KVR Audio

zvenx · Post by **zvenx** » Mon Sep 22, 2014 1:02 am

make sure it is showing the keyboard,
bottom left
rsp

mlt · Post by **mlt** » Mon Sep 22, 2014 1:04 am

loungepanda wrote:
mlt wrote:Truly amazing synth.
The only think that concerns me is the CPU load, especially in some presets that use long releases.
When I shorten the release or decrease the polyphony the CPU load comes down, so I think it would be nice
if Steve could add a function, as an option, that automatically "kills" voices (or something similar) in order to lighten the CPU load. Other things like unison doesn't seem to load the CPU much, so it is mainly the polyphony.
I have a fairly old system so this was quit expected. I wonder if someone who has a modern CPU (like 4770, 4790) could tell us about how much is the CPU load with some of the heavier patches.
Tried finding some of the heavier patches to see what load is like on my 4970k (stock 4.0, turbo not on):

PD Big Minor Seventh Pad [SN], 2 OSC with 10 Unison on both: 18-19% holding one note.

Other unison pads/sequences were around 8-12% s

Mine reaches around 26% on the same sound, but 19% is quite a lot (especially if you have i7 4790k).
I hope Steve will be able to optimize a bit the CPU load.

The synth sounds a bit a "cold" in some patches but I think that's because of the very clear top end that the oscillators are able to produce. If for example you add a "German filter" (with the cutoff full open) in the FX rack the sound becomes much warmer.
In general there is a lot of flexibility on how you can make it sound.

TheKid · Post by **TheKid** » Mon Sep 22, 2014 1:21 am

Can anyone try to reproduce a plugin crash, please? While playing switch oversampling from 2x to 4x back to 2x. When reducing oversampling back Ableton Live crashes. Live 9.1.5 32bit on Windows 7 64bit. I'll report this to Steve.

I also have a pretty high load, spikes up to 40% on my i7 4770.

bitcrusher · Post by **bitcrusher** » Mon Sep 22, 2014 1:31 am

if Steve could add a function, as an option, that automatically "kills" voices (or something similar) in order to lighten the CPU load.

Yes there are 2 features here already:

1) The Polyphony limit in the lower-right corner labelled "POLY". This setting defines the maximum number of "notes" you can have active not voices (I felt nobody wants to play math games with how many notes you can press, like you have to in some other synths). You'll see the literal # of voices used in the display below.

2) The preference "Limit Polyphony to 1 of any given note number". This means if E.g. C3 is in release, and you press it again, the old C3 will fade out quickly and the new C3 will take its place. This can have a huge benefit in the number of voices 'hanging around'. Obviously if you're sliding your finger across the keyboard low-to-high it won't make a difference, but in can also have a sonic benefit in addition to CPU reduction, because 2 voices playing the same MIDI note will phase against each other.

As for "Cold" sound, I hear it as 'sharp, clean, and pure'. People have mentioned pitch drift and filtering the highs as possiblities. There are obviously ways to post-process as well. Additional filter types with different sorts of internal drive or character are of course a possibility. As is an Arp, of course. I have an impressive list of feature requests going in a database and I'm weighing them as customers chime in and in particular echo each other's wishes (I'm trying to be democratic since I am resource limited - and bug fixes are coming first).
Im aware of the "changing oversampling during playback" issue, it should be an easy fix, for the moment just wait for all voices to stop playing to negate the potential of that happening.

woodsdenis · Post by **woodsdenis** » Mon Sep 22, 2014 2:25 am

Serum sounds like Serum, its "sharp, clean, and pure" sound is what makes it different from others, If you want an "analog" sounding synth do not buy this. Its really as simple as that. I do have quite a few other synth plugins so personally Serum is different,unique and very exciting addition to my arsenal. If you want "Diva" style analogue then buy Diva, don't expect any synth to be a jack of all trades. On that note I really don't want an arp if it takes away from more important things.

Serum really shines in its workflow, I want to make music and not have to think to hard to edit or make a sound. Serum like Nerve is really intuitive and quick.

I would vote for a draft mode as an option to cut down on CPU, especially for pad and multiple voice patches.

swamiill · Post by **swamiill** » Mon Sep 22, 2014 3:07 am

It's showing. Usually when you click browse parameters and move a knob, the linkable object is brought up in the browser. Nothing happens when I move the pitch wheel for some reason. There isn't a Pitch Wheel listed in the browser CC's either

btw I forgot to say nice work steve

billcarroll · Post by **billcarroll** » Mon Sep 22, 2014 3:19 am

Phase47 · Post by **Phase47** » Mon Sep 22, 2014 4:10 am

bitcrusher wrote:As for "Cold" sound, I hear it as 'sharp, clean, and pure'. People have mentioned pitch drift and filtering the highs as possiblities. There are obviously ways to post-process as well. Additional filter types with different sorts of internal drive or character are of course a possibility. As is an Arp, of course. I have an impressive list of feature requests going in a database and I'm weighing them as customers chime in and in particular echo each other's wishes (I'm trying to be democratic since I am resource limited - and bug fixes are coming first).
Im aware of the "changing oversampling during playback" issue, it should be an easy fix, for the moment just wait for all voices to stop playing to negate the potential of that happening.

The best part about Serum to me is that it can be cold and digital, but also quite warm and fat and VA-like. An arp I don't need, because there a million arps available now, but more filter types would be welcome.

