Spectralayers Song Unmix (high) unuseable on new PC

SpectraLayers
1

After upgrading to a new Intel Core Ultra 9 285k PC. pretty much every app I use has shown a significant performance boost, including Cubase Pro 15.

Spectralayers Pro 12.0.40, in sharp contrast, completely crashed my new PC twice and will be removed from this machine as totally unpredictable danger zone, when using Song Unmix in high quality.

  • the first crash happened with my passive Nvidia 3050 KalmX with only 6Gb of RAM still switched on by default for automatic rendering and didn’t come unexpected due to the GPU memory demands of Spectralayers
  • After fixing the setting to pure CPU rendering, I was shocked to see a second total crash after 2-3 minutes of rendering.

While I can still use Spectralayers Song Unmix or other jobs in high quality on my AMD notebook, Spectralayers Pro 12 is practically unuseable on my brand new Intel PC. Of course I find that state of things to be completely intolerable.

P.S. after a lot of checking I still don’t like this state of things, but refrain from any software judgement, see below in the thread.

1 Like
3

SL12 seems to be much more sensitive to RAM amounts, I used to use this card and it did work OK with an older version of SL … possibly 11. In SL12 there’s also an option to use “Shared memory” which might help, but there are several reports of RAM-related crashes, which shouldn’t happen.

I know this doesn’t help, but I have had zero problems on what is also a new build, but with 64GB system RAM and NVIDIA RTX 3060 Ti 16GB. I totally agree it should not crash with the 3050 KalmX 6GB though. Running slower is one thing, but complete crashing is another. Be sure to log a support ticket though.

4

It happens to me as well. During initialization in the Unmix Songs.., Windows Activity Manager shows that RAM usage ramps up to the maximum and then SL12 crashes. I then tried unchecking the options in the Unmix Song.. Process and then checking just one stem for unmixing at a time: in my case it worked with the vocals, drums, bass, sax. The crash occurs in my case if I ask to separate either the guitar or piano stems. In the ufficial communications, they said they did modifications to those algorythms (guitar and piano). I spent one afternoon to do these checks, I wrote to the support one week ago, no answer so far.

5

My new Intel system has 64 Gb of RAM like yours. Why it would still crash with system preference set to CPU is beyond me.

I had no such CPU rendering problems with my former AMD 5090X system with 64Gb RAM.

6

should really tag @Robin_Lobel for this one, no?

7

Hopefully, the next release, version 13, will be better. The release of version 12 was, imho, a disaster, to be honest. Just an expensive bunch of bugs with a deep feel of fermented milk.

For our usecases (rather scientific than artistic), the program in its actual state of technical realisation is plain unusable.
image

And no! I do expect the developer(s) to figure out the bugs themselves with thorough testing before release. We do not want to figure out tedious “workarounds” for basic functions to work as they are named and therefore should behave as can be bravely expected.

8

Mmmm … my new build uses an AMD Ryzen 7 9700X (65W TDP).

One last thing, just to rule out any RAM/mainboard hardware issue: try running memtest86+ overnight.

You’d be amazed at the number of times RAM issues cause crashes that never show up in other applications. The fact is that most applications will only ever use tiny fractions of a 64GB system, and the problem might never show up until an application tries to use very large amounts of RAM intensely.

9

Thanks for the proposal, MrSoundman. I doubt it’s a RAM problem, because I get get no problems when running several large sample libraries.

But I will run HCI Memtest (not Memtest86Plus, because I don’t want to boot from a stick and deactivate and activate secure boot in bios) overnight just to be quite sure.

10

In that case I recommend you use the built-in Windows Memory Diagnostic instead.

11

Hmm, I notice its been a bit ‘quiet’ from him on here of late (in comparison to how things used to be - engaging/responding to users posts). Hope all’s ok…

1 Like
12

Indeed, I agree, hope all is well :slight_smile:

1 Like
13

@Robin_Lobel

I have done some more research over the weekend with the limited means and technical understanding I have as Spectralayers user.

- meanwhile I have run a memory test and my 64Gb Kingston memory seems ok. As I said, I use some big sample libraries without any problems on this PC and would probably have noticed RAM problems anyway, if there had been any.

- the next thing I found out will probably make Robin and the Spectralayers team feel a little better: my new Intel PC not only had crashes with Spectralayers 12 stem separation, but as it turned out then also with RipX DAW Pro and Song Master Pro. This PC works fine with everything else, like rendering big files with DaVinci Resolve, using Cubase with CPU heavy plugins for hours, CPU heavy Abbyy Finereader OCR rendering of big books, some gaming, whatever: all runs smooth and fast. The exclusive exception seems to be stem separation with certain algorithms.

- the Windows error protocol for the crashes names rather unspecific Kernel-power-41 errors as cause for the crashes. This can mean all kinds of component and software related malfunctioning resulting in abrupt crashes (blackscreen without error message) and does not help to find the culprit.

