Thorn: Dmitry Sches' new synth!

[BONES]

i won't go into this anymore, you're wrong and you have an agenda and its f**king annoying.

I am not wrong and my only agenda is truth.

Except reality is that M1 slaughters every one of that chip in single core performance (very beneficial for audio) and it's about 10% worse in multicore, while being passively cooled, while consuming 60% less power.

Except it isn't. Here's another set of benchmarks for single core performance. Notice that where you have provided a useless table that doesn't offer any details of what was tested, this is benchmarking actual computers that you can buy -



As you can see, the M1 does perform very well but it is beaten by a 4 core Core i7 (the only 11th gen Intel machine on test). Given that 6 core i7's are more common in performance machines, and there are Core i9s above that, it puts the M1 very much in the mid tier only.

Why is power consumption meaningful? Because power consumption directly affects how much can you squeeze out of a chip before it throttles.

Which is why it makes sense to test actual computers you can buy, as has been done in the article I linked to. So power saving or not, the M1 still doesn't perform anywhere near as well as Apple claims.

[Ploki]

Except in i.e logic, M1 gets a bigger track count than i9 8 core (16”) running same synthetic test. (While being whisper quiet)
Even when running m1 @64 buffersize and i9 @1024.

It simply smokes a desktop 6-core irl - i already told you i tested it myself extensively, so dont be so f**king stubborn- you’re WRONG, deal with it.

I didn’t blindly buy it. I tested it, and if it didn’t perform I’d return it. Instead i sold my i7 mini

[BONES]

You tested against your old computer, it's pointless. My new computer can run projects that used to max out the old one without the CPU meter getting over 25%. It's how it works and you can see it in the benchmark image I posted above, where the 4 core 11th gen i7 outperforms the 6 core 10th gen laptops.

A better test would simply be to run the same project across different machines and see what the CPU usage is. That's pretty much how Cinebench works, using Cinema 4D, which is what makes it a good indicator. Of course, Logic is useless for that so it would need to be something cross-platform, like Cubase, Live, Bitwig or Studio One using a set of common plugins.

[LeVzi]

https://techjourneyman.com/blog/apple-m ... d-ryzen-9/

M1 isnt always better than the 5900X in ALL single core tests, the cinebench test it was worse, but in the geekbench it beat it.

It obviously beat it for the one with the graphics as the 5900X doesnt have onboard graphics. So its decent for single core performance. No denying that.

[Ploki]

You tested against your old computer, it's pointless. My new computer can run projects that used to max out the old one without the CPU meter getting over 25%. It's how it works and you can see it in the benchmark image I posted above, where the 4 core 11th gen i7 outperforms the 6 core 10th gen laptops.

Yeah well, i pick Geekbench:
https://browser.geekbench.com/processor-benchmarks
https://browser.geekbench.com/mac-benchmarks

just ran it on my lap, battery power

Screenshot 2021-10-12 at 10.56.38.png

here's a brand new (2021) 45W (3x TDP) Ryzen 5900HX for example.

Screenshot 2021-10-12 at 11.11.44.png

A better test would simply be to run the same project across different machines and see what the CPU usage is. That's pretty much how Cinebench works, using Cinema 4D, which is what makes it a good indicator. Of course, Logic is useless for that so it would need to be something cross-platform, like Cubase, Live, Bitwig or Studio One using a set of common plugins.

That's what real world is tho - how well software works on hardware. I did a test with fabfilter/valhalla plugs running logic while someone copied routing/setup on 9900K on pro tools, and i got a bit more tracks with it, although the synthetic benchmark of 9900K (even on geekbench) is higher. If ProTools wastes CPU cycles by existing, eh?
Anyway, someone with iMac 9900K did the "new logic benchmark" and got 125 tracks (i got 110) which about 10% more, which is exactly how higher geekbench benchmark for 9900k is.


You can have a super powerful engine, but if the car is clumsy and heavy, an agile, lighter car will still outperform it on a track.

But since S1 is M1 ready, i guess i can make an S1 project with Thorn now I guess.

https://techjourneyman.com/blog/apple-m ... d-ryzen-9/

M1 isnt always better than the 5900X in ALL single core tests, the cinebench test it was worse, but in the geekbench it beat it.

It obviously beat it for the one with the graphics as the 5900X doesnt have onboard graphics. So its decent for single core performance. No denying that.