V0RT3X · Post by **V0RT3X** » Mon Sep 22, 2014 4:26 am

TheKid wrote:Can anyone try to reproduce a plugin crash, please? While playing switch oversampling from 2x to 4x back to 2x. When reducing oversampling back Ableton Live crashes. Live 9.1.5 32bit on Windows 7 64bit. I'll report this to Steve.

I also have a pretty high load, spikes up to 40% on my i7 4770.

This happened to me too on the OSX version using the Audio Unit.

bitcrusher · Post by **bitcrusher** » Mon Sep 22, 2014 6:37 am

It's showing. Usually when you click browse parameters and move a knob, the linkable object is brought up in the browser. Nothing happens when I move the pitch wheel for some reason. There isn't a Pitch Wheel listed in the browser CC's either

Pitch wheel is dedicated to CC#1. Pitch Bend is dedicated to Pitch Bend MIDI messages. I would expect this is how many (if not most?) softsynths operate.

Again, I'm aware of a potential crash changing oversampling while voices are active.The work around is to ensure there are no voices playing, and then it won't happen. I'll be tackling that and some other potential crashes tomorrow, so it should be fixed very soon.

Ingonator · Post by **Ingonator** » Mon Sep 22, 2014 7:11 am

Just checked the Serum demo in Live 9.1.5 (already checked the AAX in PT 11). Like others mentioned the CPU use is very high. A single pad with 4 notes could be close to max out my CPU.
I admit my CPU is quite old but the amount of CPU use seesm to be comparable to other CPU intense plugins like e.g. Diva (actually Diva in some cases even seems to use less than Serum).

The patches of Serum are playable here but with some of them my CPU is close to the limit. Of course this also depend on the Unison settings. While 7 voices unison still seems to be OK at around 10 voices i really reach the limit of the CPU.

4damind · Post by **4damind** » Mon Sep 22, 2014 7:15 am

Maybe that multithread support could help with this high CPU usage?

Ingonator · Post by **Ingonator** » Mon Sep 22, 2014 7:24 am

4damind wrote:Maybe that multithread support could help with this high CPU usage?

Actually multithread support is a bit like "cheating". It does not really reduce the actual CPU use but shifts it to multiple cores. For a single patch it is OK but for a track with multiple instances it will lead to trouble sooner or later.
BTW when i compared the CPU use to Diva i switched the multithread support in Diva off and still Serum uses the same or even more amount of CPU. Of course this changes when switching to Divine quality.

Also had a look at DUNE 2 which for a comparable patch including Unison uses only around 25-30% of the CPU use of Serum. Even less if the "Multisaw" feature (which actually works for all waveforms and also wit hthe Fm mode) is used instead of the "real" Unison.
With Largo the CPU use is mostly below 10% of a single core here, even if Unison is used. With PPG Wave 3.V it is comparable to Largo or a bit more (also depends on the unison setting like e.g. QUAD mode and the True PPG mode ; in both cases with the new 64-bit updates).
Also with PPG Wavegenerator i hardly reach 10% here.

Don't get me wrong, Serum is a great synth with lots of interesting features and a very nice sound quality but CPU use is still important if you actually want to use the synth with more than on instance.
The wavetable editor is cool indeed while the free Audio Term tool worked quite nicely too so far (and works for multiple synths).

If you got a modern computer with a fast CPU Serum should be a great choice. Can't really say anything bad about the sound quality and feature set.

TonyVegas · Post by **TonyVegas** » Mon Sep 22, 2014 9:04 am

Ingonator wrote:but CPU use is still important

But you said you have an old PC. So you admit the problem is not the synth, but instead your failure to stay up to date. I suppose it is good to know if a small part of the audience can't afford newer technology, but I'm afraid you'll just have to get a part time job and save up if you want to enjoy modern applications.

Anyway, I love the visual feedback, drag/drop linking and visual ADSR/LFO's. Personally I prefer synths with less features and a narrower range of sounds. The does-it-all mega-synth era is just over for me. Still there are some great ideas here with the UI that other developers should be learning from (Uhe, I'm looking at you).

Ingonator · Post by **Ingonator** » Mon Sep 22, 2014 9:15 am

TonyVegas wrote:
Ingonator wrote:but CPU use is still important
But you said you have an old PC. So you admit the problem is not the synth, but instead your failure to stay up to date. I suppose it is good to know if a small part of the audience can't afford newer technology, but I'm afraid you'll just have to get a part time job and save up if you want to enjoy modern applications.

Anyway, I love the visual feedback, drag/drop linking and visual ADSR/LFO's. Personally I prefer synths with less features and a narrower range of sounds. The does-it-all mega-synth era is just over for me. Still there are some great ideas here with the UI that other developers should be learning from (Uhe, I'm looking at you).

You could have a look at the last sentence i just added to my post.

Besides that my "good old" machine served well for several betatests, soundsets (free, commercial or factory presets) and audio demos so far.

Anyway i agree that having tons of features is not everything. I got several wavetable synths but my favorite is still PPG Wave 3.V which feature wise maybe is much limited than others like e.g. Serum.
Even the good old PPG Wave 2.V could still sound great if you program it properly.
FWIW I had released a free bank for both PPG 2.V and 3.V recently:
http://www.kvraudio.com/forum/viewtopic ... 3#p5843263

Of course it depends on how complex the patches should be, especially if you need certain filter modes besides a LPF. In that case some wavetable synths like e.g. PPG 3.V and also PPG WaveGenerator are maybe the wrong choice.

Official Serum thread!