Super 8 - New polysynth by Native Instruments

[telecode]

Using your settings, I am getting 36%.

[ghettosynth]

When I was measuring, on my old Mac Mini, S8 played through Reaktor 6 was using about 10%. For the sample demo, 3 Instances of S8 played through Maschine was using hovering about 18%. In comparison, Arturia DX-7 and Jup-v3 was around 5%. (Unless I am looking at the wrong thing. I am looking at user space CPU usage in activity monitor).

I think that's probably why, but, using performance monitors external to Reaktor is the only way to compare it with non-reaktor instruments.

[telecode]

When I was measuring, on my old Mac Mini, S8 played through Reaktor 6 was using about 10%. For the sample demo, 3 Instances of S8 played through Maschine was using hovering about 18%. In comparison, Arturia DX-7 and Jup-v3 was around 5%. (Unless I am looking at the wrong thing. I am looking at user space CPU usage in activity monitor).

I think that's probably why, but, using performance monitors external to Reaktor is the only way to compare it with non-reaktor instruments.

yeah. in my new tests using the top tool, monark is using 11%, super 8 31%, serum 21% and pigments just 12%.

[Biome_Digital]

Sounds pretty good, but was I the only one hearing that clicking sound? Was it because of the envelope settings? I should just get the demo version and check it out myself.

Did get this clicking as well while playing and it wasn't wanted

I was wondering why the person who made the video didn't notice, and just left it that way. I always listen back to my tracks of music or presets I made many times over to get it how I'd want, so it seems strange.

indeed

[Sinisterbr]

It's quite heavy on mine. Not as heavy as Monark, but, surely in the ballpark of Diva.

Strange , my pc is almost 8 years old
Super 8 takes 20% cpu without effects , 25% with eff. (running at 44khz)
Monark just 10% ( which is running @ 88Khz by default )

If you switch Super8 to one voice and double the sample rate then the statement makes sense. On my older machine I get about 5 to 6 % on super8 and about 9% with Monark. Super8 is about 20% on that same machine with the 8 voices at 44.1 and about 33% with 16 voices.

Wait, what? Super 8 only goes up to 8 voices

Like all Reaktor instruments that aren't intrinsically limited to some particular polyphony, i.e., mono for, e.g., Monark, or, i.e., fixed poly, for, e.g., ensembles like the old Reaktor version of a popular Oberheim emulation, sorry, blanking on the name right now, you can change the polyphony of the instrument to whatever you want.

Out of the box it is set to eight voice polyphony. You can increase this at the expense of more CPU. I find that 16 voices is a great balance for my use and that varies between 25%(ish) and 35%(ish) cpu depending on which of my studio machines that I'm using. On my older i7, 32 voices is about max. Max with Reaktor is usually something on the order of about 70% CPU on its meter because it will go up dynamically depending on what the ensemble is doing.

Compare that to 2-Osc, a decent synth in the factory library that uses core with oversampling, but not ZDF and I can get 56 voices from that to get about the same CPU usage as 32 in Super8. Finally, compare that with SoundSchool, also a factory library 2 osc analog synth that only uses primary filters, and it takes about 900 voices to get to about the same CPU level on the same machine.

I see, but I cannot make this tweak if I only have Reaktor player, can I?

[gentleclockdivider]

Using your settings, I am getting 36%.

Wow , and what kind of macbook is that ?

[telecode]

Mac Mini

[v1o]

CPU usage is not a good measurement. What you should be measuring is DSP time.

[telecode]

CPU usage is not a good measurement. What you should be measuring is DSP time.

what tool do you use to measure DSP?

[ghettosynth]

CPU usage is not a good measurement. What you should be measuring is DSP time.

what tool do you use to measure DSP?

To compare Monark with Super 8 just use Reaktor's DSP meter in the upper right hand corner. To compare these with something like Diva, use whatever performance meter that your host has. This should be close enough to give you a valid comparison.

[LoveEnigma18]

Got Super 8 yesterday. Awesome synth, and quite fast and fun to work with!

About CPU usage, FL Studio meter showed it as under 20%. On Windows 10 and Core i7-7700 CPU here.

[ghettosynth]

Few if any of the CPU readings seem all that surprising. It seems pretty consistent. I wonder what the target CPU usage is that NI considers acceptable for the bulk of their customer base?

Is anyone here over 50% with eight voices? I'm curious how old you have to go before you get there. I don't have anything old enough here to test with.

[Gamma-UT]

Is anyone here over 50% with eight voices? I'm curious how old you have to go before you get there. I don't have anything old enough here to test with.

Looking at some of the discrepancies between Monark and Super8 on older machines posted here, I'd take a guess the later instrument is designed to take advantage of the newer AVX instructions and probably goes off a cliff somewhere around the Nehalem/Sandy Bridge divide.

[ghettosynth]

Is anyone here over 50% with eight voices? I'm curious how old you have to go before you get there. I don't have anything old enough here to test with.

Looking at some of the discrepancies between Monark and Super8 on older machines posted here, I'd take a guess the later instrument is designed to take advantage of the newer AVX instructions and probably goes off a cliff somewhere around the Nehalem/Sandy Bridge divide.

chk071 reported, but didn't provide data, unless I missed it. In any case S8 is fully core and I haven't found anything locked and I've traveled all the way down the filter. So, if it's using AVX then it's taking advantage of some aspect of core that the compiler leverages and that isn't clear to the end user, well, at least to me..

But one thing that you have to keep in mind is that there is an overhead for the first voice. An 8 voice poly synth built with two oscillators and a Monark filter takes about 30 percent CPU (according to Reaktor) on the same machine that S8 takes 11 percent. However, pull that Monark filter out and replace it with a Paul filter and the usage drops to 16 percent. So core ZDF based polys are not impossible on modern machines, meaning, I'm not that surprised about Super 8. Monark uses so much CPU largely because the Monark filter is a beast. We don't know what's inside of course, but we can compare Paul to S8 and both are relatively straightforward.

I think that you're probably right though on the divide. I don't have any machines that have a pre-sandy bridge i-whatever and, even though I have some core2 machines, none of them run windows.

[telecode]

Got Super 8 yesterday. Awesome synth, and quite fast and fun to work with!

About CPU usage, FL Studio meter showed it as under 20%. On Windows 10 and Core i7-7700 CPU here.

Same here. Its usable on my old Mac i7 and its fast to get good sounding results, which is important to me. I doub I will ever have more than 3 or 4 tracks of S8 in one project ever with the stuff I ma doing.