That's crazy. Before M1 nobody would even dream of comparing a 100W+ desktop chip to a low-power mobile chip

[himalaya]

Hello fellow Thornrites,

Having been tempted by the current PA discount, I downloaded a demo of Thorn for a spin. Loaded the plugin, tweaked a few knobs and couldn't' believe the richly digital, lush and shimmering sounds that flowed so effortlesly out of this synth. Somehow, it reminds me of a PPG and RGCAudio Z3ta+ in equal measures. I thought that I might as well try Thorn as it's marketed as a Spectral Synthesiser and since LinPlug are no longer with us, its fantastic Spektral synth is no longer supported, so I thought Thorn could be a nice replacement. I bought it without hesitation. So far, I am impressed with the sounds I'm getting out of it.

A few questions for the seasoned Thornrites since my quick perusal of the manual did not bring any answers:

1. Is there no bi-polar modulation in the mod-matrix? I wanted to assign 'random' to pan, but I do not see a bi-polar option anywhere to get the random mod-source to modulate -/+. Assigning 'random' to pan twice and have each set to - and + respectively nullifies the pan mod. So what's the trick? The LFOs have a bipolar switch, but I can't see a similar option in the mod-matrix.

2. I can't see where the oscillator 'fine tune' option is. I normally use it in conjunction with LFOs/random sources to get some random detuning between oscillators. All I see is Octave and Semitone parameters.

3. I'm missing a flip-flop mod source. Is there?

[kev2525]

Hi all, I am a new owner of thorn. Are there any good resources/tutorials to learn this synth?

[tactile_coast]

Hi all, I am a new owner of thorn. Are there any good resources/tutorials to learn this synth?

Producertech do quite a good one which I found useful in understanding how the synth works. It is a little pricey to buy but they do a free 14 day streaming trial and regular free promotional periods.

Hello fellow Thornrites,

1. Is there no bi-polar modulation in the mod-matrix? I wanted to assign 'random' to pan, but I do not see a bi-polar option anywhere to get the random mod-source to modulate -/+. Assigning 'random' to pan twice and have each set to - and + respectively nullifies the pan mod. So what's the trick? The LFOs have a bipolar switch, but I can't see a similar option in the mod-matrix.

2. I can't see where the oscillator 'fine tune' option is. I normally use it in conjunction with LFOs/random sources to get some random detuning between oscillators. All I see is Octave and Semitone parameters.

3. I'm missing a flip-flop mod source. Is there?

If you are using Bitwig its not a problem as you can just assign an appropriate modulator as required for each of your steps to these parameters (Pan/Fine Tune) as they are displayed as automatable. Then save as the default assignment for either the preset or device.

Other, lesser DAWS, may also have a similar albeit more convoluted and less elegant solution.

[Teksonik]

This could be a clue why he might not be looking for DAW workarounds but for a solution when making patches:

himalaya.png

Anyway to answer the questions I'm not seeing any way to do Bi-polar modulation with anything other than the LFO's in Thorn.

Although you could try Random>MSEG 1> Osc Pan and set the MSEQ to alternating on/off. Kind of works but it's wonky at best.

As for Fine Tune...."DTN" will work even when Unison voices are at 1.

[EnochLight]

If you are using Bitwig its not a problem..

Other, lesser DAWS…

Hahahaha - spotted the Bitwig fanboy!

[himalaya]

If you are using Bitwig its not a problem ...

I know, I know. If I was using the SpaceX rocket I could even play Thorn while floating in Orbit. Alas....
But thank you for the suggestion nonetheless.

[himalaya]

Although you could try Random>MSEG 1> Osc Pan and set the MSEQ to alternating on/off. Kind of works but it's wonky at best.

Thank you for the suggestion. I kind of max out the MSEQs before I could use them as you suggest.

As for Fine Tune...."DTN" will work even when Unison voices are at 1.

Aha! This works! I didn't realise that DTN goes into the negative numbers. Thanks!
I thought that it only works as a positive (unipolar) modulation parameter in conjunction with Unison. Thanks again.

[GaryG]

I'm just demo-ing the full version (used the CM one on and off for a while) and have an issue modulating the filter cutoff; seems to introduce noise (distortion?). I tried emulating a patch that works fine in the CM version, going back to basics to make sure I had nothing else effecting it.

I can replicate it by taking the Pads Keys preset and adding an mseg to modulate filter 1 cutoff (filter 2 is off). I used steps 10 in the mseg, mod level ~80%. If you turn the filter res up high then in CM i get these little high freq 'dings' punctuating the sound but in the full version it's just nasty. Missing something maybe? Cheers.

[Ploki]

Could you share the patch?

[GaryG]

Sure (zipped so I could attach it directly here).

If you disable the MSEG and play around with the filter cutoff around 12- 1 o'clock then that's the kind of sound I get in CM version. With it enabled I get just noise really, like the filter is broken when modulated... I do get similar results with the LFO too.

Probably my lack of understanding but curious if there's really a difference or what really.

Thanks for looking.