

so only had a limited running so far, but as I said everything works as expected, but Ive not done any exhaustive performance testing… though no doubt I will run a few tests for fun, and compare to another organelle.

Currently, Im running without a heatsink, so Im going to keep a close eye on temperatures and again comparing to another organelle… initial tests, showed it running at 55c, so thats about the same as previously… but I need to check again when I put it under more serious load, and also run it for longer.

I suspect, even if temps rise, the cpu will automatically scale back… so preventing damage.
however, if core temps rise, Ive already noted there is plenty of space for the heatsink that was supplied with the cubox. (I just need some paste to connect it)

anyway, very happy… it was a bit of a risk, but it seems to have paid off handsomely.

UPDATE:
ok, so been running for quite a while, with a patch running constantly, at around 45% cpu on both organelles.
the quad core is actually generally running about 6C lower than the factory unit.
I can only assume this is down to some of the cpu load being put onto other cores. (2-3% typically).
Ive also to be mean, been doing parallel builds on it to raise the temperature (interestingly its critical is listed at 93 vs 83C) and seemed to recover pretty quickly.

anyone know of a really heavy CPU patch, I want to run the factory organelle at really high cpu load so i can see what happens to the temp… I’m assuming it will reach critical and then scale back, but with the governor set to performance I’m not 100% sure!


Organelle software open source?
Suppose I want to add RAM to the organelle
#2

Moreganelle :wink:


#3

Sounds great :slight_smile:

I see you went straight for 4 cores instead of 2. Nice :slight_smile:

Does Organelle already have the 2gb of memory installed? Or you takw that frpm the MicroSOM you mention? Sorry this is new territory to me.


#4

organelle microSOM has 512kb.

the MicroSOM is complete, comes with cpu, ram and wifi, its all on one board , nothing can be removed/altered.

think of it a bit like a rPI board, its all soldered/surface mounted components, its a bit different though to a pi as the ‘hardware’ is on a separate ‘carrier’ board - due to it being a ‘manufactures’ solution, rather than aimed at end-users.

so the only thing you can do is detach/replace the microSOM from the carrier board.


#5

Thanks but how it is going to be used with Organelle? You take out the Micro SOM from Cubox4 and put it into organelle?

I think 150€ is bit high price. If one don’t want this, it is still possible to use the 4 cores right?


#6

yes.

I dont understand?

to get 4 cores, you need to get a new microSOM… and this was the cheapest way I found to get a quad core.

there are other cheaper options. an alternative i considered was a dual core no wifi for 66 euro.
( I thought this was good, but i wanted wifi to free up a usb slot)

theoretically, there are a range of options… including dual or quad core, wifi/no wifi, with or without carrier board.
in particular the options are for Solid Run Hummingbird or CuBox or standalone microSOM.

but, options are also governed by what you can get from suppliers e.g. i dont think i found a good price for dual core with wifi. similarly microSOM only were really expensive, but perhaps a non-european supplier has more options, if you are willing to wait.

anyways, you pays your money and takes your choice (or not :wink: )

important note: I can only say what worked for me (Cubox i4Pro), any other combination should work, but I don’t know. i.e. there is a risk. in particular, Solid Run current microSOM is v1.5. the Organelle is 1.3 which is now not manufactured. the 1.5 should be compatible (except the wifi!) , however I was ‘lucky’ that the Cubox i4Pro that I got, came with v1.3… but Ive no idea if this is common or if it was ‘old stock’.

(btw: correction on cubox i4pro price , I paid 137 euro inc vat, forgot my order had something else on it)

definitive info is here, you are interested in the iMX6 range.


#7

Ah I though that MiroSOM was only needed for the memory upgrade.

But yeah would be great with a super charged Organelle.

Anyway STILL patching here but getting closer to actually put some more serious stuff on to the Organelle and test how much I can actually run of the “homebrewed” patches :clock9:

But great work :wink: Nice to know that these things are possible down the road when i get to the point where it is necessary .


#8

Yeah is a ‘system on a module’

I have also been looking at some others benefits:

  • memory is 64bit rather than 32bit
  • memory 25% faster
  • more powerful gpu ( not that i need this )

#9

This looks cool. Makes perfect sense for a developer setup. However I find that the main beauty of the Organelle is that everyone is using the same closed hardware system. This makes development much easier to spread out on different people, all patches work for everyone and so on. Also I really like to think about the Organelle as something that is not a computer. Or rather get all the it stuff out of the way. For me this is not about the fact that it actually IS a computer, but that it doesn’t FEEL like a computer. Even when pluging in a monitor, keyboard and mouse it is still different enough that I don’t feel like checking Facebook real quick.

