M+ computer internals + info
Comments
-
To be fair, I dont know what exact models they use or if they are superior to the M+ CPU or not but i doubt theres that much raw power difference. Maybe for the X model since it's expensive af?
More importantly AKAI develops plugs and such with their MPCS in mind, while NI is porting things over that were originally never meant for a standalone, so it's easier for AKAI to be more CPU efficient / optimised, this might be more important than just raw power, brute force is not always the answer.
Thats kind of the "innovation" on the M+, the fact it can run desktop plugins, advanced stuff like Kontakt or Reaktor but thats both a gift and curse atm due to the cpu effiency and lack of GUI to fully use those instruments. Im not sure if internally NI sees their CPU choice as "weak", I think whoever made the call probably stands by it :S
1 -
I guess, choice of CPU was the compromise to fit to energy envelope needed for passive cooling. And also to fit into price window....
Development including manufacturing may take several years. They had to use what was on the table back then. It was time of Intel´s 4 cores is the topend that fits all....
During preparation period AMD Zen came and Apple Silicon came and those have changed the rules...
0 -
According to a thread I found on the MPC forum the X and Live have a Rockchip RK3288, quadcore 1.8 GHz quadcore GPU.
I haven't found a direct comparison to what's inside the M+, but here's a comparison against a different model Atom https://gadgetversus.com/processor/intel-atom-x5-z8350-vs-rockchip-rk3288/ you can use that as a basis to adjust up/down towards the target board.
0 -
Interesting... But isn't that a bit like comparing oranges to apples since it's different architectures? I mean it's an average of results pulled from Geekbench (for android??) which is quite a general purpose benchmark. I personally never correlated any Geekbench results to actual Audio performance to know tbh so I must admit full ignorance here.
0 -
Yeah 100%. MIPS is the only real universal standard but most of those tests eventually come down to how optimized the compiler is for that platform. AMPERE has run into that issue with their 100+ core ARM server systems.
But real world performance depends entirely on the program. I'm sure if you're into gaming you've run into those games that are not optimized for multicore performance? And even with Kontakt, you can only pick up to 16 cores, so I'm assuming anything beyond that isn't used?
0 -
Poking around the M+ this past weekend I figured out how to download the system files for the latest update.
It turns out if you download the update, but tell it not to perform the update then the files are still on the SDcard.
However the entropy check was extremely high, meaning the files appear encrypted or at best compressed with a non-standard compression method, or possible both lol.
Interestingly all the .nu files have the same header as system-reset-v2.nu2. They all start with a magic value of 0x4E4950570100
Unfortunately this is as far as I got with the files. I came up empty handed with Google searches for the hex value and ascii value of the magic. A binwalk run on the files also came up with no useful info, just a few false positives.
3 -
Now thats some really valuable info right there!! I hope this gets the attention of some indie devs with reverse engineering skills.
0 -
Maybe compressed and simply XORed? Or XORed by pseudorandom series? Part of magic header might be a seed value for pseudorandom generator....
Just guessing, I do not have any experience with such things. It is the way I might do it, if I wanted to make things harder using simple method.
0 -
That's an interesting thought. XOR was used on the popular Netgear Nighthawk M1. It's possible the files are XOR'd, but I'm not sure how you would determine the key without any patterns to go off of. It's unlikely that the key is only 6 bytes, 4 bytes would make more sense, but that's awfully short for an XOR key. Usually they're extremely long like 64-256 bytes.
Here's a link to the article where they talk about the Netgear XOR https://www.pentestpartners.com/security-blog/breaking-bad-firmware-encryption-case-study-on-the-netgear-nighthawk-m1/
0 -
No, my idea is that part of those 6 bytes are seed for pseudorandom numbers generator. And the series of pseudorandom numbers would be used for XOR.
Depending on random number generation algorithm the pseudorandom series could be pretty long (thousands of bytes). It might create high entropy data stream even from noncompressed data. I guess.
Decoding would be. Get seed, calculate pseudorandom series of numbers using known algorithm. XOR data using pseudorandom. You get data. They may be compressed... So, decompress in that case.
But maybe that NI has used something more complex.... It depends how much they desire to keep it "sectet".
One could try simple algorithm for generating pseudorandom, and for each possible seed try to "decipher" data and calculate entropy. Maybe it starts dropping here and there. Another complication is that you do not know how much of pseudorandom use. I would start with all. The series repeats and repeats. It is simpler not to bother with the length.
But hard to say, if it brings any results....
And beside that. There may be sort of CRC or it may be electronically signed. So, it could be difficult to create your own modified file....
0 -
Incredible that M+ will work as normal MK3 when the internal computer is simply removed... IMO for standalone experience MK3 with some headless computer is the way to go. Anybody tried something like that?: https://www.youtube.com/watch?v=45mluu4p4Go
0 -
I'm pretty sure I've seen a custom MK3 controller made into a custom standalone Maschine.. it was on some blog, years ago..
0 -
@Seqsual @tetsuneko please do not let this thread derail into another „what is better than the M+“ bashfest.
This thread is about something completely different: A journey of investigation about the inner workings of the M+.
And AFAICR it’s only the second educated and intelligent thread on this topic ever since the M+ appeared. While there have been plenty of the other kind.
Thank you very much for considering this.
3 -
what a great thread – thanks guys!
the whole mpc line uses the same arm cortex chip (introduced with the mpc live 2016). the newly released MPC X SE and MPC Key61 using a 4gb ram version of the same chip, instead of the "old" 2gb version.
rumors has it, that they bought a big load of those cortex chips, which arent available anymore (since they released the first mpc live acutally). so i think, they want to sell as many mpc's as possible (well, obviously) with this configuration before moving on to something new.
how difficult would it be, to release a sampler/groovebox with an updated processor? i mean, for both, akai's cortex and maschines atom chip, new versions are available.
0 -
The Congatec carrier board is on the slow boat from the UK. Estimated arrival date is Feb 6 :(
0
Categories
- All Categories
- 19 Welcome
- 1.4K Hangout
- 60 NI News
- 724 Tech Talks
- 3.8K Native Access
- 15.6K Komplete
- 1.9K Komplete General
- 4.1K Komplete Kontrol
- 5.4K Kontakt
- 1.5K Reaktor
- 361 Battery 4
- 804 Guitar Rig & FX
- 410 Massive X & Synths
- 1.2K Other Software & Hardware
- 5.4K Maschine
- 6.9K Traktor
- 6.9K Traktor Software & Hardware
- Check out everything you can do
- Create an account
- See member benefits
- Answer questions
- Ask the community
- See product news
- Connect with creators