- when I did some more internet research on the latest Intel CPUs, mainboards, and RAM, I found out about a problem with memory modes. Modern PCs/mainboards normally run in so called XMP mode: that’s essentially an automatic overclocking mode to get the highest possible clock rating from your RAM. This can cause instabilities in fringe situations with high and fast changing RAM loads. So I switched off XMP mode in BIOS to get a stable DDR5 clock within normal specs.

- as a result, Song Master Pro now works fine again again, with high quality and fast stem separation of 6 stems.

- But this did not change the rendering crashes of Spectralayers 12 and RipX Daw.

Bottom line: there seems to be a lasting instability issue between my new Intel 285 PC (or perhaps with this processor, which had a lot of issues when it first appeared, almost all of which seem to be fixed by now) and certain types(!) of stem rendering, while at least one stem separation tool works fine here now. I have no explanation and can only speculate if the issue happens with certain stem algorithms (like Demux V.4) exclusively, or whatever. I simply don’t want to invest more time into this irritating crap, because I simply want to make music. But at least I wanted to inform Robin Lobel and others about my findings so far, while I have to restrict use of Spectralayers 12 and RipX to my AMD notebook, thankfully still doing the job. without any problems.

Back to making music now. :winking_face_with_tongue:

1 Like
14

Hi Dirk (@jim_knopf ) - nice lot of further research and time spent reporting your findings.

With the lack of visibility of @Robin_Lobel on these pages of late (last recorded ‘activity’ as being Oct 5th 2025), it might be an idea to log all this directly with Steinberg Tech Support (opening a support ticket via your MySteinberg account), as well.

They may be able to answer themselves or, in the very least pass the info on to the SLP dev team/support for examination.

(PS:- after some time spent actually making some music.!!)

1 Like
15

then tag him!!! @Robin_Lobel

16

hmmm, I wonder…is this a PT integration related timing? If memory serves, seems to be around that time.

Robin, did you sell and have a NDA? Anyway, I (very selfishly) hope SLP will continue to be developed. I dread the alternative.

17

Good detective work, and thank you for sharing the results! :clap:

Only read the following after you’ve made some more music and feel like a bit more hardware tinkiering! :grinning_face_with_big_eyes:

This is probably not what you want to hear, and we’ll probably have to agree to differ on this, but all of the issues you mention in relation to the Intel Core Ultra 9 285k are what led me to go with an AMD CPU for the first time ever on my recent build.

I’m not suggesting the 285k will never work for stem separation, but I still think the issue may be RAM-related. I’m not suggesting your RAM is defective, but the fact that one of the three products mentioned now works after changing the XMP profile and lowering the speed the RAM is operated at, suggests to me that it would be worth trying a bit more troubleshooting, for example:

  • check whether your RAM modules are on the verified list of your motherboard manufacturer
  • lowering the speed of the RAM even more (look up the JEDEC specifications of the RAM and manually set the lowest timings in your BIOS, if it allows for that).
  • if your 64GB consists of 4 modules, try removing 2 as a test
  • if possible, borrow some other RAM modules and test

Where I am going with this, is that it sounds to me like there may be an incompatibility between your CPU/mainboard and your current RAM.

1 Like
18

@MrSoundman

This won’t cost me much time, since I had already checked these points:

- I HAD to go with an Intel build despite my high satisfaction with my previous 4 year old AMD 5900X PC, simply because there are no more AMD mainboards with onboard Thunderbolt (like my Gigabyte Vision D back then) available - and I use an UAD Apollo x6 Gen.2 soundcard with more UAD hardware.

- despite the stem separation issues I am glad that after some hesitation I went for the 285k: it is not just leading in DAW bench tests with a surprisingly significant lead over AMD right now, but also delivers roughly double the Asio CPU power I had before in Cubase. That’s much more important to me than any stem splitter, and my notebook does the RipX and Spectralayers stem splitting fast and fine (with network export to my PC available anytime).

- my RAM modules are only two and verified for my main board

All in all, I will neither lower the RAM JEDEC specs even more (at the moment I have practically no perceivably slower machine than I had) nor would I want to change or exchange a VERY fast and smooth running PC system for the sake of two stem splittings apps, which run fine on my notebook. It’s quite a small trade in for a very big step up in my overall computer performance which I really enjoy.

1 Like
19

Fair enough, in that case I don’t think I can help any more, but one question remains: how can this be a software issue (with either SpectraLayers or RipX) if they run OK on a different PC?

1 Like
20

Both apps crashing did run on my old PC, so I’m aware of some incompatibility of soft- and hardware. But it’s beyond my possibilities to find out what exactly might be going wrong. My computer build has been done by the experienced team of Silentmaxx, so the whole setup has been done on expert level.

At the moment I simply refrain from any judgement and don’t blame Spectralayers or RipX for this weird kind of crashes, not knowing what combination of factors causes them.

Just moving on now.

2 Likes
21

P.S. As it turned out, Song Master Pro crashes as well and disabling xmp has had no result. I got the wrong impression, because a song I tested was short enough to survive the stem separation process. But I get crashes with other songs.

1 Like