Don’t really think this post is contributing much to this thread but I guess I just want to say that I really like the Organelle and hope that Critter and Guitari keeps it going for a long time and not get tempted to change out the hardware just because they can, and then ends up having user develop for two different systems, and that one system can get “outdated”.

When that is said I support all experementation and development around this really cool plattform! Keep on blowing us away @thetechnobear !


#10

I absolutely agree, and said as much in this post recently, a common form is what differentiates Organelle from similar diy solutions.

this ‘hack’ exists for three reasons:

  • I’m doing a lot of development on the Organelle at the moment, which means long compilations.
  • I’m fortunate enough to have a second Organelle which I keep as a ‘user machine’
  • I can’t help myself …
    If an idea occurs to me, I need to work out if it can be done, how to do it, and how to resolve any issues that might occur… I love a challenge. boundaries are there to be explored and pushed :wink:

so yeah, its a special case… a hack, not a recommended upgrade path for all.


#11

I’d like to add, I think i misunderstood all this upgrading the other day. From what I understood, I thought that Organelle had all ready 4 cores and more memory, I just thought that it wasnt being utilized, I didnt know I had to buy/install all kinds of stuff into it.

The reason actually I changed from Axoloti to Organelle was actually to avoid all that modification stuff, as Christopher also said, it is nice t have a closed device that actually feels like an instrument, without having to do all kinds of mods to it. To be honest I don’t care about DIY at all and not interested in it, I care about making music and experimenting with musical algorithms and stuff like that… Not mess around with wires etc.

As I wrote, I thought actually this was something that could be done with the existing device. I didnt think I had to install new memory and so on, so I am not sure I am going down this road.


#12

I made a kind of super ‘Organelle’ too (well, it’s a clone TBH):


#13

Hi there,

I guess you are in the UK, where did you get this from?

Andy


#14

Nope, I live in southern Spain…
but I did order it from the uk… iirc I ordered from newit.co.uk
(there aren’t that many suppliers, and the price varies a lot)

If i was ordering another, id probably try the dual core… as mention on another thread, the quad has a ‘startup issue’, its only minor… but it may be its not an issue with the dual.
and the dual would be fine, unless your doing heavy compiling, and possibly better for thermal reasons.

Id also get the one with built in wifi (though, see my notes about being lucky on this) , this has been awesome for me… as i also use the SD card for patch storage, I have 2 USB ports completely free… handy for a wireless keyboard/mouse and midi controllers. (without having to resort to a usb hub hanging off the organelle)


#15

Aha, so much for my guessing!

I had seen the i4Pro on eBay, looks like from the same company but a tenner more than your link so thanks very much for that.

I was going to get the wifi one, fingers crossed I guess that it will be a 1.3 rather than the 1.5 otherwise I will have a look at building the kernel when it arrives, all fun and games!

The two core is only 15 quid cheaper and out of stock so I ordered a 4 core.

Thanks again for all your help.

Andy


#16

ok, just be aware with the 4 core, you will need to plug in power, remove power (after 1 sec), then plug in again

building kernel, yeah, Ive tried a few times, but there are some issues… but if you have some experience with these things, then we can talk about it.
(but ive lost quite a few hours trying in the past, so honestly… its only going to happen if some one can supply some new input :slight_smile: )


#17

I do have experience building/altering Linux kernels but I’m not an expert.

I know how painful it can be!

So fingers crossed mine is the 1.3.


#18

yeah, you can see my other thread… there are two main issues…

a) we dont have the exact commit and source code used for the organelle/etc build… so we can’t just go from there… and I tried as close as we know, and it didnt work (see other thread)

b) ideally we’d move to the latest kernel possible, which has many fixes - but when doing this , ive not got past uboot, for some reason it doesn’t find the sdcard (though uboot does), this is suppose to be due to partition number changes, but even accounting for that it doesn’t work.

anyway, im sure one rainy day, when my girlfriend is away… i’ll have another go, but its not a trivial thing.
(and im used to building kernels without much trouble)


#19

Just had a look at what I guess is your other thread: Kernel compilation looks like fun!

When you say “latest” are you talking about 3.18, or the nightmare of 4.n?


#20

I have been thinking about the power issue at startup while cooking dinner.

Why do you think you need that power cycle at startup? Is it the same if the board is already warm?