KRONOS RAM UPGRADE 2Go to 4 Go

[NuSkoolTone]

So has anyone verified, that the upgrade actually WORKS (i.e. in use)?

Why did you decide to replace both as a pair instead of just adding one? (My memory fails at the moment, Something to do with it being DDR ram right?)

I'll say this, Even if it is "only" 1GB+ of ram you can end up using, that's still BETTER than 273 (or whatever) mb that was available before! You can actually DO something useful with that amount.

VERY Happy to read this!

[McHale]

did you NOT read the thread? This was discussed... IN THE 6TH POST.

[shap]

Someone should try an ITX board with an i7 processor. Then, the Kronos's polyphony could approach 1,000.

Whether this works would depend on a lot of low-level details of the microkernel implementation. It's not just that the clock and instruction issue rate are faster. It's also that there are more CPU scaling options for the microkernel to manage.

I haven't seen Korg identify their microkernel supplier anywhere. It may be proprietary. In that case, it probably will not have all of the necessary support needed to run on arbitrarily selected processors. You might get away with shifting to D525, because it's at least in the same family. I'd certainly expect that to work before a core-i7.

If you do try core-i7, don't neglect concerns of heat dissipation. You'll want a ULV version of the processor. You might do better in this case with a core-i3.

And needless to say, opening the case voids your warranty, and has been known to cause cancer in laboratory rats.

[shap]

It's much more like an embedded system which happens to use an Intel CPU as the DSP.

Dan: I was going through the tear-down photos that showed up on the web again, and given the cabling I was starting to wonder if the Intel board isn't acting as a slave to the OMAP processor. That is: that your description above is fairly precise. Is this the case?

And if you are free to comment, how is audio getting from the Intel board back to the audio board? Over USB?

[shap]

The speedometer of my car indicates 300 km/h and I will never get to use it

Try pushing that little gas pedal thingy harder. If that doesn't work for you, call customer service.

[shap]

Well of course. If I put 4 GB in Kronos, I would want to use ALL 4 GB that are in there, NOT A MEGABYTE LESS.

That doesn't make sense to you?

We NEVER EVER GET TO USE THE ENTIRE SIZE OF A HARD DRIVE THAT WE INSTALL... HOW IS THIS SITUATION ANY DIFFERENT?

Speaking as an OS kernel developer: because the two situations have absolutely nothing to do with each other. It actually is a bit odd that the UI is reporting 3.25GB. Bald Eagle explains why this is happening. There is a reserved window in the physical address space for I/O addressing, just as he says. PAE mode can resolve that without going to a 64-bit OS, but it slows address translation down a bit to do that, and it's a nuisance to implement the necessary kernel support. Even under PAE mode, you won't get more than 3GB available to any single process (e.g. the sampling process) because of virtual address space allocation issues. The linux kernel uses the rest of the virtual address space.

Addendum: Chrutil has also answered this, and in a bit more detail.

The underlying kernel support to manage large physical page spaces within the virtual address constraints of a 32-bit kernel is a serious pain in the ass. The kernel needs a table of per-page data structures for management purposes. It's a pretty big table, and it's really hard to fit that table into the kernel's virtual address region. Because of this, in practice, PAE mainly gets used in 32-bit systems for the sake of it's extended large-page mapping features and the NX protection bit.

When we looked at this during the Coyotos kernel development process, we concluded that we actually could handle up to 16G of physical memory within our kernel virtual space constraints (we only have 0.5GB of total KVA), but we also concluded that the result was a pretty ass-backwards configuration that would be an ongoing support burden, and that users who needed that much memory would be better advised to shift to a 64-bit platform. We therefore elected not to implement that.

[jimknopf]

In my view all this is just as interesting as practically irrelevant, as long as we get around 1,3Gb available sample space.

I am completely comfortable with that dimension, because all I want is the possibility to get space for one or two big multisample libraries plus some space for smaller instruments and samples. The Kronos is so well equipped, that hardly anyone of us will be in real need for more.

The present 2Gb limitation hurts, because it practically forces you to decide between a fully mounted Kronos and a severely cut down version allowing some sample space.

I will gladly invest that small amount of money for 2x2Gb to get sampling going, besides all the other great stuff on board of the Kronos, and I don't care if I will miss 0.75 Gb this way.

It is essential to me to get things going, and not to reach possible limits.

[shap]

But honestly, there are no "security tabs" or "anti tamper" when you open a keyboard. You gotta be pretty dumb to screw something up like putting in new ram.

Not everybody is as mechanically/electrically inclined as you are. For someone who is comfortable taking cases apart, and keeping track of the screws, and paying attention to the wires, and taking care not to strain the cables (though Korg did a nice job with cable management in this build), it's not a hard upgrade to do. But you have to remember where all the screws came from. Certain other manufacturers (cough, Yamaha, cough) use screws that look very close but have subtly different threads. Sounds like Korg didn't make this mistake, which is great, but there's a reason that the cartoon with the auto mechanic staring at the engine with one leftover part in his hand seems familiar to all of us.

All I'm saying is: try not to call people stupid because they lack the benefit of your experience. And especially, try not to encourage the people who are overly bold but lack the experience to use their brand-new keyboard as their first project. Might be wiser to disassemble an old PC first and see how that goes.

[Lou]

Don't know much here, and I'm not sure if applies to K, but isn't it so that more ram makes things run better?

[shap]

Why did you decide to replace both as a pair instead of just adding one? (My memory fails at the moment, Something to do with it being DDR ram right?)

Two different DIMMs can meet the specification without having identical timings. Even with nominally identical timings, I've seen strange stuff happen on PCs when non-identical RAMs are used on paired slots. Not sure the latter is an issue on this motherboard. Manufacturers usually recommend using matched sets.

Look at it this way: you can buy 2x2GB DDR-800 RAM at Newegg, today, for $50. At worst, you've blown $25 that you didn't need to spend. If you simply stick in another DIMM, and it doesn't work for some reason, what's the value of your time to disassemble the board again and wait three days for the second DIMM to ship to you? Not to mention the value of having your board down while you wait?

[shap]

Don't know much here, and I'm not sure if applies to K, but isn't it so that more ram makes things run better?

That's mostly an issue for systems that run general-purposes (varying) workloads. As the workloads get close to the size of physical ram, the OS is forced to trade things off, and the best strategies for doing this aren't very good.

I would not expect this to be an issue for KRONOS. KRONOS runs a fixed, previously known workload (the engines), and has been engineered to run that workload well within the RAM already available. While there might be in internal improvement in some hypothetical sense, I'd be very surprised if you would experience any practical difference from the outside.

But I'm speaking purely from general OS principles here. Hopefully Dan or somebody else from Korg will chime in if I haven't got my KRONOS workload assumptions quite right.

[Randelph]

robbinhood wrote:

But honestly, there are no "security tabs" or "anti tamper" when you open a keyboard. You gotta be pretty dumb to screw something up like putting in new ram.

I would imagine that voiding the warranty for opening up the chassis is when obvious/egregious mods have been made. Like you said, if there's no security tabs to look out for, and you're careful, how would anyone know?

Of course, if you're sending it in for warranty repair because of some modification you did....

When I had my Motif XS I replaced 2 keys and a circuit panel that had 5 sliders on it myself. I bought the Yamaha service manual for $20 on dvd, and it was close to a slam dunk because of the clear and easy instructions for opening everything up; but I made sure that I kept track of all those little screws!!!

Randy

[Lou]

Don't know much here, and I'm not sure if applies to K, but isn't it so that more ram makes things run better?

That's mostly an issue for systems that run general-purposes (varying) workloads. As the workloads get close to the size of physical ram, the OS is forced to trade things off, and the best strategies for doing this aren't very good.

I would not expect this to be an issue for KRONOS. KRONOS runs a fixed, previously known workload (the engines), and has been engineered to run that workload well within the RAM already available. While there might be in internal improvement in some hypothetical sense, I'd be very surprised if you would experience any practical difference from the outside.

But I'm speaking purely from general OS principles here. Hopefully Dan or somebody else from Korg will chime in if I haven't got my KRONOS workload assumptions quite right.

Thanks for the breakdown shap, like I said, this isn't my area, but I recall buying a laptop and after only a week or so added some ram and noticed a difference out of the gate.
That's why I questioned it, and obviously, there have been sluggish screen complaints, etc. ..

[shap]

That's why I questioned it, and obviously, there have been sluggish screen complaints, etc. ..

I'm probably about to get myself in trouble again, but here goes nothing.

All of you know that I've been looking at the KRONOS both as a musician and as someone asking how to build a better keyboard. From the latter perspective, a bunch of things have puzzled me. The missing ethernet port, the screen, and so forth. I assumed that "running linux" meant "as the primary system", and linux graphics subsystems haven't had that kind of bug in a really long time. I've also wondered about the font issues.

This afternoon I spent some time tracing cables through the tear-down photos. And I went from puzzled, to bewildered, to "nah, that can't be right", to "why on earth?" to "hey, but it works, so don't go jumping to any judgments here." From the photos, I was starting to think that the linux PC in here is acting as a slave to the board with the ARM chip. Which seemed hard to believe, but the wiring sure suggests it. And then Dan said:

It's much more like an embedded system which happens to use an Intel CPU as the DSP.

And the more I dig at those hardware photos, the more inclined I am to take him literally. And if Linux isn't driving the boat, then a bunch of things that struck me as strange about it make a lot more sense to me.

And you know, it's not how I would have approached it, but I'm sitting in front of it, and, it's really hard to argue with something that works this reliably.

But I can't wait to tear down my Motif XF at this point and see what the "other guys" did.

[kotklegg]

Amen to that brother!! I think the Kronos is the most awesome keyboard I've ever played!! No complaints here!!

Anyways, would be nice to have an OS update that will enable Kronos to see and use all 4 GB of RAM, without the stupid 3.25 limitation.

This is proof that PEOPLE WILL NEVER BE SATISFIED.... they say it would only be able to read 2 gb... someone figures out a way to get it to read 3.25 gb and FOLKS STILL WANT MORE!!!!!

I for one (if I was korg or any other product manufacturer) would stop listening to people that respond like that. Products HAVE limitations. Believe it or not. If companies spent time trying to overcome EVERY limitation, we'd never have products to COMPLAIN ABOUT AND NEVER BE SATISFIED WITH...."